WO2024233290A1

WO2024233290A1 - Respiration sensing

Info

Publication number: WO2024233290A1
Application number: PCT/US2024/027579
Authority: WO
Inventors: Mark SCHATZA; Christopher Thorp; James MANNE-NICHOLAS; Jesse FINE; David DIEKEN
Original assignee: Inspire Medical Systems Inc
Current assignee: Inspire Medical Systems Inc
Priority date: 2023-05-05
Filing date: 2024-05-03
Publication date: 2024-11-14
Anticipated expiration: 2025-11-05
Also published as: AU2024270078A1

Abstract

An example device includes a sensor, a filter, an electrode, and a control portion. The sensor is securable to a patient to provide a sensor signal. The sensor is to sense changes in acceleration indicative of respiration of the patient. The filter is to detect zero crossings of the sensor signal. The electrode is to deliver electrical stimulation to the patient. The control portion is to apply the electrical stimulation based on the detected zero crossings.

Description

RESPIRATION SENSING

Background

[0001] A significant portion of the population suffers from various forms of sleep disordered breathing (SDB). In some patients, external breathing therapy devices and/or mere surgical interventions may fail to treat the sleep disordered breathing behavior.

Brief Description of the Drawings

[0002] FIG. 1A is a block diagram illustrating one example of a device for detecting respiration.

[0003] FIG. 1 B is a block diagram illustrating one example of a device for detecting respiration and/or applying stimulation based on the detected respiration.

[0004] FIG. 2 is a diagram including a side view, schematically representing an example method and/or example device for detecting respiration via an acceleration sensor.

[0005] FIG. 3 is a diagram, including a side view, schematically representing an example method and/or example device for detecting respiration via an acceleration sensor at a chest wall.

[0006] FIG. 4 is a diagram including a graph schematically representing an example filtered, sensed acceleration signal.

[0007] FIGS. 5A and 5B are diagrams including graphs schematically representing example filtered, sensed acceleration signals for two different sensor orientations.

[0008] FIGS. 5C and 5D are diagrams schematically representing examples of stimulation timing including hard refractory and soft refractory.

[0009] FIG. 6 is a flow diagram illustrating one example of a method for treating sleep disordered breathing. [0010] FIGS. 7A and 7B are each a diagram, including a side view, schematically representing an example method and/or example device for detecting respiration via an acceleration sensor.

[0011] FIG. 8 is a diagram schematically representing example acceleration sensing elements.

[0012] FIG. 9A is a diagram including a side view schematically representing an example method and/or example device for detecting respiration with a patient relative to an angled support.

[0013] FIGS. 9B and 9C are each a diagram schematically representing an example method and/or example device including a sensing element extending at a particular angle relative to a gravity vector.

[0014] FIG. 10 is a diagram including a side view schematically representing an example method and/or example device for detecting respiration with a patient relative to an upright support.

[0015] FIG. 11 is a diagram including a front view schematically representing an example method and/or example device in which different sensing elements of an acceleration sensor are oriented relative to a patient’s body.

[0016] FIG. 12 is a diagram including a side view schematically representing an example method and/or example device in which different sensing elements of an acceleration sensor are oriented relative to a patient’s body.

[0017] FIG. 13 is a diagram including a front view schematically representing an example method and/or example device including an implantable medical device comprising an acceleration sensor.

[0018] FIG. 14A is a diagram schematically representing an example method of applying electrical stimulation based on detected zero crossings.

[0019] FIG. 14B is a diagram schematically representing an example method of excluding a detected zero crossing.

[0020] FIG. 14C is a diagram schematically representing an example method of determining a respiration period.

[0021] FIG. 14D is a diagram schematically representing an example method of determining an average respiration rate. [0022] FIG. 14E is a diagram schematically representing an example method of identifying positive and negative half cycles of a sensor signal.

[0023] FIG. 14F is a diagram schematically representing an example method of identifying expiratory and inspiratory phases of a sensor signal.

[0024] FIG. 14G is a diagram schematically representing an example method of determining a confidence factor of a sensor signal.

[0025] FIG. 14H is a diagram schematically representing an example method of determining a confidence factor of an expiratory and/or inspiratory phase.

[0026] FIG. 141 is a diagram schematically representing an example method of applying electrical stimulation in an open or closed loop mode.

[0027] FIG. 14J is a diagram schematically representing an example method of determining whether a zero crossing is a midpoint of a slope of an inspiratory or expiratory phase.

[0028] FIGS. 14K-14M are diagrams schematically representing example methods of determining a current respiration period.

[0029] FIGS. 14N and 140 are diagrams schematically representing example methods of determining an average respiration period.

[0030] FIG. 14P is a diagram schematically representing an example method of determining a median respiration period.

[0031] FIG. 14Q is a diagram schematically representing an example method of resetting a confidence factor.

[0032] FIGS. 14R and 14S are diagrams schematically representing example methods of resetting values in response to detecting a change in posture of the patient.

[0033] FIG. 14T is a diagram schematically representing an example method of detecting a change in posture of the patient.

[0034] FIGS. 14U-14Y are diagrams schematically representing example methods of determining a polarity of a sensor signal.

[0035] FIGS. 15A and 15B are diagrams schematically representing example methods of determining a confidence factor of a sensor signal.

[0036] FIGS. 15C-15E are diagrams schematically representing other example methods of determining a polarity of a sensor signal. [0037] FIG. 15F is a diagram schematically representing an example method of determining a confidence factor of an identified polarity of the sensor signal.

[0038] FIGS. 15G and 15H are diagrams schematically representing example methods of identifying inspiration and expiration half cycles of the sensor signal.

[0039] FIGS. 15I-15R are diagrams schematically representing example methods of scheduling start and stop times of electrical stimulation based on detected zero crossings of the sensor signal.

[0040] FIG. 15S is a diagram schematically representing an example method of switching from a closed loop mode to an open loop mode.

[0041] FIGS. 16A-16G are diagrams schematically representing other example methods of scheduling start and stop times of electrical stimulation based on detected zero crossings of the sensor signal.

[0042] FIGS. 16H and 161 are diagrams schematically representing example methods of selecting a sensor signal for detecting zero crossings.

[0043] FIGS. 16J and 16K are diagrams schematically representing example methods of applying electrical stimulation in an open or closed loop mode.

[0044] FIG. 16L is a diagram schematically representing an example method of adjusting stimulation start and/or stop times based on an electrical stimulation duty cycle.

[0045] FIG. 17A is a diagram including a front view schematically representing a patient’s body including example implantable components and example external elements of example methods and/or example devices.

[0046] FIG. 17B is a diagram including a front view schematically representing an example method and/or example device including a medical device, including an accelerometer, implanted in a patient’s chest.

[0047] FIG. 17C is a diagram schematically representing example patient anatomy and example locations for stimulating an IHM-innervating nerve and/or hypoglossal nerve.

[0048] FIG. 18A is a block diagram schematically representing an example method and/or example device for determining respiration information via a sensed acceleration signal. [0049] FIG. 18B is a block diagram schematically representing an example confidence factor portion.

[0050] FIG. 18C is a block diagram schematically representing an example feature extraction portion.

[0051] FIG. 18D is a block diagram schematically representing an example inspiratory phase prediction function.

[0052] FIG. 18E is a block diagram schematically representing an example noise model parameter.

[0053] FIGS. 19A and 19B each are a block diagram schematically representing an example control portion.

[0054] FIG. 20 is a block diagram schematically representing an example user interface.

Detailed Description

[0055] In the following detailed description, reference is made to the accompanying drawings which form a part hereof, and in which is shown by way of illustration specific examples in which the disclosure may be practiced. It is to be understood that other examples may be utilized and structural or logical changes may be made without departing from the scope of the present disclosure. The following detailed description, therefore, is not to be taken in a limiting sense. It is to be understood that features of the various examples described herein may be combined, in part or whole, with each other, unless specifically noted otherwise.

[0056] At least some examples of the present disclosure are directed to detecting respiration information. In some such examples, the detection of respiration may be based, at least in part, on sensing rotational movement at a respiratory body portion caused by breathing and/or attempted breathing. In some examples, the respiratory body portion may comprise a chest (e.g., chest wall), abdomen (e.g., abdominal wall), and/or other body portion exhibiting rotational movement indicative of respiration or attempted respiration. Throughout the present disclosure, references to breathing and respiration also include, in at least some examples, attempted breathing and attempted respiration, respectively.

[0057] In some examples of the present disclosure, respiration information is detected by sensing changes in acceleration indicative of respiration of a patient to provide a sensor signal. In some examples, respiration information may be detected by sensing changes in impedance and/or pressure (e.g., via internal or external sensors) indicative of respiration of a patient to provide a sensor signal. In some examples, respiration information may be detected by an acoustic sensor or video sensor to provide a sensor signal indicative of respiration of a patient. In some examples, respiration information may be detected by a radar-based sensor, a laser, and/or a light detection and ranging (LIDAR) sensor to provide a sensor signal indicative of respiration of a patient. The sensor signal may be filtered (e.g., bandpass filtered) to detect zero crossings (or other extracted features) of the sensor signal. The zero crossings of the sensor signal may be used to determine respiration information, such as a respiration period, a respiration rate, inspiratory and/or expiratory phases, etc. In some examples, the respiration information may be used for applying electrical stimulation to the patient to treat sleep disordered breathing. In other examples, however, the respiration information may be used for diagnostic, measuring disease burden, therapeutic, etc. purposes without necessarily applying electrical stimulation. The respiration information may also be applied to contexts in addition to, or instead or, care of sleep disordered breathing.

[0058] FIG. 1A is a block diagram illustrating one example of a device 10A for detecting respiration. Device 10A includes a sensor 12A and a control portion 14A electrically coupled to the sensor 12A. The sensor 12A is securable to a patient to provide a sensor signal indicative of respiration of the patient. In some examples, sensor 12A includes an implantable accelerometer to sense changes in acceleration indicative of respiration of the patient. In some examples, sensor 12A includes an external accelerometer to sense changes in acceleration indicative of respiration of the patient. In some examples, sensor 12A may include an implantable or external gyro sensor to sense changes in rotational velocity indicative of respiration of the patient. In some examples, sensor 12A includes an implantable impedance sensor to sense changes in impedance indicative of respiration of the patient, In some examples, sensor 12A includes an implantable or external pressure sensor to sense changes in pressure indicative of respiration of the patient. In some examples, sensor 12A includes an implantable or external acoustic sensor (e.g., microphone) to sense changes in acoustic signals indicative of respiration of the patient. In some examples, sensor 12A includes a video sensor, a radar-based sensor, or laser (or LIDAR) sensor to sense changes in patient movement (e.g., chest movement) indicative of respiration of the patient. The control portion 14A may extract features from the sensor signal to determine each respiratory phase of the patient. In one example, the control portion 14A determines a midpoint (e.g., zero crossing of a bandwidth filtered sensor signal) of each respiratory phase (e.g., inspiratory phase and/or expiratory phase) of the patient based on the sensor signal from sensor 12A. The extracted features may be used by control portion 14A to control (e.g., synchronize, trigger, etc.) the delivery of therapy relative to the appropriate respiratory phase of the patient, for diagnostic storage and/or display, and/or for other purposes.

[0059] In one example, sensor 12A may include a single axis accelerometer having one measurement access to provide one corresponding sensor signal as will be described below with reference to at least FIG. 2. In some examples, sensor 12A may include a multiple axis accelerometer having more than one measurement axis, such as at least two orthogonal measurement axes to provide two corresponding sensor signals or three axes (e.g., a three axis accelerometer) to provide three corresponding sensors signals as will be described below with reference to at least FIGS. 7A-8. Each sensor signal may be filtered by linear filters (e.g., lowpass, highpass, bandpass) or non-linear filters (e.g., median filter) to recover the low-frequency respiration signal while rejecting measurement noise, muscle noise, cardiac noise, and/or other noise.

[0060] FIG. 1 B is a block diagram illustrating one example of a device 10B for detecting respiration and/or applying stimulation based on the detected respiration. Device 10B includes a sensor 12B and a control portion 14B. Sensor 12B may comprise one example implementation of sensor 12A (FIG. 1A) and/or control portion 14B may comprise one example implementation of control portion 14A (FIG. 1A). In this example, sensor 12B senses changes in acceleration indicative of respiration of the patient as indicated at 16.

[0061] Control portion 14A (FIG. 1A) and/or control portion 14B (FIG. 1 B) may comprise one example implementation of control portion 900 in FIG. 19A. Control portion 14B includes portions 18-28 to implement functions based on the sensor signal. At portion 18, control portion 14B may detect zero crossings (and/or extract other features) of the sensor signal (e.g., as illustrated and described below with reference to at least FIGS. 4-5B). If sensor 12B provides more than one signal, zero crossings and/or other features may be detected and/or extracted from each sensor signal independently. In one example, the zero crossings and/or other features detected and/or extracted from one sensor signal may be selected for identifying each respiratory phase. The sensor signal may be selected based on a confidence factor to be described below, may be predetermined, or may be selected by a clinician. In some examples, the zero crossings and/or other features detected and/or extracted from each sensor signal may be combined to provide a composite signal (e.g., average of zero crossings and/or other features, median of zero crossings and/or other features, etc.) as will be described below with reference to at least FIG. 8.

[0062] In one example, the detected zero crossings include a midpoint of each respiratory phase of the patient, such as the midpoint 153 of each inspiratory phase and/or the midpoint 152 of each expiratory phase as will be described below in more detail at least with reference to FIGS. 5A and 5B. In some examples, the extracted features include the curvature of each peak (e.g., 148 of FIG. 4) versus each trough (e.g., 146 of FIG. 4) of the sensor signal. In some examples, the extracted features include relative peak-to-trough timing (e.g., from 146 to 148 and from 148 to 146 of FIG. 4) and/or peak-to-trough slope of the sensor signal. In some examples, the extracted features include the relative slope of the sensor signal between peak and trough (e.g., from 148 to 146 of FIG. 4) versus between trough and peak (e.g., 146 to 148 of FIG. 4).

[0063] At portion 20, control portion 14B may identify a polarity of the sensor signal. The polarity of the sensor signal may be used to identify each respiratory phase of the patient. As described in further detail later with reference to at least FIGS. 5A and 5B, the polarity of the sensor signal indicates whether the slope of the inspiratory phase is positive or negative and/or whether the slope of the expiratory phase is negative or positive. One example of identifying whether the slope of each expiratory phase is positive or negative is based on the detected midpoints and will be described below at least with reference to FIGS. 5A and 5B. In some examples, control portion 14B may identify whether the slope of each inspiratory phase is positive or negative. One example of identifying whether the slope of each inspiratory phase is positive or negative is based on the detected midpoints and will be described below at least with reference to FIGS. 5A and 5B. Identifying the polarity of the sensor signal is further described later with reference to at least FIGS. 14U-14Y and 15C-15E.

[0064] At portion 22, control portion 14B may identify inspiratory and expiratory half cycles of the sensor signal. In some examples, the control portion 14B identifies each expiratory half cycle and/or each inspiratory half cycle based on the polarity of the sensor signal identified at portion 20. In one example, the midpoints (e.g., zero crossings of a highpass filtered or bandpass filtered sensor signal) of each respiratory phase may be used to identify each inspiratory phase and/or each expiratory phase since the time between the midpoint of an expiratory phase followed by the midpoint of an inspiratory phase is typically longer than the time between the midpoint of an inspiratory phase followed by the midpoint of an expiratory phase (e.g., due to an expiratory pause of each expiratory phase).

[0065] In some examples, the feature of curvature of peak versus trough may be used to identify each inspiratory phase and/or each expiratory phase since the peak at an inspiratory phase to expiratory phase transition (e.g., 148 of FIG. 4) will typically have a sharper peak than the peak at an expiratory phase to inspiratory phase transition (e.g., 146 of FIG. 4). That is, the transition from inhalation to exhalation is typically much quicker than the transition from exhalation to inhalation. In some examples, the feature of relative peak-to-trough timing and/or peak-to-trough slope may be used to identify each inspiratory phase and/or each expiratory phase since the inspiratory phase is typically shorter than the expiratory phase, such that relative phase may be determined. In some examples, the features of relative slope of the sensor signal between peak and trough versus between trough and peak may be used to identify each inspiratory phase and/or each expiratory phase since the slope of the expiratory phase (e.g., 144B of FIG. 4) is typically smaller than the slope of the inspiratory phase (e.g., 144A of FIG. 4). Identifying inspiratory and expiratory half cycles is further described later with reference to at least FIGS. 14V and 15G-15H.

[0066] At portion 24, control portion 14B may determine a confidence factor for the sensor signal. Control portion 14B may determine the confidence factor associated with the identified polarity from portion 20 and/or the identified inspiratory half cycle and/or the identified expiratory half cycle from portion 22. The confidence factor is an estimated probability of correctness of the polarity, identified half cycle, and/or respiratory phase determination. The confidence factor may be determined for each extracted feature and/or each signal/axis as described in more detail below and with reference to at least FIG. 18B.

[0067] In one example, control portion 14B may determine the confidence factor associated with each respective identified expiratory half cycle and/or identified inspiratory half cycle from portion 22 using a statistical test. The statistical test may be used for an inspiration/expiration timing ratio not equal to 1 .0 (i.e. , relative duration is not 1 :1 ) over one or more historical median calculations using the historical median calculation data points as the sample population for a student’s T-test (or other statistical test) for a difference from a mean of 0. This statistical test uses both the sign and magnitude information from the median calculations to inform the confidence level (e.g., noisy signal centered on 0 would have low confidence, a clean signal that is either strongly positive or strongly negative would have high confidence). The t-test (or other statistical test) may be based on a lookup table and may optionally include a saturation or extrapolation at high and/or low confidence levels. In some examples, the confidence factor may be partly or fully determined based on a signal to noise ratio, relative signal amplitude between axes, and/or absolute signal amplitude. In some examples, the confidence factors may be programmatically biased (e.g., by design or by a clinician) for each axis, such as to modify and/or override other confidence factors.

[0068] In some examples, confidence factors may be determined based on an expected number and/or timing of zero crossings (e.g., used to extract midpoints of respiratory phases). In one example, an expected number of zero crossings may be based on a running respiration rate, such as the expected number of zero crossings in an estimated respiration period. In some examples, an expected number of zero crossings may be fixed by a physiologic interval (e.g., expect no more than 3 zero crossings per a physiologically realistic interval). In some examples, confidence factors may be determined based on the expected timing from one zero crossing to the next zero crossing. In some examples, a violation of the expected number of zero crossings and/or the timing of zero crossings may be termed an “excess zero” and may be rejected on that basis. In some examples, excess zeros may be used to inform the confidence factor (e.g., more excess zeros equals lower signal quality) or may itself be used as a confidence factor. More excess zeros may occur in one respiratory phase than in the other respiratory phase (e.g., an expiratory phase may have more excess zeros than an inspiratory phase), which may be used to inform the confidence in the current polarity determination described below with reference to at least FIGS. 3-5B.

[0069] In some examples, the control portion 14B may compute a confidence factor as a statistical test for a median of the sensor signal(s) (e.g., data points of the sensor signals(s)) being different than a mean of the sensor signal(s). In some examples, the control portion 14B may compute a confidence factor based on a magnitude of an AC component of the sensor signal(s). In some examples, the control portion 14B may compute a confidence factor as a magnitude of an AC component of the sensor signal(s) divided by a root mean square of the sensor signal(s). In some examples, the control portion 14B may compute a confidence factor as a magnitude of a first harmonic of the sensor signal(s) divided by a sum of higher order harmonics of the sensor signal(s). In some examples, the control portion 14B may compute a confidence factor as a ratio of an AC component of the sensor signal(s) to a noise floor of the sensor 12B. In some examples, the control portion 14B may compute a confidence factor based on the average amplitude of the sensor signal(s) (e.g., a bandpass filtered sensor signal). If the average amplitude is below a threshold, the confidence factor may be set to zero. In some examples, the control portion 14B may compute a confidence factor as a linear or non-linear combination of any of the above example confidence factors.

[0070] In some examples, the control portion 14B determines a confidence factor associated with each of three respective sensor signals (e.g., from a three axis accelerometer), identifies which respective sensor signal exhibits the greatest confidence factor, and determines the midpoint (or other features) of each respiratory phase of the patient based on the identified sensor signal exhibiting the greatest confidence factor. Determining a confidence factor associated with a sensor signal is further described later with reference to at least FIGS. 14G- 14H, 15A-15B, 15F, and 16H-16I.

[0071] At portion 26, control portion 14B may schedule the start and stop of stimulation based on the detected zero crossings from portion 18, the identified polarity from portion 20, and/or the identified inspiratory and expiratory half cycles from portion 22. In one example, control portion 14B predicts a midpoint (e.g., zero crossing) of a future inspiratory phase based on previous midpoints of each respiratory phase, which may be used to schedule the start and/or stop of stimulation. In some examples, the midpoint of a future inspiratory phase may be predicted based on one or more previous respiratory cycles and the respiratory rate of one or more previous respiratory cycles. For example, the midpoint of a future inspiratory phase may be predicted based on the mean of previous midpoints, the median of previous midpoints, a linear extrapolation of previous midpoints, or a non-linear extrapolation of previous midpoints. Previous values, which are advantageous as “look back” techniques, may be used to improve accuracy of feature extraction and to reduce susceptibility to a noisy signal (i.e., non-respiratory transients) due to limb movement, bed partner movement, etc. For example, zero crossing detection (e.g., as one example of midpoint detection for a highpass or bandpass filtered signal) accuracy may be improved by using data points before and after the zero crossing. In some examples, the midpoint of a future inspiratory phase may be predicted based on similarity (e.g., crosscorrelation) with a reference respiratory morphology.

[0072] With the midpoint (e.g., zero crossing) of the future inspiratory phase predicted, the beginning of the future inspiratory phase (which may be used to schedule the start and/or stop of stimulation so that stimulation is applied during the inspiratory phase) may be predicted by subtracting a first predetermined interval from the predicted midpoint and the end of the future inspiratory phase may be predicted by adding a second predetermined interval to the predicted midpoint. In some examples, the future inspiratory phase may be predicted based on other previously extracted features, such as previous curvatures of peak versus trough, previous relative peak-to-trough timing and/or peak-to-trough slope as previously described. Scheduling the start and stop of stimulation is further described later with reference to at least FIGS. 15I-16G.

[0073] At portion 28, control portion 14B may apply stimulation. In one example, device 10B may include an electrode (e.g., electrode 672 to be described with reference to at least FIG. 17B) to deliver electrical stimulation to an upper airway patency-related nerve of the patient or another nerve of the patient. In one example, the control portion 14B may apply the electrical stimulation based on the predicted midpoint (e.g., zero crossing) of the future inspiratory phase. The control portion 14B may apply the electrical stimulation with a time or ratio-based offset from the predicted midpoint of the future inspiratory phase. For example, the control portion 14B may apply the electrical stimulation starting relative to the predicted midpoint by a first predetermined interval and ending after the predicted midpoint by a second predetermined interval. The first predetermined interval may be selected based on a preselected, or computed, relative position (e.g., 20% or 30%) within the mean or median of previous inspiratory phase durations plus optionally an absolute additional amount (e.g., 200 ms or 300 ms) to ensure the stimulation is started just prior to the inspiratory phase. The second predetermined interval may be selected based on a preselected, or computed, relative position within the mean or median of previous inspiratory phase durations plus optionally an absolute additional amount to ensure the stimulation is ended at or just after the inspiratory phase. The relative position may be responsive to optimize therapy based on detected sleep disordered breathing like obstructive sleep apnea, central sleep apnea, multiple-type sleep apneas, etc., and may be moved forwards or backwards from the relative position to optimize the efficacy of the stimulation. In some examples, control portion 14B may apply stimulation based on future inspiratory phases predicted based on other previously extracted features, such as previous curvatures of peak versus trough, previous relative peak-to-trough timing and/or peak-to-trough slope as previously described. Additional examples of the application of stimulation will be described below with reference to FIGS. 5A-5B.

[0074] In some examples, the stimulation portion 28 controls stimulation of target tissues, such as but not limited to an upper airway patency-related nerve (e.g., hypoglossal nerve, infrahyoid muscle (IHM)-innervating nerve, other and/or other respiratory-related targets such as a phrenic nerve, to treat sleep disordered breathing (SDB) behavior. Further details regarding the IHM-innervating nerve is provided below in association with at least FIG. 17C. In some examples, the stimulation portion 28 comprises a closed loop parameter to deliver stimulation therapy in a closed loop manner such that the delivered stimulation is in response to and/or based on the sensor signal(s).

[0075] In some examples, the closed loop parameter may be implemented using the sensed information to control the particular timing of the stimulation according to respiratory information, in which the stimulation pulses are triggered by or synchronized with specific portions (e.g., inspiratory phase) of the patient’s respiratory cycle(s), in some examples. In some such examples and as previously described, this respiratory information may be determined via the sensor 12A or 12B.

[0076] In some examples in which the sensor signal enables determining at least respiratory phase information, the closed loop parameter may be implemented to initiate, maintain, pause, adjust, and/or terminate stimulation therapy based on (at least) the determined respiratory phase information. In some examples, the stimulation is started prior to an onset of the inspiratory phase and the stimulation is stopped exactly at the end of the inspiratory phase or stopped just after the end of the inspiratory phase. [0077] In some examples the stimulation portion 28 comprises an open loop parameter by which stimulation therapy is applied without a feedback loop. In some such examples, in an open loop mode the stimulation therapy is applied during a treatment period without (e.g., independent of) information sensed regarding the patient’s sleep quality, sleep state, respiratory phase, and/or AHI, etc. In some such examples, in an open loop mode the stimulation therapy is applied during a treatment period without (i.e., independent of) particular knowledge of the patient’s respiratory cycle information. However, in some such examples, some sensory feedback may be utilized to determine, in general, whether the patient should receive stimulation based on a severity of sleep apnea behavior.

[0078] In some examples the stimulation portion 28 comprises an auto-titration parameter by which an intensity of stimulation therapy can be automatically titrated (i.e., adjusted) to be more intense (e.g., higher amplitude, greater frequency, and/or greater pulse width) or to be less intense within a treatment period. In some such examples, such auto-titration may be implemented based on sleep quality, which may be obtained via sensed physiologic information, in some examples. It will be understood that such examples may be employed with synchronizing stimulation to sensed respiratory information (e.g., closed loop stimulation) or may be employed without synchronizing stimulation to sensed respiratory information (e.g., open loop stimulation). In some examples, at least some aspects of the auto-titration parameter may comprise, and/or may be implemented, via at least some of substantially the same features and attributes as described in Christopherson et al., US 8,938,299, SYSTEM FOR TREATING SLEEP DISORDERED BREATHING, issued January 20, 2015, and which is hereby incorporated by reference in its entirety.

[0079] In some examples, the stimulation portion 28 of control portion 14B may comprise an “off period” function by which a user or clinician may adjust the time that stimulation will remain off and which may be expressed as a percentage of the previous “on period.” In some examples, the “off period” for stimulation coincides with the expiratory active phase. However, in some examples, once enabled by the user or clinician, the “off period” (i.e., no stimulation) setting is implemented regardless of detected phases. In some examples, the “on period” may sometimes be referred to as a stimulation period and the “off period” may sometimes be referred to a non-stimulation period. In some examples, a stimulation cycle may comprise one stimulation period followed by one nonstimulation period. In some examples, the stimulation cycle may have a duration corresponding to a duration of a respiratory cycle.

[0080] In some examples, the stimulation portion 28 of control portion 14B may comprise a “maximum stimulation” function which may be used by a patient or clinician to adjust a maximum time for an “on period” of stimulation for a given stimulation cycle, after which an “off period” takes place. The “on period” may extend for a selectable, predetermined period of time. In some examples, the “on period” for stimulation coincides with (or at least partially coincides with) the inspiratory phase for at least some respiratory cycles. In some such examples, once enabled by the user or clinician, the “on period” of stimulation is implemented regardless of detected phases. Applying stimulation is further described later at least with reference to FIGS. 14A, 141, and 16J-16L.

[0081] FIG. 2 is a diagram including a side view, schematically representing an example method 100 and/or example device including a sensor 104A. In one example, sensor 104A may provide sensor 12A or 12B of FIGS. 1A and 1 B. As shown in FIG. 2, example sensor 104A is chronically, subcutaneously implanted within a chest wall 102A of a patient’s body. During breathing, the chest wall 102A will exhibit rotational movement (B2) as at least some portions of the chest wall 102A move (e.g., rise and fall) during inspiration and expiration, wherein inspiration corresponds to expansion of the rib cage and expiration corresponds to contraction of the rib cage. During this expansion and contraction of the rib cage during breathing, at least some portions of the chest wall 102A exhibit rotational behavior, which may in turn may be sensed upon the sensor 104A experiencing such rotational movement (as represented via directional arrow B1 ), which in turn provides respiration information as described above and as further described below. It will be understood, as further described later, that the rotational movement of the sensed portion of the chest wall 102A is not necessarily or strictly limited to rotational movement in a single plane. In some examples, sensor 104A may comprise a portion of a larger device, as further described later in association with at least FIG. 17B.

[0082] As further shown in FIG. 2, the sensor 104A may sense the rotational movement of at least a portion of the chest wall 102 (as represented via directional arrow B2) relative to an earth gravitational field (arrow G), i.e. , gravity vector. For illustrative simplicity to depict at least some examples, FIG. 2 depicts the chest wall 102A as if the patient’s body was in a generally horizontal sleep position. It will be understood that at least some example devices and/or example methods will be effective in detecting respiration information regardless of whether the generally horizontal sleep position is a supine position, a prone position, or a side-laying (i.e., lateral decubitus) position. Moreover, at least some example methods and/or example devices also will be effective in detecting respiration information if, and when, the patient is in positions other than a generally horizontal position, such as sitting in a chair in a vertically upright position, in a reclining position, etc. as further described later in association with at least FIGS. 9A-10.

[0083] Moreover, determining respiration information via acceleration-based sensing of rotational movements (at a portion of a chest wall of the patient) does not include, or depend on, determining (e.g., via sensing) a body position of the patient. Accordingly, while such respiration information may be determined in any one of several different sleeping body positions, such determination may be performed without determining the particular sleeping body position at the time the sensing of the rotational movements is being performed.

[0084] However, in some such examples, the particular sleeping body position occurring at the time of the determining the respiration information (via acceleration-based sensing of rotational movements of a portion of a chest wall during breathing) may be determined and used as a parameter to augment the determined respiration information and/or other general patient physiologic information, in some instances.

[0085] In some examples, securing the implantable acceleration sensor(s) comprises mechanically coupling the sensor(s) relative to a respiratory body portion. In some examples, securing the implanted acceleration sensor(s) comprises securing the acceleration sensor relative to tissue on top of a muscle layer of the respiratory body portion, while in some examples the sensor may be secured directly to a muscle layer of the respiratory body portion. In some examples, the acceleration sensor may be secured subcutaneously within the respiratory body portion without securing the acceleration sensor on the muscle layers. In some examples, the respiratory body portion may comprise the chest. In some such examples, the respiratory body portion may comprise a portion of the chest, such as but not limited to a portion of a chest wall. In some instances, the portion of the chest wall may correspond to a portion of the rib cage. In some examples, such aspects of securing the sensor(s) relative to a muscle layer or subcutaneously are also applicable to securing the sensor at other respiratory body portions, such as an abdomen (e.g., abdominal wall) by physically (e.g., mechanically) coupling the sensor relative to the abdomen to sense rotational movement at the abdomen during breathing.

[0086] FIG. 3 is a diagram 120, including a side view, schematically representing an example method and/or example sensor 104A. As shown in FIG. 3, in some examples the sensor 104A may comprise a sensing element 122A, which is arranged to measure an inclination angle (Q) upon rotational movement of the sensing element 122A caused by breathing. As represented in FIG. 3, upon rotational movement of at least a portion of the chest wall 102A during breathing, the sensing element 122A may rotationally move between a first angular orientation YR1 (shown in solid lines) and a second angular orientation YR2 (shown in dashed lines). In some such examples, the first angular orientation YR1 (shown in solid lines) of sensing element 122A may correspond to a peak expiration of a respiratory cycle (e.g., rib cage contracted) and the second angular orientation YR2 (shown in dashed lines) of sensing element 122A may correspond to a peak inspiration of the respiratory cycle (e.g., rib cage expanded). [0087] With reference to at least FIG. 3, it will be understood that the sensing element 122A moves with at least a portion of the chest wall 102A as depicted in dashed lines. Accordingly, sensing element 122A does not move relative to the chest wall 102A, but rather the sensing element 122A rotationally moves along with (e.g., in synchrony with) the rotational movement of at least the portion of the chest wall 102A (in which the sensor 104A, including sensing element 122A), is implanted) during breathing.

[0088] In some examples, the sensing element 122A comprises an accelerometer, which may comprise a single axis accelerometer in some examples or which may comprise a multiple-axis accelerometer in some examples. Via the accelerometer, the sensing element 122A can determine absolute rotation of sensor 104A (and therefore rotation of the portion of the chest wall 102A) with respect to gravity (e.g., earth gravity vector G), rather than instantaneous changes in rotation. In some examples, element 122A may comprise a single axis accelerometer to measure (at least) a value of, and changes in the value of, the above-noted inclination angle (Q) associated with movement of at least a portion the chest wall 102 caused by breathing.

[0089] FIG. 4 is a diagram including a graph schematically representing a filtered acceleration signal 142 sensed via a sensor, such as sensing element 122A in FIG. 3. As shown in FIG. 4, signal 142 corresponds to a respiratory waveform exhibited through several respiratory cycles during breathing. Each respiratory cycle 143 comprises an inspiratory phase (Ti) 144A, an expiratory active phase (TEA) 144B, and an expiratory pause phase (TEP) 144C. An expiratory phase includes both the expiratory active phase and the expiratory pause phase. It will be understood that the example respiratory waveform in FIG. 4 represents a typical respiratory waveform for at least some patients during normal breathing, but not necessarily for all patients at all times.

[0090] With FIG. 4 in mind, and with further reference to FIG. 3, it will be noted that the first angular orientation YR1 of sensing element 122A (shown in solid lines) may correspond generally to a peak expiration 146 (e.g., end of the expiratory active phase (TEA)) while the second angular orientation YR2 of sensing element 122A (shown in dashed lines) may correspond to a peak 148 of inspiratory phase (Ti), i.e., the end of inspiration just at or before the onset of expiration.

[0091] With further reference to FIG. 3, upon rotation (B2) of at least a first portion of chest wall 102A, as represented by directional arrow B2 and the depiction of chest wall in dashed lines 102B, such as during inspiration, the sensing element 122A rotates by the inclination angle (Q) with chest wall 102A to a position or orientation YR2 shown in dashed lines 122B. Upon the end of inspiration, and the ensuing expiration, the chest wall 102 will rotate back into the position shown in solid lines (e.g., end of expiration) such that the sensing element 122A will sense a change in inclination angle (Q) from the position YR2 (shown in dashed lines) back to the position YR1 (shown in solid lines).

[0092] In sensing the inclination angle (Q) through successive respiratory cycles, the sensing element 122A obtains an entire respiratory waveform, which may comprise information such as the duration, magnitude, etc. of an inspiratory phase (Ti), expiratory active phase (TEA), and expiratory pause phase (TEP) of respiratory cycles of the patient, and/or other information (e.g., respiratory rate, etc.) as represented in FIG. 4 and/or as further described later. With this in mind, in some examples the obtained respiratory waveform (e.g., respiration morphology) also comprises a respiratory period, which includes the inspiratory phase, the expiratory active phase, and the expiratory pause phase. In one aspect, the respiratory period corresponds to a duration of a respiratory cycle, with this duration comprising a sum of a duration of the inspiratory phase, a duration of the expiratory active phase, and a duration of the expiratory pause phase.

[0093] In some examples, the identified respiration morphology comprises identifying (within the respiratory waveform morphology) a start of the inspiratory phase, i.e. , onset of inspiration. In some examples, this start of the inspiratory phase also may at least partially correspond to an expiration-to-inspiration transition. In some examples, a method of identifying the start of the inspiratory phase within the identified respiratory waveform morphology further comprises performing the identification (of the start of the inspiratory phase) without identifying an end (e.g., offset) of the inspiratory phase, which may enhance the accuracy of identification (of the start of the inspiratory phase) in the presence of noise, in contrast to identification of more than one phase transition (e.g., inspiratory-to-expiratory or expiratory-to-inspiratory) per respiratory cycle where each transition is subject to misidentification due to noise. In some such examples, the end (e.g., offset) of the inspiratory phase corresponds to a start (e.g., onset) of the expiratory active phase.

[0094] In some examples, at least some aspects of such identification, prediction, etc. of features (e.g., start of inspiratory phase, end of expiration, etc.) within a respiratory waveform may be implemented via at least some of substantially the same features and attributes as later described in association with at least FIGS. 18A-18E and/or various examples throughout the present disclosure, such as but not limited to identifying inspiratory phase (e.g., 752 in FIG. 18A), inspiratory phase prediction (e.g., 860 in FIG. 18D), etc.

[0095] With further reference to FIG. 3, it will be understood that the second angular orientation YR2 of sensing element 122A is not a fixed position, but rather corresponds to a temporary position at one end (e.g., a second end) of a range of rotational movement of the portion of the chest wall 102A, such as peak inspiration 148 (FIG. 4) during breathing. This second end of the range of rotational movement (and therefore the second angular orientation YR2) may vary depending upon whether the patient is exhibiting normal/relaxed breathing, forced breathing (such as more forceful inspiration), and/or disordered breathing. Moreover, this second end of the range of rotational movement may exhibit some variance from breath-to-breath even during relaxed breathing.

[0096] Similarly, the first angular orientation YR1 of sensing element 122A shown in FIG. 3 does not comprise a fixed position, but rather the first angular orientation YR1 corresponds to a temporary position at an opposite other end (e.g., a first end) of a range of rotational movement of the portion of the chest wall 102A, such as peak expiration 146 (FIG. 4) during breathing. This first end of range of rotational movement (and therefore the first angular orientation YR1 ) may vary depending upon whether the patient is exhibiting normal/relaxed breathing, forced breathing (such as more forceful expiration), and/or disordered breathing. Moreover, this first end of the range of rotational movement may exhibit some variance from breath-to-breath even during normal/relaxed breathing.

[0097] While there may be some variances in the exact rotational positions of the respective second angular orientation YR2 and/or the first angular orientation YR1 , it will be understood that there consistently will be a significant difference (i.e. substantial difference) between the first angular orientation YR1 and the second angular orientation YR2, by which respiration morphology (e.g., shown in FIG. 4) can be determined as the sensor moves through the full range of rotational movement between the first and second angular orientations, YR1 and YR2 during patient breathing. Moreover, the variances in the particular rotational position of the first angular orientation YR1 , and of the second angular orientation YR2, at the ends of the range of rotational movement of the sensing element 122A may yield valuable information regarding variances in respiration, such as variances in amplitude of inspiration and/or expiration, variances in respiratory rate, etc. In some examples, the ends of the range of angular movement between the two orientations YR1 , YR2 may correspond to the ends of a range of values of an AC signal component of the acceleration signal from the sensor.

[0098] In some examples, the first angular orientation YR1 may sometimes be referred to as a reference angular orientation, at least to the extent that the first angular orientation YR1 may correspond to an orientation which is the closest to being generally perpendicular to the gravity vector G for at least some sleeping body positions, such as but not limited to a generally horizontal sleep position.

[0099] With further reference to FIG. 3, in this example implementation and in general terms, the sensor 104A may be implanted in a manner to cause the first angular orientation YR1 (i.e., base orientation) of the measurement axis of the sensing element 122A to be generally parallel to a superior-inferior (S — I) orientation of at least the chest region of the patient’s body, and generally perpendicular to an earth gravitational field G, such as when the patient is in a generally horizontal position. In this example implementation, the measurement axis of the sensing element 122A also may be understood as having an orientation generally perpendicular to an anterior-posterior (A — P) orientation of at least a portion of the chest region 102A of the patient’s body.

[0100] In the example of FIG. 3, rotational movement of sensing element 122A, which has a Y-axis orientation, occurs roughly near or within a plane P1 defined by the anterior-posterior orientation (A — P) and by the superior-inferior orientation (S — I) of the patient’s body. This rotational movement is primarily indicative of rotational movement of the rib cage during breathing, such as during a treatment period in which a patient is sleeping. Additional examples later describe additional/other aspects in which rotational movements of the rib cage are further indicative of breathing, and therefore respiratory morphology.

[0101] It will be understood that, due to patient-to-patient variances in anatomy and/or due to the particular location where the sensor 104A is actually implanted along the chest wall 102A, the sensing element 122A may extend in an orientation which is not exactly parallel to a superior-inferior orientation of the chest wall (or entire patient as a whole), and not exactly perpendicular to an anterior-posterior orientation of the patient’s body (and gravity vector G when laying in a generally horizontal position).

[0102] Nevertheless, in some example implementations, by arranging the measurement axis of the sensing element 122A to have an orientation as close as possible to being generally perpendicular to the gravity vector (G) for at least some patient body positions (e.g., generally horizontal sleep position), the sensitivity of the AC signal component of the acceleration sensing element 122A is maximized (and absolute value of the DC signal component is minimized), which in turn may increase the effectiveness of measuring changes in the inclination angle (Q) of sensing element 122A caused by, and during, breathing by the patient. In one aspect, the AC signal component of the acceleration sensing element 122A may be understood as the time-varying portion of the output signal of the acceleration sensing element 122A.

[0103] In particular, by arranging the sensing element 122A within the chest wall 102A to be as close as reasonably practical to being generally perpendicular to the earth gravitational field G (at least when the patient is in a primary sleep position), the sensed inclination angle will correspond to a maximum value of a measured AC component of the acceleration signal and a minimum absolute value of a measured DC component of the acceleration signal. Stated differently, when a measurement axis of the acceleration sensing element 122A is generally perpendicular (or as close as reasonably practical) to an orientation in which it would otherwise measure a maximum value (e.g., 1 g, such as when parallel to an earth gravity vector), the absolute value of the DC component will be negligible or minimal. In this situation, changes in value of the AC component of the acceleration signal become more prominent, being of a magnitude and/or reflecting significantly measurable changes as an orientation of the (measurement axis of the) acceleration sensing element changes (e.g., inclination angle) as the portion of the chest wall exhibits rotational movement caused by breathing.

[0104] It will be understood that based on the particular orientation at which the sensor 104A (e.g., sensing element 122A) is actually implanted, based on the varying position of the patient once the sensor 104A has been implanted, and/or based on other factors described further below, the measurement axis of the sensing element 122A at the chest wall 102A may not be perpendicular to the earth gravitational field G at the time of performing the sensing during breathing and hence the sensitivity of the AC component of the acceleration signal may not be at a maximum value. Nevertheless, at least some example methods (and/or devices) may perform the sensing (e.g., of the inclination angle of the sensing element 122A) to obtain the desired respiration information provided that the sensed signal provides a sufficiently high degree of sensitivity of a measured AC component of the acceleration signal. In some such examples, the methods and/or devices may employ magnitude criteria by which it may be determined if, and/or when, a sufficiently high degree of sensitivity of the measured AC signal is present. For instance, in some non-limiting examples, a sufficiently high degree of sensitivity corresponds to a measured AC signal having adequate signal to noise ratio in order to determine respiration.

[0105] In some such examples, an output acceleration signal of sensing element 122A corresponds to a sine of the angle between the accelerometer measurement axis (i.e., orientation of Y) and a generally horizontal orientation (which is generally perpendicular to gravity vector G).

[0106] Because of variances in patient-to-patient anatomy, in some example methods/devices, an absolute magnitude of the AC signal component is not used to determine respiration information. Rather, by using the difference in magnitude of the value of the AC signal component between the first angular orientation (YR1 ) and the second angular orientation (YR2), the example methods/devices can determine a respiratory waveform, morphology, etc. [0107] In some examples, depending on the particular angle at which the device and sensor are implanted in a particular patient, and/or depending on the particular sleeping position in which the patient is arranged, the inspiration identified from the sensed respiratory waveform may have a positive slope or may have a negative slope. In some such examples, the positive slope or the negative slope of the inspiratory phase may sometimes be referred to as a polarity of the slope of the inspiratory phase. It will be understood that in accounting for the particular slope of the inspiratory phase, the slope of the other phases of the respiratory cycle will be accounted for as well.

[0108] With these features in mind regarding the slope of the inspiratory phase of the respiratory waveform signal, at least some of the example methods and/or devices of the present disclosure may accurately capture and determine respiratory information regardless of how the patient may be moving in space, e.g., regardless of the direction of the sensor rotation in space or regardless of rotation of the patient (including the sensor) with respect to gravity. Accordingly, the example methods and/or devices may produce accurate, reliable determination of respiration information.

[0109] Accordingly, at least some example methods comprise implanting sensor 104A (including sensing element 122A) in a manner to maximize sensitivity of the AC component of the sensed acceleration signal by establishing an orientation (e.g., YR1 ) of sensing element 122A which is closest to being generally perpendicular to the gravity vector G, for at least some body positions such as a common sleep position (e.g., generally horizontal). In some situations, the sensor 104A (including sensing element 122A) may be implanted in a position in which the sensitivity of the AC component of the sensed acceleration signal is not maximized but which is sufficient to effectively and reliably determine respiration information based on sensed rotational movement at a first portion of a chest wall (or other physiologic location as described below). In some such examples, a sufficient sensitivity of the AC component of the sensed acceleration signal may comprise having an adequate signal-to-noise ratio.

[0110] With further regard to the observation that the device (including the acceleration sensor(s)) may be implanted at various particular orientations (e.g., angles) which are not parallel to an ideal reference orientation (e.g., superiorinferior), it will be understood that in some examples, the example methods/devices determine the respiration information (e.g., using accelerationbased sensing of rotational movement of a portion of a chest wall, etc.) without calibrating the measured inclination angle signal (of the acceleration sensor) relative to any difference between the ideal reference orientation (e.g., superiorinferior) and the actual implant orientation. However, in some examples, such calibration may be performed and/or such differences may be considered in using the sensed information.

[0111] As noted elsewhere, in at least some examples, determining respiration information based on acceleration sensing of rotational movement (of a portion of the chest wall) does not depend on the sensor having an ideal implant orientation (e.g., with respect to the gravity vector or with respect to the patient’s body), does not depend on knowing the actual implant orientation (e.g., with respect to the gravity vector or with respect to the patient’s body), and/or does not depend on accounting for differences between the ideal implant orientation and the actual implant orientation.

[0112] In some later example implementations, a sensor comprises multiple sensing elements such that the example methods may comprise determining which of the multiple sensing elements has an orientation which is closest to being generally perpendicular to the gravity vector, and therefore which may provide the most sensitivity and effectiveness in sensing respiratory information. In some such examples, the multiple sensing elements may be oriented orthogonally relative to each other or may be oriented at other angles (e.g., 45 degrees) relative to each other.

[0113] With further reference to at least FIG. 3, in some examples, the term “generally perpendicular” may comprise the first angular orientation YR1 being at some angle relative to the gravity vector G (e.g., 85, 86, 87, 88, 89, 91 , 92, 93, 94, 95 degrees) which varies slightly from an exactly perpendicular angle (e.g., 90 degrees) relative to the gravity vector G. Moreover, as noted above and/or further described below, the effectiveness of measuring respiration by changes in the inclination angle (Q) between the first and second orientations (YR1 , YR2) does not strictly depend on the first angular orientation YR1 being exactly perpendicular to the gravity vector G.

[0114] However, the first angular orientation YR1 may be at angles other than generally perpendicular relative to the gravity vector (G), such as in example implementations in which the first angular orientation YR1 of a sensing element (e.g., 122A) is positioned to be about 135 degrees relative to the gravity vector G (i.e. , 135 degrees to an anterior-posterior (A — P) orientation of patient’s body. In some such examples, the second angular orientation YR2 of sensing element 122A would still extend at an angle (Q) relative to the first angular orientation YR1 , with it being understood that angle (Q) varies according to the variances in respiration of the patient which occur in normal breathing, forced breathing, and/or disordered breathing, as previously described. As further described later, establishing the first orientation YR1 at angles other than 135 degrees are contemplated as well.

[0115] As further described later, in noise model parameter 870 in FIG. 18E the example device(s) and/or example method(s) may perform such measurements in a manner to exclude (e.g., filter) measurements of gross body motion, measurement noise, muscle noise, cardiac noise, other noise, etc. such that the remaining sensed or measured acceleration signal is primarily representative of movement of at least a portion of the chest wall 102. In some such examples, the measured acceleration signal is representative solely of movement of the chest wall 102. In particular, in some such examples, the measured acceleration signal corresponds to rotational movements of at least a portion of the chest wall 102 as sensed by sensor 104A (B1 in FIG. 2) caused by and/or occurring during breathing.

[0116] FIGS. 5A and 5B are diagrams including graphs schematically representing filtered acceleration signals 142A and 142B sensed via a sensor, such as sensing element 122A in FIG. 3, having two different orientations (e.g., due to patient body position), respectively. The filtered acceleration signal 142A of FIG. 5A has a similar orientation as filtered acceleration signal 142 of FIG. 4, while filtered acceleration signal 142B of FIG. 5B has the opposite orientation from filtered acceleration signal 142A of FIG. 5A. In some examples, the filtered acceleration signal 142A of FIG. 5A may be referred to as having a positive polarity since the slope of the inspiratory phase 144A is positive (and the slope of the expiratory phase 144B is negative), while filtered acceleration signal 142B of FIG. 5B may be referred to as having a negative polarity since the slope of the inspiratory phase 144A is negative (and the slope of the expiratory phase 144B is positive). As shown in FIGS. 5A and 5B, signals 142A and 142B correspond to a respiratory waveform exhibited through several respiratory cycles during breathing. Each respiratory cycle 143 comprises an inspiratory phase 144A, an expiratory active phase 144B, and an expiratory pause phase 144C. In FIG. 5A, the slope of the expiratory phase 144B is negative as indicated at 155A, and in FIG. 5B, the slope of the expiratory phase 144B is positive as indicated at 155B. [0117] The graphs of FIGS. 5A and 5B include a signal midpoint (e.g., zero crossing) as indicated at 150 and a current time as indicated at 151. The signal midpoint 150 may be provided by a threshold of a bandpass filter, such that zero crossings of the sensor signal indicate the midpoints of each respiratory phase. The bandpass filter may be auto-centering, such that the signal midpoint 150 is maintained despite patient movement that results in changes to the sensor signal. As shown in FIGS. 5A and 5B, the midpoint of each expiratory phase (TE.MID) 152 occurs at the zero crossing of each expiratory phase, and the midpoint of each inspiratory phase (TI.MID) 153 occurs at the zero crossing of each inspiratory phase. The detection of zero crossings optionally includes hysteresis to prevent false triggering on noise. The hysteresis threshold may be determined by one or more of a: fixed threshold, fraction of recent N peak to peak values (where “N” is an integer), fraction of signal RMS value, dynamic threshold with linear or exponential decay, and time duration since the previous zero crossing.

[0118] The graphs of FIGS. 5A and 5B also include positive half cycles, corresponding to portions of the sensor signal above the signal midpoint 150, and negative half cycles, corresponding to portions of the sensor signal below the signal midpoint 150. A positive half cycle may be either an inspiration half cycle or an expiration half cycle depending upon the polarity of the sensor signal. Likewise, a negative half cycle may be either an inspiration half cycle or an expiration half cycle depending upon the polarity of the sensor signal. For example, for filtered acceleration signal 142A of FIG. 5A having a positive polarity, half cycle 154B is a positive half cycle and an inspiration half cycle, while half cycle 154C is a negative half cycle and an expiration half cycle. In contrast, for filtered acceleration signal 142B of FIG. 5B having a negative polarity, half cycle 154B is a negative half cycle and an inspiration half cycle, while half cycle 154C is a positive half cycle and an expiration half cycle. The expiration half cycle includes the expiratory pause 144C.

[0119] In some examples, the zero crossings may be used to identify the polarity of the sensor signal by identifying the slope of the expiratory phase and/or the inspiratory phase. The slope of the expiratory phase and/or the inspiratory phase may be identified by determining the median of data points of the sensor signal between a current midpoint (e.g., Ti.Mio at current time 151 in FIGS. 5A and 5B) and at least two previous midpoints, such as a multiple N of two previous midpoints corresponding to N complete respiratory cycles. In some examples, more than two previous midpoints may be used, such as 4, 6, 8, etc. In some examples, many (e.g., 11 or more) previous midpoints may be used to determine the median, and deviation from a multiple N of two previous midpoints corresponding to N complete respiratory cycles is acceptable (i.e. , an odd number of half cycles may be used). By determining the median, the inspiration half cycle and the expiration half cycle may be identified since the location of the median above or below zero corresponds to the location of the expiration half cycle above or below the zero crossing, respectively. Accordingly, the slope of the expiratory phase is identified as negative (e.g., as indicated at 155A in FIG. 5A) and the slope of the inspiratory phase is identified as positive in response to the median being less than 0. The slope of the expiratory phase is identified as positive (e.g., as indicated at 155B in FIG. 5B) and the slope of the inspiratory phase is identified as negative in response to the median being greater than 0. In some examples, if the absolute value of the median is less than 1 e-5, the median may be rounded to 0. In this case, the median is not used to identify the polarity of the sensor signal.

[0120] As described further below, an inspiration to expiration (l/E) ratio may be calculated by dividing the duration of an inspiration half cycle 154B by the duration of an expiration half cycle 154C. The sign of the median may indicate the sign of exhalation slope since respiration typically has an l/E ratio that is less than 1.0. [0121] In addition, a respiratory period may be calculated based on the midpoints of expiratory phases and/or inspiratory phases by calculating the difference between each pair of subsequent TE.MID events and/or each pair of subsequent TI.MID events. In other words, in one example, the respiratory period equals the midpoint (TI.MID) of an inspiratory phase minus the midpoint (TI.MID) of an immediately previous inspiratory phase or equals a running average of the midpoint (TI.MID) of an inspiratory phase minus the midpoint (TI.MID) of an immediately previous inspiratory phase for N previous respiratory periods, where “N” is an integer number of respiratory periods. In some examples, the respiratory period equals the midpoint (TE.MID) of an expiratory phase minus the midpoint (TE.MID) of an immediately previous expiratory phase or equals a running average of the midpoint (TE.MID) of an expiratory phase minus the midpoint (TE.MID) of an immediately previous expiratory phase for N previous respiratory periods. A respiratory rate may be calculated as 1 divided by the calculated respiratory period or 1 divided by a running average of N previous respiratory periods. A duration of an expiratory to inspiratory half cycle (e.g., 154C) may be calculated as the midpoint (TI.MID) of an inspiratory phase minus the midpoint (TE.MID) of the immediately previous expiratory phase or calculated as a running average of the midpoint (TI.MID) of an inspiratory phase minus the midpoint (TE.MID) of the immediately previous expiratory phase over N previous respiratory periods. A duration of an inspiratory to expiratory half cycle (e.g., 154B) may be calculated as the midpoint of an expiratory phase (TE.MID) minus the midpoint (TI.MID) of the immediately previous inspiratory phase or calculated as a running average of the midpoint of an expiratory phase (TE.MID) minus the midpoint (TI.MID) of the immediately previous inspiratory phase over N previous respiratory periods. An inspiration to expiration (l/E) ratio may be calculated by dividing the duration of the inspiratory to expiratory half cycle (e.g., 154B) by the duration of the expiratory to inspiratory half cycle (e.g., 154C).

[0122] Based on the midpoint of one or more previous expiratory phases and/or one or more previous inspiratory phases, the midpoint of a future inspiratory phase (PTI.MID) 156 may be predicted. Based on the predicted midpoint of the future inspiratory phase, a stimulation duration indicated at 157 may be determined. The stimulation may begin at the beginning of the inspiratory phase or just prior to the inspiratory phase. The stimulation may end at the end of the inspiratory phase or just after the inspiratory phase. In some examples, the stimulation duration 157 is at least 500 milliseconds with no more than 10 seconds between stimulation durations 157.

[0123] In one example, the stimulation is started an amount of time prior to the predicted midpoint PTI.MID by a first predetermined (e.g., fixed) interval. In some examples, the stimulation is started an amount of time prior to the predicted midpoint PTI.MID computed as a percentage of the respiratory rate. In some examples, the stimulation is started an amount of time prior to the predicted midpoint PTI.MID computed as a percentage of the respiratory rate plus a predetermined (e.g., fixed) amount of time. In some examples, the stimulation is started an amount of time prior to the predicted midpoint PTI.MID computed as a percentage of the expiratory to inspiratory half cycle. In some examples, the stimulation is started an amount of time prior to the predicted midpoint PTI.MID computed as a percentage of the inspiratory to expiratory half cycle. In some examples, the stimulation is started an amount of time prior to the predicted midpoint PTI.MID computed as a percentage of the respiration period.

[0124] In one example, the stimulation ends after the predicted midpoint PTI.MID by a second predetermined (e.g., fixed) interval as previously described. In some examples, the stimulation ends after the predicted midpoint PTI.MID by an amount of time computed as a percentage of the respiratory rate. In some examples, the stimulation ends after the predicted midpoint PTI.MID by an amount of time computed as a percentage of the respiratory rate plus a predetermined amount of time. In some examples, the stimulation ends after the predicted midpoint PTI.MID by an amount of time computed as a percentage of the expiratory to inspiratory half cycle. In some examples, the stimulation ends after the predicted midpoint PTI.MID by an amount of time computed as a percentage of the inspiratory to expiratory half cycle. In some examples, the stimulation ends after the predicted midpoint PTI.MID by an amount of time computed as a percentage of the respiration period. In some examples, the stimulation may start and/or end based on a combination of any of the above examples.

[0125] FIG. 5C is a diagram schematically representing an example of stimulation timing 1100A including hard refractory and soft refractory. In some examples, hard refractory is a period during which stimulation cannot occur, and soft refractory is a period during which stimulation may or may not occur (e.g., based on a probability, morphology, or decision criteria threshold that reduces the likelihood of stimulation occurring). The hard refractory period and the soft refractory period may be set to a predefined percentage of the respiratory period. In the example illustrated in FIG. 5C, stimulation is turned on at 1102, remains on for a stimulation period 1104, and is turned off at 1106. Immediately following the stimulation being turned off at 1106, a hard refractory period 1108 begins. While in this example, the hard refractory period is set to 25% of the respiratory period, in some examples, the hard refractory period may be set to another suitable percentage (e.g., within a range between 20% and 75%). Immediately following the hard refractory period 1108, a soft refractory period 1110 begins. In the example illustrated in FIG. 5C, stimulation also doesn’t occur during the soft refractory period. While in this example, the soft refractory period is set to 13% of the respiratory period, in some examples, the soft refractory period may be set to another suitable percentage (e.g., within a range between 10% and 30%). Stimulation may be started any time after the soft refractory period 1110 ends. In this example, stimulation is turned on again at 1112a after the soft refractory period 1110, remains on for a stimulation period 1114, and is turned off at 1116a, after which another hard refractory period begins. An average stimulation off to stimulation off period is indicated at 1120. A stimulation period (e.g., 1114) may be scheduled to start (e.g., at 1112a) after the hard refractory period 1108 plus the soft refractory period 1110 has occurred.

[0126] FIG. 5D is a diagram schematically representing another example of stimulation timing 1100B including hard refractory and soft refractory. In the example illustrated in FIG. 5D, stimulation is turned on at 1102, remains on for a stimulation period 1104, and is turned off at 1106. Immediately following the stimulation being turned off, a hard refractory period 1108 begins. Immediately following the hard refractory period 1108, a soft refractory period 1110 begins. In the example illustrated in FIG. 5D, stimulation may be started during the soft refractory period, such that stimulation may be started any time after the hard refractory period 1108 ends. In this example, stimulation is turned on again at 1112b during the soft refractory period 1110, remains on for a stimulation period 1114, and is turned off at 1116b, after which another hard refractory period begins. An average stimulation off to stimulation off period is indicated at 1120. A stimulation period (e.g., 1114) may be scheduled to start (e.g., at 1112b) after the hard refractory period 1108 has occurred. In some examples, hard refractory and/or soft refractory are considered when scheduling the start and stop of stimulation further described later with reference to at least FIGS. 15I-16G.

[0127] FIG. 6 is a flow diagram illustrating one example of a method 158 for treating sleep disordered breathing. At 158A, method 158 includes sensing changes in acceleration indicative of respiration of the patient to provide a sensor signal. For example, sensor 12A of FIG. 1A, 12B of FIG. 1 B, or 104A of FIGS. 2 or 3 may sense changes in acceleration of a chest wall 102A of the patient indicative of respiration to provide a sensor signal. At 158B, method 158 includes filtering the sensor signal to detect zero crossings of the sensor signal. For example, the sensor signal may be bandpass filtered to detect zero crossings TE.MID 152 and TI.MID 153 as indicated by filtered acceleration sensor signals 142A and 142B of FIGS. 5A and 5B. At 158C, method 158 includes applying electrical stimulation to the patient based on the detected zero crossings. For example, electrical stimulation as indicated at 157 in FIGS. 5A and 5B may be applied to the patient based on the detected zero crossings TE.MID 152 and TI.MID 153. Additional methods for treating sleep disordered breathing are described below with reference to at least FIGS. 14A-16L.

[0128] FIGS. 7A, 7B, and 8 are diagrams which schematically represent an example method and/or example sensor 204 which may comprise three sensing elements 122A (Y), 162 (Z), 164 (X) arranged orthogonally relative to each other. In some examples, the sensor 204 (including at least sensing element 122A) comprises at least some of substantially the same features and attributes as sensor 104A previously described in association with at least FIGS. 2-4 in which just one sensing element 122A (Y) is present. However, as shown in FIGS. 7A- 7B, in addition to sensing element 122A (Y), in some examples sensor 204 also may comprise acceleration sensing element 162 having orientation Z (Z-axis) which is perpendicular to sensing element 122A. As implanted, this Z-axis orientation is generally perpendicular to a superior-inferior (S — I) orientation of the chest wall 102A, and is generally parallel to an anterior-posterior (A — P) orientation of the chest wall 102A.

[0129] Meanwhile, as shown in FIG. 7B, in addition to comprising sensing element 122A (Y), in some examples sensor 204 also may comprise acceleration sensing element 164, having orientation X (X-axis) which is generally perpendicular to sensing element 122A. As implanted, this X-axis orientation is generally perpendicular to a superior-inferior (S — I) orientation of the chest wall 102A, and generally perpendicular to an anterior-posterior (A — P) orientation of the chest wall 102A. In some such examples, sensing element 164 may sense rotational movement of chest wall 102A (as represented by directional arrow B5) in a plane defined by the anterior-posterior orientation (A — P) and by the lateral-medial orientation (L — M), according to changes in an inclination angle (as represented via directional arrow B4) of sensing element 164. Each of the respective sensing elements 162 (Z), 164 (X) may provide additional sensing of rotational movement of the chest wall 102A to provide further respiration information. With further reference to FIGS. 7A, 7B, and 8, in some examples, sensor 204 may comprise all three sensing elements 122A (Y), 162 (Z), and 164 (X).

[0130] As schematically represented in the diagram 250 of FIG. 8, the sensed acceleration signal information from each of the three sensing elements 162, 122A, 164 of sensor 204 may be combined to provide composite rotational change information (252). In some examples, the composite rotational change information 252 may sometimes be referred to as a virtual vector representing the overall rotational movement (e.g., according to at least two orthogonal axes) caused by breathing. In some such examples, the composite rotational change information 252 corresponds to sensing the AC component of the multidimensional acceleration vector (e.g., a virtual vector) with respect to gravity. [0131] In some examples, at least two of the three orthogonally-arranged sensing elements may be used to perform determination of composite rotational movement and therefore respiration information at least based on an AC component of a multi-dimensional acceleration vector produced by the n singleaxis sensing elements.

[0132] In some such examples associated with FIG. 8, the virtual vector corresponding to the composite rotational change (252) may exhibit higher sensitivity to respiration than any single vector of a physical sensing element 122A (Y), sensing element 162 (Z), or sensing element 164 (X). In some such examples, the virtual vector (252) may exhibit a higher signal-to-noise ratio (e.g., signal quality) than any single physical vector, such as single sensing element 122A (Y) or single sensing element 162 (Z) or single sensing element 164 (X) by virtue of combining the signals of the multiple sensing elements.

[0133] In some such examples, the virtual vector (e.g., 252) effectively excludes non-physiologic motion of the chest wall. At least some examples of such nonphysiologic motion may comprise motion of a vehicle (e.g., car, airplane, etc.) within which the patient is riding, of patient swinging in a hammock, and the like. Accordingly, determining respiration information via the virtual vector in such example methods and/or devices may produce respiration information which is generally insensitive to non-physiologic motion of the patient.

[0134] In some examples, respiration detection may be based on a sum of two of the vectors from among the three orthogonally-arranged sensing elements. In some examples, respiration detection may be based on a sum of signals from all three orthogonally-arranged sensing elements. In some examples, respiration detection may be determined by looking independently at each of the three vectors or from among the three vectors.

[0135] In some examples, a method and/or device may employ control portion 900 (FIG. 19A) to select the virtual vector (e.g., 252) or a physical vector from one of the sensing elements 122A, 162, or 164 for use in determining respiration information. In some such examples, the method and/or device may evaluate the robustness of the determined respiration information and automatically convert operation among the virtual vector (e.g., 252 in FIG. 8) and any one of the physical vectors (e.g., 122A/Y, 162/Z, 164/X) to consistently use the most robust, accurate signal source in determining respiration information.

[0136] In association with the examples of at least FIGS. 7A, 7B, and 8, in some examples the signal-to-noise ratio of a virtual vector and/or physical vector may be enhanced via excluding noise, such as later described in association with at least noise model parameter 870 (FIG. 18E).

[0137] In some examples, the above-described measuring of rotational movement (of a portion of a chest wall via acceleration sensing) per sensing element 162 (Z-axis) may be likened to a pitch parameter, measuring rotational movement per sensing element 122A (Y-axis) may be likened to a yaw parameter, and measuring rotational movement per sensing element 164 (X-axis) may be likened to a roll parameter. Because of variances in anatomy from patient to patient, the particular implant orientation, and/or the particular implant location (e.g., front vs. side of the chest), the pitch parameter, yaw parameter, and/or roll parameter may bear a rough or general correspondence to the ideal definition for such respective parameters in which the pitch parameter may correspond to rotational movement of the portion of the chest wall in a first plane defined by an anterior-posterior orientation and by a superior-inferior orientation of the patient’s body. Similarly, the yaw parameter may roughly or generally correspond to rotational movement of the portion of the chest wall in a second plane defined by the anterior-posterior orientation and by a lateral-medial orientation of the patients’ body. Similarly, the roll parameter may roughly or generally correspond to rotational movement of the portion of the chest wall in a third plane defined by the lateral-medial orientation and by the superior-inferior orientation of the patient’s body.

[0138] In examples in which the patient’s body position corresponds to a primary sleeping position (e.g., generally horizontal), then the magnitude of changes in the AC signal component from rotational movement (B3) sensing element 162 (Z axis) during breathing will be negligible and the magnitude of changes in the AC signal component from rotation (arrow B4) of sensing element 164 (X axis) during breathing may be relatively small at least compared the magnitude of changes in the AC signal component of sensing element 122A (Y-axis) during breathing (as described in association with FIGS. 2-4).

[0139] However, in some example situations the patient’s body position may correspond to a secondary or alternate sleep position, such as sitting upright against a support 273 (e.g., ordinary chair, airplane chair, etc.) as shown in FIG. 10 or in a partially reclined position (e.g., torso is 45 degrees from horizontal) against a support 263 (e.g., recliner chair, recliner bed, etc.) which is at angle (A) relative to generally horizontal (e.g., floor) as shown in FIG. 9A. In some such examples, such as the partially reclined position in FIG. 9A, at least the respective sensing element 162 (Z axis) may yield significant magnitude of changes in the AC signal component during breathing instead of and/or in addition to sensed changes in the AC signal component of sensing element 122A (Y-axis) during breathing. In this example, the sensing element 122A may comprise a first angular orientation (like YR1 in FIG. 3 for peak expiration) which is 45 degrees (o in FIG. 9B) relative to the gravity vector G (and which is 45 degrees relative to a generally horizontal plane, which typically is a primary sleep position). While the first orientation (e.g., YR1 ) of the sensing element 122A may not be generally perpendicular to the gravity vector G as in FIG. 3, at the first orientation of 45 degrees (o in FIG. 9B) relative to the gravity vector G, the acceleration sensing element 122A still exhibits sufficient sensitivity in the AC signal component to produce meaningful measurements in changes of the inclination angle (e.g., Q in FIG. 3) of sensing element 122A between the first and second orientations (e.g., YR1 and YR2) during breathing to enable determining respiration information.

[0140] In this example, the sensing element 162 may comprise a first orientation (like YR1 in FIG. 3), which extends at an angle of 135 degrees (6 in FIG. 9C) relative to the gravity vector G (and which is 45 degrees relative to a generally horizontal plane, which typically is a primary sleep position). While the sensing element 162 may not be generally perpendicular to the gravity vector G (as was sensing element 122A in the example of FIG. 3), at the first orientation of 135 degrees (6 in FIG. 9C) relative to the gravity vector G, the acceleration sensing element 162 exhibits sufficient sensitivity in the AC signal component to produce meaningful measurements in changes of the inclination angle (like Q in FIG. 3) of sensing element 162 between its first orientation (peak expiration) and second orientation (peak inspiration) during breathing to enable determining respiration information.

[0141] In a manner similar to that shown in FIG. 8, the sensed rotational movement from at least the multiple sensing elements (e.g., 162/Z-axis and 122A/Y-axis in FIGS. 9A-9C) may be combined to yield a composite value of sensed rotational movement of sensor 204 in order to produce sensing of a respiratory waveform while the patient is in the partially reclined position.

[0142] It will be further understood that, in some examples, the sensing element 164 (X-axis) also may be used in addition to sensing elements 122A, 162 (and in a manner similar to that described for sensing elements 122A, 162 in FIGS. 9A- 9C) to provide further sensing by which the determination of respiration information can be made, with the rotational sensing information being combined, similar to that shown in FIG. 8. With this in mind and with further reference to at least the examples of FIGS. 9A-9C and 10, it will be understood that employing a three axis accelerometer (in which the three axes are orthogonally-arranged) will ensure that at least one of the three axes will have an output signal of magnitude sufficient to reliably determine respiration (e.g., based on rotational movement of the sensor in correspondence with rotational movement of a portion of the chest wall during breathing as described in various examples).

[0143] It will be understood that in some examples, the particular angle A of reclination in FIG. 9A may be angles other than 45 degrees, and may be variable over time in some instances, depending on the type and manner of support 263 (e.g., adjustable bed, chair). In some such examples, a determination of respiration information may be based on the particular respective sensing element(s) (e.g., 122A (Y-axis), 162 (Z-axis), 164 (X-axis)) having the orientation(s) closest to being generally perpendicular to the gravity vector G for the particular angle A at a particular point in time.

[0144] Moreover, in this example arrangement of FIGS. 9A-10, if and/or when the patient moves to another sleeping position, such as generally horizontal position (e.g., FIG. 3), then the sensing element 122A (or sensing element 164) may become the sole or primary signal source for detecting respiration in some examples. Accordingly, example arrangements of multiple single-axis acceleration sensing elements in orthogonal relationship to each other may provide robust sensing of respiration which enables adaptability in response to a patient moving among different sleep positions within a single treatment period or among multiple, different treatment periods.

[0145] As previously noted, the FIG. 10 schematic represents at least a chest wall 102A of a patient’s body in a generally vertically upright position, such as if the patient were sitting on a support 276 with their torso against a vertical support 273. In this example arrangement, both the acceleration sensing elements 162 (Z-axis) and 164 (X-axis) of sensor 204 may have a first orientation which is generally perpendicular (or reasonably close to being generally perpendicular) to gravity vector G, whereas the acceleration sensing element 122A (Y-axis) of sensor 204 has a general orientation which is generally parallel to gravity vector G. Accordingly, for substantially similar reasons presented with respect to FIGS. 3 and 7A-9C, one or both of the sensing elements 162 (Z-axis), 164 (X-axis) may provide the most sensitive sensing elements by which respiration information determination may be performed.

[0146] In particular, upon rotational movement of the patient’s chest wall 104A during breathing within a treatment period, rotational movement of Z-axis sensing element 162 between a first orientation (e.g., like YR1 in FIG. 3) and a second orientation (e.g., like YR2 in FIG. 3) may be sensed as a range of values of an AC signal component from which a respiratory waveform (including respiratory phase timing/details) may be determined as shown in FIGS. 4-5B. Moreover, rotational movement of X-axis sensing element 164 may provide similar information and may be used to determine respiration information. The respiration information may be determined solely from the Z-axis sensing element 162, solely from the X-axis sensing element 164, or from a combination of information sensed via both of the Z-axis sensing element 162 and the X-axis sensing element 164. While the Y-axis sensing element 122A would generally be expected to produce negligible or minimal respiration information (because of being parallel to the gravity vector G), in some examples, information sensed from Y-axis sensing element 122A may be combined with rotational information sensed via the sensing elements 162, 164.

[0147] FIG. 11 is a diagram 300 including a front view schematically representing different measurement axes of an example sensor 304 and/or related example method. In some examples, the sensor 304 may comprise at least some of substantially the same features and attributes as the sensors, sensing elements, and related example methods as previously described in association with FIGS. 1 A-10. As shown in FIG. 11 , sensor 304 is implanted within a wall of chest region 306 of torso 307 below a neck 124 and head 302. The sensor 304 comprises multiple sensing elements 122A (Y-axis orientation), 162 (Z-axis orientation), 164 (X-axis orientation), which may be independent such as three separate singleaxis accelerometers, or these sensing elements may be combined into a single arrangement, such as a three-axis accelerometer. FIG. 12 is a diagram 350 including a side view schematically representing the sensor 304 of FIG. 11 , highlighting the orientation of the sensing elements 122A, 162.

[0148] FIG. 13 is a diagram 400 including an isometric view schematically representing an implantable device 402 comprising an accelerometer-based sensor 304, which may comprise at least some of substantially the same features and attributes as the sensors, sensing elements, and related example methods as previously described in association with FIGS. 1A-12. It will be understood that the sensor (and sensing elements) described in FIGS. 1A-12 may be implemented as being on or within device 402. In some examples, sensor 304 is enclosed within a sealed housing (e.g., can) of the device 402. However, the sensor 304 may be external to the housing 405 of device 402, whether located on the housing or extending from the housing 405 on a lead.

[0149] With further reference to FIG. 13, in some examples, device 402 may comprise an implantable device, which includes circuitry and power elements to operate the sensor 304 to sense physiologic phenomenon, such as but not limited to respiration information. In some examples, the circuitry and power may be implemented within or as part of a control portion 900 and/or related portions, elements, functions, parameters, engines, as further described later in association with at least FIGS. 18A-18E. Among other attributes, via the control portion, the device 402 may be used to monitor and/or diagnose physiologic phenomenon, patient conditions (e.g., respiratory health, cardiac health, etc.), with one such patient condition including sleep disordered breathing (SDB). In some examples, device 402 may comprise an implantable pulse generator (IPG), which may implement neurostimulation in association with respiration detection in order to treat sleep disordered breathing and/or other patient health conditions. In some such examples, the device 402 may also sense translational movements of the chest wall and/or associated body tissue in order to sense, monitor, diagnose, etc. the various physiologic phenomenon, patient conditions, etc. whether the sensed translational movement is obtained instead of, or in addition to, the sensed rotational movement of the portion of the chest wall.

[0150] With the examples of FIGS. 1A-13 in mind, it will be understood that in some examples the sensor signal which will be used to determine respiration information may be selected from among multiple sensing elements, such as but not limited to, the individual/respective axes of the three-axis accelerometer. In some examples, the sensor signal used to determine respiration information may be predetermined or selected by a clinician. Zero crossings of the sensor signal may be detected and used to determine when to apply electrical stimulation. Accordingly, at least some example methods and/or devices as described in association with at least FIGS. 14A-16L further describe such methods. In some examples, the methods may be implemented by a device (e.g., 10A, 10B, 402, etc.) including a sensor (e.g., 12A, 12B, 104A, 122A, 162, 164, 204, 304, etc.) securable to a patient to provide a sensor signal. The sensor signal senses changes in acceleration indicative of respiration of the patient (e.g., as described with reference to at least FIGS. 2-5B and/or 7A-13). The device may also include a filter to detect zero crossings (e.g., TE.MID 152 and TI.MED 153 of FIGS. 5A and 5B) of the sensor signal and an electrode (e.g., electrode 672 to be described with reference to at least FIG. 17B) to deliver electrical stimulation to the patient. The device may also include a control portion (e.g., 14A, 14B, 900) to control the operation of the device by implementing the methods described in association with at least the following FIGS. 14A-16L. [0151] Accordingly, as shown at 500 in FIG. 14A and as previously described, some example methods and/or devices for applying electrical stimulation may comprise applying electrical stimulation based on detected zero crossings (e.g., T E,MID 152 and T I.MED 153 of FIGS. 5A and 5B). The steeper slope of the sensor signal at each zero crossing is less sensitive to noise compared to the flat tops/bottoms of the peaks/troughs, upon which the application of electrical stimulation may be based in some examples. Therefore, applying electrical stimulation based on the detected zero crossings may more precisely align the electrical stimulation with at least a portion of the inspiratory phase for at least some respiratory cycles. As shown at 501 in FIG. 14B, in some examples, the methods and/or devices further comprise ignoring a currently detected zero crossing in response to the currently detected zero crossing being within a predetermined period of a previously detected zero crossing. For example, after the previously detected zero crossing, if a new zero crossing is detected within a range between 0.1 seconds and 2 seconds depending on the respiration rate, the currently detected zero crossing may be considered noise and skipped.

[0152] In some examples, the filter for detecting the zero crossings of the sensor signal may comprise a bandpass filter. In some examples, the bandpass filter may be part of filtering 714 of FIG. 18A described later. The bandpass filter may remove high frequency noise and DC offset from the sensor signal. The bandpass filter may include corner frequencies, for example, a lower corner frequency within a range between 0.01 Hz and 1 Hz, and an upper corner frequency within a range between 0.25 Hz and 10 Hz. In some examples, the lower corner frequency is within a range between 0.01 Hz and 0.5 Hz, and the upper corner frequency is within a range between 1 Hz and 1 .5 Hz. The corner frequencies may be adjusted based on frequency domain analysis of the sensor signal to remove noise from the sensor signal. In some examples, the frequency domain analysis comprises power spectral density (PSD) analysis. In some examples, the device may also comprise a decimation filter to reduce a sampling rate of the sensor signal. In some examples, the decimation filter may be a low pass decimation filter to reduce a sampling rate of the sensor signal and increase an amplitude resolution of the sensor signal. The decimation filter may reduce the sampling rate of the sensor signal to within a range, for example, between 40 Hz and 60 Hz. In some examples, the decimation filter may also be part of filtering 714 of FIG. 18A described later. Since respiration is at a slow frequency (e.g., less than 1 Hz), filters inject a large amount of group delay into the outputted data making real-time stimulation decisions computational intensive as the decisions may be made over 1 second in the past. Accordingly, by reducing the sampling rate of the sensor signal by using a decimation filter, computational saving may be achieved. In some examples, a decimation filter may be used prior to the bandpass filter. Thus, in some examples described herein, the “sensor signal” may refer to the outputted data from a decimation filter.

[0153] As shown at 502 in FIG. 14C and as previously described with reference to FIGS. 5A and 5B, in some examples, the methods and/or devices comprise determining a respiration period (e.g., 154A) based on the detected zero crossings (e.g., TE,MID 1 52 and TI, MED 153 of FIGS. 5A and 5B). In some examples, the corner frequencies of the bandpass filter used to detect the zero crossings of the sensor signal may be adjusted based on the respiration period. For example, if the determined respiration period is reduced (e.g., due to faster breathing of the patient), the corner frequencies of the bandpass filter may be widened to prevent loss of respiration information. As shown at 503 in FIG. 14D and as previously described with reference to FIGS. 5A and 5B, in some examples, the methods and/or devices comprise determining an average respiration rate (e.g., 1 divided by the average respiration period described below with reference to FIG. 14N) based on the detected zero crossings.

[0154] As shown at 504 in FIG.14E, in some examples, the methods and/or devices comprise identifying each positive half cycle and each negative half cycle of the sensor signal based on the detected zero crossings. For example, referring to FIG. 5A, the positive half cycle is identified at 154B as the half cycle above the zero crossing 150, and the negative half cycle is identified at 154C as the half cycle below the zero crossing 150. Referring to FIG. 5B, the positive half cycle is identified at 154C as the half cycle above the zero crossing 150, and the negative half cycle is identified at 154B as the half cycle below the zero crossing 150. As previously described with reference to FIGS. 5A and 5B and further described below with reference to at least FIGS. 14V, 14X, 14Y, 15A, and 15H, each identified positive half cycle and negative half cycle may be further used to determine a polarity of the sensor signal, identify each inspiration half cycle and expiration half cycle of the sensor signal, and/or determine a confidence factor of the sensor signal.

[0155] As shown at 505 in FIG. 14F, in some examples, the methods and/or devices comprise identifying each expiratory phase and/or each inspiratory phase of the sensor signal. For example, referring to FIGS. 5A and 5B, the expiratory phase is identified at 144B and the inspiratory phase is identified at 144A. By identifying each expiratory phase and/or each inspiratory phase of the sensor signal, the polarity of the sensor signal may be determined and electrical stimulation may be applied to correspond with at least a portion of the inspiratory phase for at least some respiratory cycles.

[0156] As shown at 506 in FIG. 14G and as previously described with reference to FIG. 1 B, in some examples, the methods and/or devices comprise determining a confidence factor associated with the sensor signal. As further described below, the confidence factor may be used to determine whether electrical stimulation is applied in a closed loop mode (e.g., in response to the confidence factor indicating that the sensor signal quality is sufficient) or in an open loop mode (e.g., in response to the confidence factor indicating that the sensor signal quality is insufficient). In some such examples associated with 505 of FIG. 14F and 506 of FIG. 14G, as shown at 507 in FIG. 14H, determining a confidence factor may comprise determining a confidence factor associated with each identified expiratory phase and/or each identified inspiratory phase.

[0157] In some examples associated with 500 of FIG. 14A, 506 of FIG. 14G, and 507 of FIG. 14H, as shown at 508 in FIG. 141, applying the electrical stimulation may comprise applying the electrical stimulation in an open loop mode in response to the confidence factor being less than a threshold value and in a closed loop mode in response to the confidence factor being greater than the threshold value. In some examples, the threshold value may be configurable by a clinician and/or physician. As previously described, in the closed loop mode electrical stimulation is applied based on feedback from the sensor signal, while in the open loop mode electrical stimulation is applied without using feedback from the sensor signal.

[0158] As shown at 509 in FIG. 14J, in some examples, the methods and/or devices comprise determining whether a current zero crossing is a midpoint of a slope of an inspiratory phase (TI.MID) or a midpoint of a slope of an expiratory phase (TE.MID). For example, referring to FIGS. 5A and 5B, the current zero crossing at the current time 151 is determined to be a midpoint of a slope of an inspiratory phase (TI.MID). AS further described below, the current zero crossing TI.MID or TE.MID may be used to determine a current, average, and/or a median respiration period as described with reference to at least FIGS. 14K-14P and/or for scheduling a start and stop of electrical stimulation as described with reference to at least FIGS. 15I-15R and 16A-16G.

[0159] As shown at 510 in FIG. 14K, in some examples, the methods and/or devices comprise determining a current respiration period based on the detected zero crossings. In some examples associated with 510 of FIG. 14K, as shown at 511 in FIG. 14L, determining the current respiration period (e.g., 154A of FIGS. 5A and 5B) based on the detected zero crossings may comprise determining a current respiration period based on a current TI.MID and a previous TI.MID (e.g., current TI.MID minus the previous TI.MID) and/or based on a current TE.MID and a previous TE.MID (e.g., current TE.MID minus the previous TE.MID). In some examples associated with 510 of FIG. 14K, as shown at 512 in FIG. 14M, determining the current respiration period (e.g., 154A of FIGS. 5A and 5B) based on the detected zero crossings may comprise determining a current respiration period based on a first period (e.g., 154C of FIGS. 5A and 5B) between a current TI.MID and a current TE.MID plus a second period (e.g., 154B of FIGS. 5A and 5B) between the current TE.MID and a previous TI.MID and/or based on a first period between a current TE.MID and a current TI.MID plus a second period between the current TI.MID and a previous TE.MID.

[0160] In some examples associated with 502 of FIG. 14C and/or 510 of FIG. 14K, as shown at 513 in FIG. 14N, the methods and/or devices may comprise determining an average respiration period for a predetermined number (e.g., 5, 10, 20, 30, 50, 100, 250, etc.) of respiration periods. In some examples associated with 513 of FIG. 14N, as shown at 514 in FIG. 140, the methods and/or devices may comprise excluding, from the determination of the average respiration period, respiration periods outside a predetermined range of threshold values. In some examples, the predetermined range of threshold values may be configurable by a clinician and/or physician. For example, a respiration period less than 0.5 times the average respiration period or greater than 1.5 times the average respiration period may be excluded from the determination of the average respiration period as being noise, which may be due to a missing or extraneous zero crossing. In some examples associated with 502 of FIG. 14C, 510 of FIG. 14K, and/or 513 of FIG. 14N, as shown at 515 in FIG. 14P, the methods and/or devices may comprise determining a median respiration period over a predetermined number (e.g., 5, 10, 20, 30, 50, 100, 250, etc.) of respiration periods. A median respiration period lessens the effect of outlier respiration periods, while an average respiration period allows larger values to have an impact on the resulting respiration period.

[0161] In some examples associated with 502 of FIG. 14C and/or 510 of FIG. 14K, as shown at 410 in FIG. 14Q is a diagram schematically representing an example method and/or device for resetting a confidence factor, which may be used to determine whether to apply electrical stimulation in a closed loop mode or in an open loop mode. As shown at 516 in FIG. 14Q, the methods and/or devices comprise determining a standard deviation over a predetermined number of respiration periods. As shown at 517 in FIG. 14Q, the methods and/or devices further comprise in response to the standard deviation exceeding a threshold value, resetting a confidence factor to a predetermined value. For example, if the standard deviation exceeds a threshold value within a range between 10 milliseconds and 5000 milliseconds (e.g., within a range between 100 milliseconds and 1000 milliseconds), the confidence factor may be set to zero. In some examples, the threshold value may be configurable by a clinician and/or physician. Setting the confidence factor to zero may result in stimulation switching (e.g., converting) from a closed loop mode to an open loop mode in response to the confidence factor falling below a threshold value greater than zero. Thus, stimulation may switch (e.g., convert) from a closed loop mode to an open loop mode in response to a confidence factor falling below a threshold value, and switch (e.g., convert) from the open loop mode back to the closed loop mode in response to the confidence factor rising back above the threshold value. Accordingly, throughout a treatment period, stimulation may continue uninterrupted and switch between the closed loop mode and the open loop mode depending upon the confidence factor (which may vary based on changing conditions, such as patient posture, movement, etc.) and the selected threshold value.

[0162] FIG. 14R is a diagram schematically representing an example method and/or device 412 for resetting values, which may be used to determine whether to apply stimulation in a closed loop mode or in an open loop mode, in response to detecting a change in posture of the patient. In some examples, the sensor to sense changes in acceleration indicative of respiration of the patient may also be used to sense posture of the patient. In some examples, one sensor may be used to sense changes in acceleration indicative of respiration of the patient and another sensor may be used to sense posture of the patient. As shown at 518 in FIG. 14R, in some examples, the methods and/or devices comprise detecting a change in the posture of the patient. As shown at 519 in FIG. 14R, the methods and/or devices further comprise in response to detecting the change in the posture of the patient, resetting the filter, and resetting the confidence factor to a predetermined value less than a threshold value. For example, in response to detecting the patient rolling over, the bandpass filter is reset such that a new zero crossing may be detected and the confidence factor of the sensor signal may be set to zero, which is less than a threshold confidence factor above which stimulation may operate in a closed loop mode. In some examples, the threshold confidence factor above which stimulation may operate in the closed loop mode may be configurable by a clinician and/or physician. Thus, setting the confidence factor to zero may result in stimulation switching from a closed loop mode to an open loop mode. Once the bandpass filter stabilizes after the patient posture change, the confidence factor may rise back above the threshold value and stimulation may switch from the open loop mode back to the closed loop mode. [0163] In some examples associated with FIG. 14N and/or FIG. 14R described above and/or FIG. 14U to be described below, as shown at 414 in FIG. 14S is a diagram schematically representing another example method and/or device for resetting values in response to detecting a change in posture of the patient. As shown at 520 in FIG. 14S, the methods and/or devices comprise detecting a change in the posture of the patient. As shown at 521 in FIG. 14S, the methods and/or devices further comprise in response to detecting the change in the posture of the patient, resetting an average respiration period and an identified polarity of the sensor signal. In this example, since the posture of the patient has changed, the respiration period and the polarity of the sensor signal may also have changed. Therefore, to ensure the average respiration period and the identified polarity of the sensor signal are correct after the change in posture of the patient, the average respiration period and the polarity of the sensor signal are newly calculated/identified each time the posture of the patient changes.

[0164] In some examples associated with FIG. 14R and/or FIG. 14S, as shown at 522 in FIG. 14T, detecting the change in posture of the patient comprises detecting the change in the posture of the patient in response to a maximum value of sensor signal samples (e.g., data points) (stored in a buffer of a predetermined length) minus a minimum value of the sensor signal samples (stored in the buffer) exceeding a threshold value. Thus, a change in the posture of the patient is detected when a difference between the maximum value and the minimum value of sensor signal samples (stored in the buffer) exceeds the threshold value. In some examples, the threshold value may be configurable by a clinician and/or physician. When the posture of a patient changes, the selected axis (e.g., of a three axis accelerometer 204 in FIGS. 7A and 7B) may exhibit a large difference between the maximum and minimum sensor signal samples due to the changing orientation of the selected axis with respect to gravity. In some examples, when a change in posture of the patient is detected, the sensor signal samples stored in the buffer may be erased and a different axis may be selected for the new posture.

[0165] As shown at 523 in FIG. 14U, in some examples the methods and/or devices comprise determining a polarity of the sensor signal. As previously described above with reference to at least FIGS. 1 B, 5A, and 5B, the polarity of the sensor signal may be used to identify each respiratory phase of the patient and indicates whether the slope of the inspiratory phase is positive or negative and/or whether the slope of the expiratory phase is negative or positive. In some examples, the polarity of the sensor signal may be determined using any one of the methods described below with reference to at least FIGS. 14V-14Y, 15C-15D, and 15H.

[0166] In some examples associated with 504 of FIG. 14E and 523 of FIG. 14U, as shown at 524 in FIG. 14V, determining the polarity of the sensor signal may comprise identifying a positive half cycle of the sensor signal as an inspiration half cycle (e.g., 154B of FIG. 5A) and/or identifying a negative half cycle of the sensor signal as an expiration half cycle (e.g., 154C of FIG. 5A) of the sensor signal in response to determining the polarity of the sensor signal is positive (e.g., 142A of FIG. 5A) and/or identifying a positive half cycle of the sensor signal as an expiration half cycle (e.g., 154C of FIG. 5B) and/or identifying a negative half cycle of the sensor signal as an inspiration half cycle (e.g., 154B of FIG. 5B) of the sensor signal in response to determining the polarity of the sensor signal is negative (e.g., 142B of FIG. 5B).

[0167] In some examples associated with 523 of FIG. 14U, as shown at 416 in FIG. 14W is a diagram schematically representing an example method and/or device for determining a polarity of a sensor signal. As shown at 525 in FIG. 14W, determining the polarity of the sensor signal comprises determining a median of samples (e.g., data points) of the sensor signal between a current zero crossing (e.g., TI.MID 153 at current time 151 in FIG. 5A or 5B) and at least two previous zero crossings. As shown at 526 in FIG. 14W, determining the polarity further comprises identifying the polarity of the sensor signal as positive (e.g., signal 142A of FIG. 5A) in response to the median being less than 0. As shown at 527 in FIG. 14W, determining the polarity further comprises identifying the polarity of the sensor signal as negative (e.g., signal 142B of FIG. 5B) in response to the median being greater than 0.

[0168] In some examples associated with 504 of FIG. 14E and 523 of FIG. 14U, as shown at 418 in FIG. 14X is a diagram schematically representing another example method and/or device for determining a polarity of a sensor signal. As shown at 528 in FIG. 14X, determining the polarity of the sensor signal comprises determining a number of samples (e.g., data points) in each of a positive half cycle (e.g., 154B of FIG. 5A or 154C of FIG. 5B) and a negative half cycle (e.g., 154C of FIG. 5A or 154B of FIG. 5B) of the sensor signal. As shown at 529 in FIG. 14X, determining the polarity further comprises identifying the positive half cycle as including a median of samples of the sensor signal between a current zero crossing (e.g., TI.MID 153 at current time 151 in FIG. 5A or 5B) and at least two previous zero crossings in response to the number of samples of the positive half cycle exceeding the number of samples of the negative half cycle. As shown at 530 in FIG. 14X, determining the polarity further comprises identifying the negative half cycle as including the median of data points of the sensor signal between the current zero crossing and the at least two previous zero crossings in response to the number of samples of the negative half cycle exceeding the number of samples of the positive half cycle. As shown at 531 in FIG. 14X, determining the polarity further comprises in response to identifying the positive half cycle as including the median of the samples of the sensor signal, determining the median of the samples of the sensor signal within the positive half cycle. As shown at 532 in FIG. 14X, determining the polarity further comprises in response to identifying the negative half cycle as including the median of the samples of the sensor signal, determining the median of the samples of the sensor signal within the negative half cycle. As shown at 533 in FIG. 14X, determining the polarity further comprises identifying the polarity of the sensor signal as positive (e.g., signal 142A of FIG. 5A) in response to the median being less than 0. As shown at 534 in FIG. 14X, determining the polarity further comprises identifying the polarity of the sensor signal as negative (e.g., signal 142B of FIG. 5B) in response to the median being greater than 0. The method of FIG. 14X improves the efficiency of the median determination since the half cycle including the most samples indicates the location of the median. The location of the median within the half cycle including the most samples is then determined according to the total number of samples from both half cycles which minimizes sorting of the samples (which may result in computational savings compared to method 416 of FIG. 14W).

[0169] In some examples associated with 504 of FIG. 14E and 523 of FIG. 14U, as shown at 420 in FIG. 14Y is a diagram schematically representing another example method and/or device for determining a polarity of a sensor signal. As shown at 535 in FIG. 14Y, determining the polarity of the sensor signal may comprise determining a number of samples in each of a positive half cycle (e.g., 154B of FIG. 5A or 154C of FIG. 5B) and a negative half cycle (e.g., 154C of FIG. 5A or 154B of FIG. 5B) of the sensor signal. As shown at 536 in FIG. 14Y, determining the polarity further comprises identifying the polarity of the sensor signal as positive (e.g., signal 142A of FIG. 5A) in response to the negative half cycle including more samples than the positive half cycle. As shown at 537 in FIG. 14Y, determining the polarity further comprises identifying the polarity of the sensor signal as negative (e.g., signal 142B of FIG. 5B) in response to the positive half cycle including more samples than the negative half cycle. In this example, the median is not determined. Thus, this example of determining polarity does not include any sorting of samples, which may result in computational saving compared to method 416 of FIG. 14W or method 418 of FIG. 14X.

[0170] In some examples associated with FIG. 14E and FIG. 14Y, as shown at 424 in FIG. 15A is a diagram schematically representing an example method and/or device for determining a confidence factor of a sensor signal. As shown at 541 in FIG. 15A, the methods and/or devices comprise incrementing a counter each time a respective positive half cycle (e.g., 154B of FIG. 5A or 154C of FIG. 5B) includes more samples than a respective negative half cycle (e.g., 154C of FIG. 5A or 154B of FIG. 5B). As shown at 542 in FIG. 15A, the methods and/or devices further comprise decrementing the counter each time a respective negative half cycle includes more samples than a respective positive half cycle. In some examples associated with FIG. 15A, as shown at 543 in FIG. 15B, the methods and/or devices may comprise determining a confidence factor based on a current value of the counter. In some examples, the confidence factor may be set equal to the absolute value of the counter. [0171] In some examples associated with 523 of FIG. 14U, as shown at 547 in FIG. 15C, the methods and/or devices comprise demeaning low passed samples of the last two half cycles of the sensor signal prior to determining the polarity of the sensor signal such that a high pass filter is excluded. Accordingly, in this case, the bandpass filter may be replaced with a low pass filter. In some examples associated with 547 of FIG. 15C, as shown at 428 in FIG. 15D is a diagram schematically representing another example method and/or device for determining a polarity of a sensor signal. As shown at 548 in FIG. 15D, determining the polarity of the sensor signal comprises determining a median of data points of the demeaned low passed samples of the last two half cycles. As shown at 549 in FIG. 15D, determining the polarity further comprises identifying the polarity of the sensor signal as positive (e.g., signal 142A of FIG. 5A) in response to the median being less than 0. As shown at 550 in FIG. 15D, determining the polarity further comprises identifying the polarity of the sensor signal as negative (e.g., signal 142B of FIG. 5B) in response to the median being greater than 0.

[0172] In some examples associated with 523 of FIG. 14U, as shown at 430 in FIG. 15E is a diagram schematically representing an example method and/or device for selecting an axis of a sensor signal, including identifying a polarity of each axis of the sensor signal. The selected axis (which may have the highest confidence factor) may be used for further calculations/determinations. As shown at 551 in FIG. 15E, determining the polarity of the sensor signal comprises determining a median of data points of a first axis (e.g., X-axis) of the sensor signal between a current zero crossing (e.g., Ti, MID 153 at current time 151 in FIG. 5A or 5B) and at least two previous zero crossings. As shown at 552 in FIG. 15E, determining the polarity further comprises identifying the polarity of the first axis of the sensor signal as positive in response to the median being less than 0. As shown at 553 in FIG. 15E, determining the polarity further comprises identifying the polarity of the first axis of the sensor signal as negative in response to the median being greater than 0. As shown at 554 in FIG. 15E, determining the polarity further comprises determining a median of data points of a second axis (e.g., Y-axis or Z-axis) of the sensor signal between a current zero crossing and at least two previous zero crossings. As shown at 555 in FIG. 15E, determining the polarity further comprises identifying the polarity of the second axis of the sensor signal as positive in response to the median being less than 0. As shown at 556 in FIG. 15E, determining the polarity further comprises identifying the polarity of the second axis of the sensor signal as negative in response to the median being greater than 0. As shown at 557 in FIG. 15E, determining the polarity further comprises comparing the identified polarity of the first axis of the sensor signal to the identified polarity of the second axis of the sensor signal. As shown at 558 in FIG. 15E, determining the polarity further comprises using the first axis of the sensor signal and/or the second axis of the sensor signal for further calculations (e.g., detecting zero crossings, calculating respiration period, respiration rate, etc.) in response to the identified polarity of the first axis of the sensor signal matching the identified polarity of the second axis of the sensor signal. In this example, the identified polarity of the first axis is confirmed by the identified polarity of the second axis, which may increase a confidence factor of the first axis (and/or the second axis) of the sensor signal.

[0173] In some examples associated with any of FIGS. 14U-14Y and 15C-15E, as shown at 559 in FIG. 15F, the methods and/or devices may comprise determining a confidence factor associated with the identified polarity of the sensor signal using a statistical test (e.g., student’s T-test).

[0174] As shown at 560 in FIG. 15G, in some examples, the methods and/or devices comprise identifying inspiration half cycles (e.g., 154B of FIGS. 5A and 5B) and expiration half cycles (e.g., 154C of FIGS. 5A and 5B) of the sensor signal. As previously described, the identified inspiration half cycles and expiration half cycles may be used to identify each expiratory phase and/or each inspiratory phase of the sensor signal.

[0175] In some examples associated with 504 of FIG. 14E and 560 of FIG. 15G, as shown at 432 in FIG. 15H is a diagram schematically representing an example method and/or device for identifying inspiration and expiration half cycles of the sensor signal. As shown at 561 in FIG. 15H, identifying inspiration half cycles and expiration half cycles comprises determining a number of samples in each of a positive half cycle and a negative half cycle of the sensor signal. As shown at 562 in FIG. 15H, identifying inspiration half cycles and expiration half cycles further comprises identifying the positive half cycle as an inspiration half cycle (e.g., 154A of FIG. 5A) and/or the negative half cycle as an expiration half cycle (e.g., 154C of FIG. 5A) in response to the number of the samples in the positive half cycle being less than the number of samples in the negative half cycle. As shown at 563 in FIG. 15H, identifying inspiration half cycles and expiration half cycles further comprises identifying the negative half cycle as an inspiration half cycle (e.g., 154B of FIG. 5B) and/or the positive half cycle as an expiration half cycle (e.g., 154C of FIG. 5B) in response to the number of the samples in the positive half cycle being greater than the number of samples in the negative half cycle.

[0176] The followings FIGS. 15I-16L relate to applying electrical stimulation based on the detected zero crossings. As previously noted, the zero crossings of the sensor signal may be used to determine respiration information, such as a respiration period, a respiration rate, inspiratory and/or expiratory phases, etc. without using the respiration information to apply electrical stimulation. In some examples, the respiration information may be used for applying electrical stimulation to the patient to treat sleep disordered breathing. In some examples, however, the respiration information may be used for diagnostic, measuring disease burden, therapeutic, etc. purposes without necessarily applying electrical stimulation. The respiration information may also be applied to contexts in addition to, or instead or, care of sleep disordered breathing.

[0177] In some examples, as previously noted, a predicted zero crossing (e.g., midpoint) of a future inspiratory phase (e.g., PTI.MID of FIGS. 5A or 5B) or a predicted zero crossing of a future expiratory phase (PTE.MID) may be used to schedule the start and/or stop of electrical stimulation. Accordingly, in some examples, predicted zero crossings PTI.MID and/or PTE.MID may be used in place of the current zero crossings TI.MID and/or TE.MID, respectively, to schedule the start and/or stop of electrical stimulation in the following FIGS. 15I-16G.

[0178] FIG. 151 is a diagram schematically representing an example method and/or device 434 for scheduling start and stop times of electrical stimulation based on detected zero crossings of the sensor signal. The current zero crossing may be TI.MID or TE.MID. In some examples where the current zero crossing is TI.MID (e.g., 153 of FIGS. 5A and 5B), as shown at 564 in FIG. 151, the methods and/or devices comprise scheduling a start of electrical stimulation (e.g., for the current inspiratory phase) based on the current TI.MID. AS shown at 565 in FIG. 151, the methods and/or devices further comprise scheduling a stop of electrical stimulation based on the current TI.MID. In some examples associated with 564 of FIG. 151, as shown at 566 in FIG. 15J, electrical stimulation may be started immediately in response to electrical stimulation not already being started. For example, if electrical stimulation is not active at the time the current zero crossing TI.MID is detected, some of the inspiratory phase (where stimulation is desired) has already been missed so electrical stimulation should be started as soon as possible. In some examples associated with 565 of FIG. 151, as shown at 567 in FIG. 15K, scheduling the stop of electrical stimulation may comprise scheduling the stop of electrical stimulation one half average respiration period (e.g., as determined at 513 of FIG. 14N) in the future. One half average respiration period in the future approximately corresponds to the end of the inspiration half cycle (i.e., the zero crossing TE.MID immediately following the current TI.MID). In some examples, the electrical stimulation may be stopped in response (e.g., directly in response) to the zero crossing TE.MID, such that the stop of electrical stimulation might not be scheduled in advance.

[0179] In some examples associated with 565 of FIG. 151, as shown at 568 in FIG. 15L, scheduling the stop of electrical stimulation may comprise scheduling the stop of electrical stimulation at the end of the current inspiration half cycle. In some examples associated with 565 of FIG. 151, as shown at 569 in FIG. 15M, scheduling the stop of electrical stimulation may comprise scheduling the stop of electrical stimulation 1.5 times the average respiration period in the future. In some examples associated with 565 of FIG. 151, as shown at 570 in FIG. 15N, scheduling the stop of electrical stimulation may comprise scheduling the stop of electrical stimulation within a range between 0.6 and 2 times the average respiration period in the future.

[0180] The methods and/or devices of FIGS. 151-15N described above and FIGS. 15O-16G described below enable the use of a high-latency filter to filter the sensor signal. In other examples, the sensor signal may be filtered by a low- latency filter. Where a low-latency filter is used, in some examples where the current zero crossing is TI.MID, the methods and/or devices comprise scheduling a stop of electrical stimulation based on the current TI.MID. The stop of electrical stimulation may be scheduled one half average respiration period in the future, at the end of the current inspiration half cycle, 1 .5 times the average respiration period in the future, or within a range between 0.6 and 2 times the average respiration period in the future. In addition, where a low-latency filter is used, in examples where the current zero crossing is TE.MID, the methods and/or devices may comprise scheduling a start of electrical stimulation based on the current TE.MID. The start of electrical stimulation may be scheduled immediately in response to the electrical stimulation not already being started, one average respiration period in the future, at the beginning of the next inspiration half cycle, or within a selectable range after a previous electrical stimulation stops.

[0181] FIG. 150 is a diagram schematically representing another example method and/or device 436 for scheduling start and stop times of electrical stimulation based on detected zero crossings of the sensor signal. In some examples where the current zero crossing is TI.MID, as shown at 571 in FIG. 150, the methods and/or devices comprise scheduling a future start of electrical stimulation (e.g., for the following inspiratory phase) based on the current TI.MID. As shown at 572 in FIG. 150, the methods and/or devices further comprise scheduling a future stop of electrical stimulation based on the current TI.MID. In some examples associated with 571 of FIG. 150, as shown at 573 in FIG. 15P, scheduling the future start of electrical stimulation may comprise scheduling the future start of electrical stimulation one average respiration period in the future. In some examples associated with 571 of FIG. 150, as shown at 574 in FIG. 15Q, scheduling the future start of electrical stimulation may comprise scheduling the future start of electrical stimulation at the beginning of the next inspiration half cycle (e.g., the next TI.MID).

[0182] In some examples associated with 571 of FIG. 150, as shown at 575 in FIG. 15R, scheduling the future start of electrical stimulation may comprise scheduling the future start of electrical stimulation within a selectable range after a previous electrical stimulation stops.

[0183] In some examples associated with FIG. 14C and/or FIG. 14Q, as shown at 577 in FIG. 15S, the methods and/or devices may comprise switching from a closed loop mode to an open loop mode in response to a standard deviation of a predetermined number of respiration periods exceeding a threshold value. For example, if the standard deviation exceeds a threshold value within a range between 10 milliseconds and 5000 milliseconds (e.g., within a range between 100 milliseconds and 1000 milliseconds), stimulation may be switched from a closed loop mode to an open loop mode. In some examples, the threshold value may be configurable by a clinician and/or physician. In some examples, the methods and/or devices may comprise switching from the open loop mode back to the closed loop mode in response to the standard deviation of the predetermined number of respiration periods falling back below the threshold value. Thus, electrical stimulation may switch between an open loop mode and a closed loop mode throughout a treatment period to ensure uninterrupted treatment.

[0184] FIG. 16A is a diagram schematically representing another example method and/or device 438 for scheduling start and stop times of electrical stimulation based on detected zero crossings of the sensor signal. In some examples where the current zero crossing is TE.MID (e.g., 152 of FIGS. 5A and 5B), as shown at 578 in FIG. 16A, the methods and/or devices comprise scheduling a start of electrical stimulation based on the current TE, MID. AS shown at 579 in FIG. 16A, the methods and/or devices further comprise scheduling a stop of electrical stimulation (or immediately stopping electrical stimulation) based on the current TE.MID. In some examples associated with 578 of FIG. 16A, as shown at 580 in FIG. 16B, scheduling the start of electrical stimulation may comprise scheduling the start of electrical stimulation one half average respiration period (e.g., as determined at 513 of FIG. 14N) in the future. One half average respiration period in the future approximately corresponds to the start of the next inspiration half cycle (i.e., the zero crossing TI.MID immediately following the current TE.MID). In some examples associated with 578 of FIG. 16A, as shown at 581 in FIG. 16C, scheduling the start of electrical stimulation may comprise scheduling the start of electrical stimulation at the beginning of the next inspiration half cycle (e.g., the next TI.MID).

[0185] In some examples associated with 578 of FIG. 16A, as shown at 582 in FIG. 16D, scheduling the start of electrical stimulation may comprise scheduling the start of electrical stimulation within a range between -2 and 2 times the average respiration period in the future relative to the next TE.MID (i.e., the TE.MID immediately following the current TE.MID). In some examples associated with 579 of FIG. 16A, as shown at 583 in FIG. 16E, scheduling the stop of electrical stimulation may comprise scheduling the stop of electrical stimulation one average respiration period in the future. In some examples associated with 579 of FIG. 16A, as shown at 584 in FIG. 16F, scheduling the stop of electrical stimulation may comprise scheduling the stop of electrical stimulation at the start of the next expiration half cycle. In some examples associated with 579 of FIG. 16A, as shown at 585 in FIG. 16G, scheduling the stop of electrical stimulation may comprise replacing (e.g., overwriting) a previously scheduled stop of electrical stimulation in response to TI MID occurring (e.g., as described with reference to FIGS. 15K-15N).

[0186] In some examples, a method of stimulating an upper airway patency- related nerve (e.g., hypoglossal nerve, infrahyoid muscle (IHM)-innervating nerve, phrenic nerve) may include chronically implanting a stimulation element (e.g., stimulation element 617 to be described with reference to at least FIG. 17A) in or near an upper airway. The method may include sensing a signal (e.g., 142 of FIG. 4) corresponding to respiration. The method may include detecting one or more zero crossings of the sensor signal (e.g., TI.MID, TE.MID). The method may include calculating a future onset of an inspiratory phase from the detected one or more zero crossings. The method may include delivering stimulation to the upper airway patency-related nerve via the stimulation element. The stimulation may start a predetermined amount of time prior to the calculated onset of the inspiratory phase.

[0187] In some examples, a method of treating obstructive sleep apnea may include chronically implanting an electrode (e.g., electrode 672 to be described with reference to at least FIG. 17B) on a nerve innervating an upper airway muscle. The method may include sensing a measure of respiration (e.g., signal 142 of FIG. 4). The method may include analyzing the measure of respiration to identify midpoints of expiration (e.g., TE,MID). The method may include calculating a respiratory period from the midpoints of expiration. The method may include predicting the midpoint of a future expiratory phase (e.g., PTE.MID). The method may include beginning stimulation of the nerve a fraction of the calculated respiratory period before the midpoint of the future expiratory phase, and continuing stimulation of the nerve during an entire inspiratory phase. In some examples, the method is performed without identifying an onset of the inspiratory phase.

[0188] FIG. 16H is a diagram schematically representing an example method and/or device 440 for selecting a sensor signal for detecting zero crossings. In some examples, the sensor to sense changes in acceleration indicative of respiration of the patient comprises an at least two axis accelerometer (e.g., 204 in FIGS. 7A and 7B). The at least two axis accelerometer provides at least two respective sensor signals. By selecting one axis of the at least two axis accelerometer, the unselected axis or axes may be excluded from further filtering and processing, thereby resulting in computational savings. In some examples as shown at 586 in FIG. 16H and as previously described with reference to at least FIG. 1 B, the methods and/or devices comprise determining a confidence factor associated with each of the at least two respective sensor signals. The confidence factor associated with each of the least two sensor signals may be determined as described below with reference to at least FIGS. 18A and 18B. As shown at 587 in FIG. 16H, the methods and/or devices further comprise identifying which respective sensor signal exhibits the highest confidence factor. As shown at 588 in FIG. 16H, the methods and/or devices further comprise detecting zero crossing of the sensor signal exhibiting the highest confidence factor. In some examples associated with 586 of FIG. 16H, as shown at 589 in FIG. 161, determining the confidence factor may comprise updating the confidence factor associated with each of the at least two respective sensor signals in response to each detected zero crossing. Accordingly, in this example, the sensor signal having the highest confidence factor may be used for further calculations/determinations, and the selected sensor signal may change throughout a treatment period.

[0189] In some examples associated with 500 of FIG. 14A and 440 of FIG. 16H, as shown at 590 in FIG. 16J, applying the electrical stimulation may comprise applying the electrical stimulation in an open loop mode or a closed loop mode. In some examples, the open loop mode is assigned a predetermined open loop mode confidence factor. Thus, the open loop mode confidence factor may be used as a threshold value for switching (e.g., converting) between the open loop mode and the closed loop mode. In some examples associated with 590 of FIG. 16J, as shown at 442 in FIG. 16K is a diagram schematically representing an example method and/or device for applying electrical stimulation in an open loop mode or a closed loop mode. As shown at 591 in FIG. 16K, applying the electrical stimulation may comprise applying the electrical stimulation in the closed loop mode in response to the respective sensor signal exhibiting the highest confidence factor comprising a confidence factor greater than the predetermined open loop mode confidence factor. As shown at 592 in FIG. 16K, applying the electrical stimulation may comprise applying the electrical stimulation in the open loop mode in response to the respective sensor signal exhibiting the highest confidence factor comprising a confidence factor less than the predetermined open loop mode confidence factor. Thus, in this example, electrical stimulation may switch (e.g., convert) between an open loop mode and a closed loop mode throughout a treatment period to ensure uninterrupted treatment.

[0190] In some examples, the open loop mode confidence factor may be adjusted over time (e.g., from treatment period to treatment period) based on sensing disease burden indicator(s) (e.g., a frequency of apnea and/or hypopnea events such as AHI) during a given treatment period. For instance, upon sensing increases in disease burden indicator(s) on a treatment-period-to-treatment period basis, one may adjust the open loop mode confidence factor to be higher or lower to increase the likelihood of selecting the stimulation source that was not used the previous treatment period. For example, if disease burden indicator(s) increased the previous night and therapy was open loop, make open loop less likely by lowering the open loop mode confidence factor. If disease burden ind icator(s) decreased the previous night and therapy was open loop, make open loop more likely by raising the open loop mode confidence factor. If disease burden indicator(s) increased the previous night and therapy was closed loop, make open loop more likely by raising the open loop mode confidence factor. If disease burden indicator(s) decreased the previous night and therapy was closed loop, make open loop less likely by lowering the open loop mode confidence factor. In this example, the open loop mode confidence factor may be adjusted using a machine learning model.

[0191] In some examples, the electrical stimulation duty cycle is less than or equal to 75% (i.e., during a treatment period, electrical stimulation is applied for up to 75% of each respiration period). In some example, the electrical stimulation duty cycle may be within a range greater than 25% and less than or equal to 75%. In some examples, the electrical stimulation duty cycle may be within a range greater than 50% and less than or equal to 75%. In some such examples associated with any of FIGS. 15I-15R and 16A-16G, as shown at 593 in FIG. 16L, the methods and/or devices may comprise adjusting stimulation start times and/or stimulation stop times based on the electrical stimulation duty cycle. By adjusting stimulation start times and/or stimulation stop times based on the electrical stimulation duty cycle, it may be ensured that the applied electrical stimulation has the selected duty cycle throughout a treatment period.

[0192] FIG. 17A is a block diagram schematically representing a patient’s body 600, including example target portions 610-634 at which at least some example sensing element(s) and/or stimulation elements may be employed to implement at least some examples of the present disclosure previously described with reference to FIGS. 1A-16L.

[0193] As shown in FIG. 17A, patient’s body 600 comprises a head-and-neck portion 610, including head 612 and neck 614. Head 612 comprises cranial tissue, nerves, etc., and upper airway 616 (e.g., nerves, muscles, tissues), etc. As further shown in FIG. 17A, the patient’s body 600 comprises a torso 620, which comprises various organs, muscles, nerves, other tissues, such as but not limited to those in pectoral region 622 (e.g., lungs 626, cardiac 627), abdomen 624, and/or pelvic region 629 (e.g., urinary/bladder, anal, reproductive, etc.). As further shown in FIG. 17A, the patient’s body 600 comprises limbs 630, such as arms 632 and legs 634.

[0194] It will be understood that various sensing elements (and/or stimulation elements) as described throughout the various examples of the present disclosure may be deployed within the various regions of the patient’s body 600 to sense and/or otherwise diagnose, monitor, treat various physiologic conditions such as, but not limited to those examples described in association with FIGS. 1A-20. In some such examples, a stimulation element 617 may be located in or near the upper airway 616 for treating sleep disordered breathing (and/or near other nerves/muscles for treating other conditions) and/or a sensing element 628 may be located anywhere within the neck 614 and/or torso 620 (or other body regions) to sense physiologic information for providing patient care (e.g., SDB, other) with the sensed physiologic information including, but not limited to, respiration information and related parameters. In some examples, sensing element 628 may comprise a sensor (e.g., 12A, 12B, 104A, 122A, 162, 164, 204, 304, etc.) having at least some of substantially the same features and attributes as previously described in association with at least FIGS. 1A-13. In some examples, sensing element 628 may comprise electrodes for impedance sensing, a microphone or piezoelectric sensor for acoustic sensing, and/or an element (e.g., electrode) for electroneurography (ENG), electromyography (EMG), or other biosignal sensing. For impedance sensing, the electrodes may be placed at multiple spaced apart locations in/on the neck and/or in/on the chest to obtain bioimpedance measurements.

[0195] In some examples, sensing may be implemented via at least some of substantially the same features as disclosed in Dieken et al., RESPIRATION DETECTION, published as US 2023/0119173 on April 20, 2023; Thorp et al., DISEASE BURDEN INDICATION, published as US 2023/0277121 on September 7, 2023; and Thorp et al., RESPIRATION SENSING, published as PCT application WO 2022/261311 on December 15, 2022 and corresponding to US National Stage S/N 18/287,205 filed on October 17, 2023, published as ; all of which are incorporated herein by reference. [0196] In some examples, the stimulation element 617 and/or additional stimulation elements (e.g., like element 617 or different from element 617) may be located more generally in the neck and/or more generally at some portion near or at airway structures not strictly confined to the upper airway. In some examples, at least a portion of the stimulation element 617 may comprise part of an implantable component/device, such as an implantable pulse generator (IPG) whether full sized or sized as a microstimulator. The implantable components (e.g., IPG, other) may comprise a stimulation/control circuit, a power supply (e.g., non-rechargeable, rechargeable), communication elements, and/or other components. In some examples, the stimulation element 617 also may comprise a stimulation electrode and/or stimulation lead connected to the implantable pulse generator.

[0197] Further details regarding a location, structure, operation, and/or use of the sensing element 628 and/or stimulation element 617 are described below in association with at least FIG. 17B.

[0198] In some examples, at least a portion of the stimulation element 617 may comprise part of an external component/device such as, but not limited to, the external component comprising a pulse generator (e.g., stimulation/control circuitry), power supply (e.g., rechargeable, non-rechargeable), and/other components. In some examples, a portion of the stimulation element 617 may be implantable and a portion of the stimulation element 617 may be external to the patient.

[0199] Accordingly, as further shown in FIG. 17A, the various sensing element(s) 628 and/or stimulation element(s) 617 implanted in the patient’s body may be in wireless communication (e.g., connection 637) with at least one external element 640.

[0200] As further shown in FIG. 17A, in some examples, the external element(s) 640 may be implemented via a wide variety of formats such as, but not limited to, at least one of the formats 641 including a patient support 642 (e.g., bed, chair, sleep mat, other), wearable elements 644 (e.g., finger, wrist, head, neck, shirt), noncontact elements 646 (e.g., watch, camera, mobile device, other), and/or other elements 648. [0201] As further shown in FIG. 17A, in some examples, the external element(s) 640 may comprise one or more different modalities 650 such as (but not limited to) a sensing portion 651 , stimulation portion 652, power portion 654, communication portion 656, and/or other portion 658. The different portions 651 , 652, 654, 656, 658 may be combined into a single physical structure (e.g., package, arrangement, assembly), may be implemented in multiple different physical structures, and/or with just some of the different portions 651 , 652, 654, 656, 658 combined together in a single physical structure.

[0202] In some examples, the external power portion 654 and/or power components associated with implanted stimulation element 617 may comprise at least some of substantially the same features and attributes of at least the stimulation arrangements, as further described below in association with at least FIG. 17B and/or other examples throughout the present disclosure. In some such examples, the respective power portion, components, etc. may comprise a rechargeable power element (e.g., supply, battery, circuitry elements) and/or non- rechargeable power elements (e.g., battery). In some examples, the external power portion 654 may comprise a power source by which a power component of the implanted stimulation element 617 may be recharged.

[0203] In some examples, the wireless communication portion 656 (e.g., connection/link at 637) may be implemented via various forms of radiofrequency communication and/or other forms of wireless communication, such as (but not limited to) magnetic induction telemetry, Bluetooth (BT), Bluetooth Low Energy (BLE), near infrared (NIF), near-field protocols, Wi-Fi, Ultra- Wideband (UWB), and/or other short range or long range wireless communication protocols suitable for use in communicating between implanted components and external components in a medical device environment.

[0204] Examples are not so limited as expressed by other portion 658 via which other aspects of implementing medical care may be embodied in external element(s) 640 to relate to the various implanted and/or external components described above.

[0205] FIG. 17B is a diagram including a front view of an example device 671 (and/or example method) implanted within a patient’s body 660. In some examples, the device 671 may comprise an implantable device 683 such as (but not limited to) an implantable pulse generator (IPG) with device 683 including a sensor 685. In some examples, sensor 685 may comprise a sensor (e.g., 12A, 12B, 104A, 122A, 162, 164, 204, 304, etc.) having at least some of substantially the same features and attributes as previously described in association with at least FIGS. 1A-13. Via such example sensing arrangements, the device 683 may determine respiration information via sensing rotational movement of the patient’s chest wall during breathing, such as but not limited to when in a sleeping body position during a treatment period.

[0206] As further shown in FIG. 17B, device 671 comprises a lead 677 including a lead body 678 for chronic subcutaneous implantation (e.g., via tunneling) and extends to a position adjacent a nerve, such as a hypoglossal nerve 665, an infrahyoid muscle (IHM)-innervating nerve (as one example of nerve 667) and/or phrenic nerve (as one example of nerve 667). The lead 677 may comprise a stimulation electrode 672 to engage the nerve (e.g., 665, 667) for stimulating the nerve to treat a physiologic condition, such as sleep disordered breathing like obstructive sleep apnea, central sleep apnea, multiple-type sleep apneas, etc. The device 683 may comprise circuitry, power element, etc. to support control and operation of both the sensor 685 and the stimulation electrode 672 (via lead 677). In some examples, such control, operation, etc. may be implemented, at least in part, via a control portion (and related functions, portions, elements, engines, parameters, etc.) such as described later in association with at least FIGS. 18A-20.

[0207] With regard to the various examples of the present disclosure, in some examples, delivering stimulation to an upper airway patency nerve (e.g., a hypoglossal nerve 665, infrahyoid muscle (IHM)-innervating nerve as one example of nerve 667, etc.) via the stimulation electrode 672 is to cause contraction of upper airway patency-related muscles, which may cause or maintain opening of the upper airway (668) to prevent and/or treat obstructive sleep apnea. Similarly, such electrical stimulation may be applied to a phrenic nerve (as one example of nerve 667) via the stimulation electrode 672 to cause contraction of the diaphragm as part of preventing or treating at least central sleep apnea. It will be further understood that some example methods may comprise treating both obstructive sleep apnea and central sleep apnea, such as but not limited to, instances of multiple-type sleep apnea in which both types of sleep apnea may be present at least some of the time. In some such instances, separate stimulation leads 677 may be provided or a single stimulation lead 677 may be provided but with a bifurcated distal portion (or trifurcated distal portions) with each separate distal portion extending to a respective one of the hypoglossal nerve 665, infrahyoid muscle (IHM)-innervating nerve (as one example of nerve 667), and/or the phrenic nerve (as one example of nerve 667).

[0208] In some examples, such stimulation electrodes 672, leads 677, and/or the IPG 683 may be arranged to deliver stimulation bilaterally to the hypoglossal nerve 665, infrahyoid muscle (IHM)-innervating nerve (one example of nerve 667), and/or phrenic nerve (one example of nerve 667) including various combinations of simultaneous, alternating, sequential stimulation patterns among the left and right sides, the different nerves, etc.

[0209] It will be further understood that in some examples, the stimulation electrode 672, lead 677, and/or IPG 683 (including sensor 685, such as an accelerometer) may be embodied as a microstimulator.

[0210] In some such examples, the contraction of the upper airway patency- related muscle(s) caused by electrical stimulation of nerves (e.g., hypoglossal nerve, infrahyoid muscle (IHM)-innervating nerve) comprises a suprathreshold stimulation, which is in contrast to a subthreshold stimulation (e.g., mere tone) of such muscles. In one aspect, a suprathreshold intensity level corresponds to a stimulation energy greater than the nerve excitation threshold, such that the suprathreshold stimulation may provide for higher degrees (e.g., maximum, other) upper-airway clearance (i.e. , patency) and sleep apnea therapy efficacy.

[0211] In some examples, a target intensity level of stimulation energy is selected, determined, implemented, etc. without regard to intentionally establishing a discomfort threshold of the patient (such as in response to such stimulation). Stated differently, in at least some examples, a target intensity level of stimulation may be implemented to provide the desired efficacious therapeutic effect in reducing sleep disordered breathing (SDB) without attempting to adjust or increase the target intensity level according to (or relative to) a discomfort threshold.

[0212] In some examples, the treatment period (during which stimulation may be applied at least part of the time) may comprise a period of time beginning with the patient turning on the therapy device and ending with the patient turning off the device. In some examples, the treatment period may comprise a selectable, predetermined start time (e.g., 10 p.m.) and selectable, predetermined stop time (e.g., 6 a.m.). In some examples, the treatment period may comprise a period of time between an auto-detected initiation of sleep and auto-detected awake-from- sleep time. With this in mind, the treatment period corresponds to a period during which a patient is sleeping such that the stimulation of the upper airway patency- related nerve and/or central sleep apnea-related nerve is generally not perceived by the patient and so that the stimulation coincides with the patient behavior (e.g., sleeping) during which the sleep disordered breathing behavior (e.g., central or obstructive sleep apnea) would be expected to occur. In some examples the initiation or termination of the treatment period may be implemented automatically based on sensed sleep state information, which in turn may comprise sleep stage information.

[0213] To avoid enabling stimulation prior to the patient falling asleep, in some examples stimulation can be enabled after expiration of a timer started by the patient (to enable therapy with a remote control), or enabled automatically via sleep stage detection. To avoid continuing stimulation after the patient wakes, stimulation can be disabled by the patient using a remote control, or automatically via sleep stage detection. Accordingly, in at least some examples, these periods may be considered to be outside of the treatment period or may be considered as a startup portion and wind down portion, respectively, of a treatment period. [0214] In some examples, stimulation of an upper airway patency-related nerve may be performed via open loop stimulation. In some examples, the open loop stimulation may refer to performing stimulation without use of any sensory feedback of any kind relative to the stimulation.

[0215] In some examples, the open loop stimulation may refer to stimulation performed without use of sensory feedback by which timing of the stimulation (e.g., synchronization) would otherwise be determined relative to respiratory information (e.g., respiratory cycles). However, in some such examples, some sensory feedback may be utilized to determine, in general, whether the patient should receive stimulation based on a severity of sleep apnea behavior.

[0216] Conversely, in some examples and as previously described in relation to at least several examples, stimulation of an upper airway patency-related nerve may be performed via closed loop stimulation. In some examples, the closed loop stimulation may refer to performing stimulation at least partially based on sensory feedback regarding parameters of the stimulation and/or effects of the stimulation.

[0217] In some examples, the closed loop stimulation may refer to stimulation performed via use of sensory feedback by which timing of the stimulation (e.g., synchronization) is determined relative to respiratory information, such as but not limited to respiratory cycle information, which may comprise onset, offset, duration, magnitude, morphology, etc. of various features of the respiratory cycles, including but not limited to the inspiratory phase, expiratory active phase, etc. In some examples, the respiration information excludes (i.e., is without) tracking a respiratory volume and/or respiratory rate. In some examples, stimulation based on such synchronization may be delivered throughout a treatment period or throughout substantially the entire treatment period. In some examples, such stimulation may be delivered just during a portion or portions of a treatment period.

[0218] In some examples of “synchronization”, synchronization of the stimulation relative to the inspiratory phase may extend to a pre-inspiratory period and/or a post-inspiratory phase. For instance, in some such examples, a beginning of the synchronization may occur at a point in each respiratory cycle which is just prior to an onset of the inspiratory phase. In some examples, this point may be about 200 milliseconds, or 300 milliseconds prior to an onset of the inspiratory phase.

[0219] In some examples in which the stimulation is synchronous with at least a portion of the inspiratory phase, the upper airway muscles are contracted via the stimulation to ensure they are open at the time the respiratory drive controlled by the central nervous system initiates an inspiration (inhalation). In some such examples, in combination with the stimulation occurring during the inspiratory phase, example implementation of the above-noted pre-inspiratory stimulation helps to ensure that the upper airway is open before the negative pressure of inspiration within the respiratory system is applied via the diaphragm of the patient’s body. In one aspect, this example arrangement may minimize the chance of constriction or collapse of the upper airway, which might otherwise occur if flow of the upper airway flow were too limited prior to the full force of inspiration occurring. In some such examples, the stimulation of the upper airway patency-related nerve may be synchronized to occur with at least a portion of the expiratory period.

[0220] With regard to at least the methods of treating sleep apnea as previously described in association with at least FIGS. 1A-16L, at least some such methods may comprise performing the delivery of stimulation to the upper airway patency- related first nerve without synchronizing such stimulation relative to a portion of a respiratory cycle. In some instances, such methods may sometimes be referred to as the previously described open loop stimulation.

[0221] In some examples, the term “without synchronizing” may refer to performing the stimulation independently of timing of a respiratory cycle. In some examples, the term “without synchronizing” may refer to performing the stimulation while being aware of respiratory information but without necessarily triggering the initiation of stimulation relative to a specific portion of a respiratory cycle or without causing the stimulation to coincide with a specific portion (e.g., inspiratory phase) of respiratory cycle.

[0222] In some examples, in this context the term “without synchronizing” may refer to performing stimulation upon the detection of sleep disordered breathing behavior (e.g., obstructive sleep apnea events) but without necessarily triggering the initiation of stimulation relative to a specific portion of a respiratory cycle or without causing the stimulation to coincide with the inspiratory phase. At least some such examples may be described in Wagner et al., STIMULATION FOR TREATING SLEEP DISORDERED BREATHING, published as US 2018/0117316 on May 3, 2018, and which is incorporated by reference herein in its entirety. [0223] In some examples, while open loop stimulation may be performed continuously without regard to timing of respiratory information (e.g., inspiratory phase, expiratory phase, etc.) such an example method and/or system may still comprise sensing respiration information for diagnostic data and/or to determine whether (and by how much) the continuous stimulation should be adjusted. For instance, via such respiratory sensing, it may be determined that the number of sleep disordered breathing (SDB) events are too numerous (e.g., an elevated AHI) and therefore the intensity (e.g., amplitude, frequency, pulse width, etc.) of the continuous stimulation should be increased or that the SDB events are relative low such that the intensity of the continuous stimulation can be decreased while still providing therapeutic stimulation. It will be understood that via such respiratory sensing, other SDB-related information may be determined which may be used for diagnostic purposes and/or used to determine adjustments to an intensity of stimulation, initiating stimulation, and/or terminating stimulation to treat sleep disordered breathing. It will be further understood that such “continuous” stimulation may be implemented via selectable duty cycles, train of stimulation pulses, selective activation of different combinations of electrodes, etc.

[0224] In some examples of open loop stimulation or closed loop stimulation, some sensory feedback may be utilized to determine, in general, whether the patient should receive stimulation based on a severity of sleep apnea behavior. In other words, upon sensing that a certain number of sleep apnea events are occurring, the device may implement stimulation.

[0225] Moreover, in some examples various stimulation methods may be applied to treat obstructive sleep apnea, which include but are not limited to: Ni et al., SYSTEM FOR SELECTING A STIMULATION PROTOCOL BASED ON SENSED RESPIRATORY EFFORT, which issued as U.S. 10,583,297 on March 10, 2020; Christopherson et al., US 8938299, SYSTEM FOR TREATING SLEEP DISORDERED BREATHING, issued January 20, 2015; and Wagner et al., STIMULATION FOR TREATING SLEEP DISORDERED BREATHING, published as US 2018/0117316 on May 3, 2018, each of which is hereby incorporated by reference herein in its entirety. [0226] In some examples, the example stimulation element(s) 672 shown in FIG. 17B may comprise at least some of substantially the same features and attributes as described in Bonde et al., U.S. 8,340,785, SELF EXPANDING ELECTRODE CUFF, issued on December 25, 2012 and Bonde et al., U.S. 9,227,053, SELF EXPANDING ELECTRODE CUFF, issued on January 5, 2016; Johnson et al., U.S. 8,934,992, NERVE CUFF issued on January 13, 2015; and Rondoni et al., U.S. 11 ,298,540, CUFF ELECTRODE, issued on April 12, 2022, each which are incorporated by reference herein in their entirety.

[0227] In some examples, the stimulation electrode 672 may be delivered transvenously, percutaneously, etc. In some such examples, a transvenous approach may comprise at least some of substantially the same features and attributes as described in Ni et al., Transvenous Method of Treating Sleep Apnea, issued as U.S. 9,889,299 on February 13, 2018, and which is hereby incorporated by reference. In some such examples, a percutaneous approach may comprise at least some of substantially the same features and attributes as described in Christopherson et al., Percutaneous Access For Systems and Methods Of Treating Sleep Apnea, issued as U.S. 9,486,628 on November 8, 2016, and which is hereby incorporated by reference.

[0228] As previously noted in various examples of the disclosure, in some examples an upper airway patency-related nerve may comprise an IHM- innervating nerve in addition to, or instead of, a hypoglossal nerve and/or other upper airway patency-related nerves.

[0229] In some examples, an IHM-innervating nerve may comprise a nerve or nerve branch which innervates (directly or indirectly) at least one infrahyoid muscle, which may sometimes be referred to as an infrahyoid strap muscle. In some examples, IHM-innervating nerves/nerve branches extend from (e.g., originates) from a nerve loop called the ansa cervicalis (AC) or the “AC loop nerve”, which stems from the cervical plexus, e.g., extending from cranial nerves C1 -C3. Accordingly, in some examples, at least some IHM-innervating nerves may correspond to an ansa cervicalis (AC)-related nerve in the sense that such nerves/nerve branches (e.g., IHM-innervating nerves) do not form the AC loop nerve but extend from the AC loop nerve. At least because the AC loop nerve is the origin for some nerves which innervate muscles other than the infrahyoid muscles, some AC-related nerves do not comprise IHM-innervating nerves. Moreover, it will be understood that in some examples, stimulation applied to a portion (e.g., superior root) of the AC loop nerve (and/or to nerves from which the AC loop nerve originates) may activate IHM-innervating nerves/nerve branches, which extend from the AC loop nerve. However, implementing stimulation (e.g., to influence upper airway patency) occurring at more proximal locations, such as along the superior root of the AC loop nerve may be more complex because of the number/type of different nerves and number/type of different muscles innervated via a superior root of the AC loop nerve such that selective activation of a particular infrahyoid muscle (via stimulation along the superior root) may be quite challenging in some circumstances.

[0230] With this background in mind, FIG. 17C is a diagram 1000 schematically representing patient anatomy and providing further details regarding example devices and/or example methods for stimulating an IHM-innervating nerve and/or hypoglossal nerve. As shown in FIG. 17C, diagram 1000 includes a side view schematically representing an AC-main nerve 1015, in context with a hypoglossal nerve 1005 and with cranial nerves C1 , C2, C3. As shown in FIG. 17C, portion 1029A of the AC-main nerve 1015 (e.g., a portion or trunk connecting to the AC loop nerve 1019) extends anteriorly from a first cranial nerve C1 with a segment 1017 running alongside (e.g., coextensive with) the hypoglossal nerve 1005 for a length until the AC-main nerve 1015 diverges from the hypoglossal nerve 1005 to form a superior root 1025 of the AC-main nerve 1015, which forms part of the AC loop nerve 1019. A portion of the hypoglossal nerve 1005 extends distally to innervate the genioglossus muscle 1004. As further shown in FIG. 17C, the superior root 1025 of the AC-main nerve 1015 extends inferiorly (i.e., downward) until reaching near bottom portion 1018 of the AC loop nerve 1019, from which the AC loop nerve 1019 extends superiorly (i.e., upward) to form an lesser root 1027 (i.e., inferior root) which joins to the second and third cranial nerves, C2 and C3, respectively and via portions 1029B, 1029C.

[0231] As further shown in FIG. 17C, several branches 1031 extend off the AC loop nerve 1019, including branch 1032 which innervates the omohyoid muscle group 1034, branch 1042 which innervates the sternothyroid muscle group 1044 and at least a portion (e.g., inferior portion) of the sternohyoid muscle group 1054. Another branch 1052, near bottom portion 1018 of the AC loop nerve 1019, innervates at least a portion (e.g., superior portion) of the sternohyoid muscle group 1054. In some examples, the collective arrangement of the AC-main nerve 1015 (including at least superior root 1025 of the AC loop nerve 1019) and its related branches (e.g., at least 1032, 1042, 1052) when considered together, or any of those elements individually, may sometimes be referred to as an IHM- innervating nerve 1016. It will be further understood that at least one such IHM- innervating nerve 1016 is present on both sides (e.g., right and left) of the patient’s body.

[0232] In some examples, stimulation of the superior root 1025 of AC loop nerve 1019 and/or at least some of the branches 1031 extending from the AC loop nerve 1019, may influence upper airway patency. However, in some examples, upper airway patency also may be increased and/or maintained by directly stimulating the above-identified muscle groups, such as the omohyoid, sternothyroid, and/or sternohyoid muscle groups. Accordingly, in some examples, such stimulation also may comprise stimulation of just a nerve portion(s), just muscle portion(s), a combination of nerve portion(s) and muscle portion(s), a neuromuscular junction of nerve portion(s) and muscle portion(s), and combinations thereof. Among other effects, in some examples stimulation of such nerves and/or muscles (and/or neuromuscular junctions, combinations, etc.) may act to bring the larynx inferiorly, which may increase upper airway patency.

[0233] Stimulation may be delivered to many different locations of an IHM- innervating nerve 1016/nerve branches. Of these various potential stimulation locations, FIG. 17C generally illustrates three example stimulation locations A, B, and C. A stimulation element may be placed at all three of these locations or just some (e.g., one or two) of these example stimulation locations. At each location, a wide variety of types of stimulation elements (e.g., cuff electrode, axial array, paddle electrode, etc.) may be implanted depending on the particular delivery path, method, etc. In some such examples, a scale of the various stimulation elements, anchors, access tools, and/or other elements may be reduced to accommodate a generally smaller diameter of the IHM-innervating nerve/nerve branches 1016 as compared to some other nerve portions, such as at least some portions of the hypoglossal nerve.

[0234] With further reference to FIG. 17C, at each example stimulation location A, B, C, a stimulation element may be delivered subcutaneously, intravascularly, etc. At each stimulation location, in some examples the stimulation element may comprise a microstimulator.

[0235] In some examples, delivering stimulation to an IHM-innervating nerve may be implemented via at least some of substantially the same features and attributes as described in the PCT application published on November 24, 2022 as WO 2022/246320 “Multiple target stimulation therapy for sleep disordered breathing”, corresponding to U.S. National Phase application , filed , published as> , and/or the PCT application published

December 2, 2021 , as WO 2021/242633 “Single or multiple nerve stimulation to treat sleep disordered breathing” corresponding to U.S. National Phase application 17/926010, filed November 17, 2022 and published as , both of which are incorporated herein by reference.

[0236] FIG. 18A is a block diagram schematically representing example method 700. In some examples, method 700 may be implemented via at least some of the devices, sensors, sensing elements, etc., and/or may comprise at least some of substantially the same features and attributes as, the examples previously described in association with FIGS. 1A-17B. For example, as described below, method 700 includes additional details and/or features for processing a signal(s) from a sensor (e.g., 12A, 12B, 104A, 122A, 162, 164, 204, 304, etc.), determining a confidence factor of the sensor signal(s) (e.g., via portion 24 of FIG. 1 B), extracting features (e.g., zero crossings as described with reference to FIGS. 5A and 5B) from the sensor signal(s), and determining respiratory phase information (e.g., via portion 22 of FIG. 1 B) from the sensor signal(s). In some examples, the example method 700 may be at least partially implemented within, and/or via, control portion 14A in FIG. 1A, control portion 14B in FIG. 1 B, control portion 900 in FIG. 19A, control portion 920 in FIG. 19B, or interface 940 in FIG. 20. [0237] As shown at 710 in FIG. 18A, the example method 700 comprises sensing acceleration signal(s) from a sensor(s) implanted within a patient’s body in a position, such as in the chest region, to detect respiration information. In some examples, just a single sensing element (e.g., 122A in FIG. 3) may be used to provide just a single sensed acceleration signal or in some examples, multiple sensing elements (e.g., 122A, 162, and/or 164 in FIGS. 7A-8) may be used to provide separate multiple sensed acceleration signals. The multiple sensing elements may be separate from, and independent of, each other, or may be colocated as part of a single device, such as a three-axis accelerometer (e.g., 204 of FIGS. 7A and 7B).

[0238] As further shown at 714, filtering is applied separately to the sensed signal(s) (710) to produce a respective separate inclination angle signal (721X, 721Y, 721Z) for each corresponding acceleration signal (e.g., X-axis, Y-axis, Z- axis). It will be understood that if just one single-axis sensing element is employed, then just one inclination angle signal will be present at 720 (e.g., as described with reference to FIG. 3). As previously described through various examples, the inclination angle signal represents the physiologic phenomenon of the patient’s breathing with a value and/or shape of the inclination angle signal varying through the different phases of a respiratory cycle (e.g., inspiratory phase, expiratory active phase, expiratory pause phase) as the patient breathes. It will be further noted that while some examples may comprise tracking inclination angle signal for multiple axes (X, Y, Z), some example methods may focus on an axis which is closest to being generally perpendicular to the gravity vector.

[0239] In some examples, in addition to applying filtering (at 714) as described above to produce a respective separate inclination angle signal (721 X, 721 Y, 721 Z) for each corresponding acceleration signal (e.g., X-axis, Y-axis, Z-axis), the filtering may further comprise subtracting (e.g., filtering, excluding) noise from the signal to increase the signal-to-noise ratio for the respiratory features of interest. In some examples, such noise filtering may be implemented as described later in association with noise model parameter 870 in FIG. 18E. It will be understood that in some examples, such noise filtering may be applied in other ways and/or at other times within the example method (and/or arrangement) in FIG. 18A.

[0240] As further shown at 740 in FIG. 18A, method 700 comprises performing a feature extraction on a signal-by-signal basis (741 X, 741 Y, 741 Z) to identify within each inclination angle signal (721 X, 721 Y, 721 Z) features indicative of respiration (and/or other features pertinent to respiratory detection, patient health, etc.), such as the expiratory phase midpoints (e.g., zero crossings TE.MID) and/or the inspiratory phase midpoints (e.g., zero crossings TI.MID) as described with reference to FIGS. 5A and 5B. As shown at 750, in some examples the method identifies at least respiratory phase information including (but not limited to) the features of an inspiratory phase 752, an expiratory active phase 754, and an expiratory pause phase 756 (e.g., as described with reference to FIGS. 4, 5A, and 5B). It will be understood that each feature (e.g., phase 752, 754, 756) may comprise a start (i.e., onset), an end (e.g., offset), duration, magnitude, and/or both a “start and end” of each respective feature. In some instances, a particular feature may sometimes be referred to as a fiducial or similar term, such as a start of a phase (e.g., inspiration) comprising a fiducial.

[0241] As shown at 730, a confidence factor may be applied to each of the feature extraction elements (741 X, 741 Y, 741 Z), such as an X-axis confidence factor 731 X, Y-axis confidence factor 731 Y, and Z-axis confidence factor 731 Z. At least some aspects of applying a confidence factor are described later in association with at least FIG. 18B. In some examples, the feature extraction elements (741 X, 741 Y, 741 Z) having the highest confidence factor are used at 745.

[0242] In some examples, upon performing feature extraction (741X, 741Y, 741Z) of respiratory phase information to each inclination angle signal, the resulting extracted feature signals are combined (e.g., fused together) at 745 to produce (i.e., determine) a composite sensed respiratory signal including respiratory phase information (750) including inspiratory phase 752, expiratory active phase 754, and expiratory pause phase 754. In some examples, the different extracted feature signals may be combined (e.g., fused) as an average of the respective features, a median of the respective features, or weighting (linear or non-linear) according to a confidence factor (e.g., 731X, 731 Y, 731Z). At least some aspects of the confidence factor(s) are described later in association with at least FIG. 18B. In some such examples, the composite sensed respiratory signal may correspond to the virtual vector as previously described in association with at least FIG. 8 and throughout various examples of the present disclosure.

[0243] As further shown in FIG. 18A, from the determined respiratory phases (752, 754, 754), additional respiratory parameters 760 may be determined. For example, an (overall) expiratory phase may comprise a sum or combination of the expiratory active phase (754) and the expiratory pause phase (756). In addition, a respiratory period may be determined from a sum of duration of the inspiratory phase 752 and a duration of the (overall) expiratory phase, including both the active and pause phases 754, 756. Meanwhile, the respiratory rate (RR) may be computed as 1 /respiratory period. Additional parameters may comprise a computed l/E ratio, such as by dividing a duration of the inspiratory to expiratory half cycle (e.g., inspiration half cycle 154B of FIGS. 5A and 5B) by the duration of an expiratory to inspiratory half cycle (e.g., expiration half cycle 154C of FIGS. 5A and 5B).

[0244] In some examples, assuming a given body position or posture and excluding translational motion along the axes (e.g., X, Y, Z), some additional parameters may be determined from the extracted features (including respiratory phase information at 750) with such additional parameters comprising: an approximation of a tidal volume as being proportional to acceleration; an approximation of respiratory flow as being proportional to a derivative of the acceleration signal with respect to time; and/or an approximation of minute ventilation as being proportional to a result of a multiplication of the computed volume and the computed respiratory rate (described above).

[0245] In some examples, determinations relating to feature extraction (740 in FIG. 18A) may further comprise the following parameters. For instance, in some examples of feature extraction, a signal midpoint (e.g., zero crossing) may be determined as previously described with reference to FIGS. 5A and 5B. In some examples of feature extraction, a signal midpoint crossing may be determined, which comprises a sample at which the signal midpoint is crossed. In some examples, the signal midpoint crossing may involve hysteresis with a hysteresis threshold being determined by a fixed threshold, a fraction of recent “n” peak-to- peak values, a fraction of signal root-mean-square (RMS) value, and/or a dynamic threshold with linear decay or exponential decay.

[0246] With further reference to FIG. 18A, in some examples determining respiration information (via sensing acceleration signals to detect rotational movements of the ribcage during breathing), the example method 700 may utilize default respiratory phase values as shown at 790 instead of using the sensed acceleration signals 710. For instance, in cases in which the sensed acceleration signal quality is poor (i.e., inadequate), the current respiratory phases of the patient may not be known from the current sensed acceleration signals or recent sensed acceleration signals. In some examples, the default respiratory phase values (790) are assigned a confidence level or factor 791 , which may have a low value to ensure that extracted features (741 X, 741 Y, 741 Z) are used when the sensed acceleration signal quality is adequate. Accordingly, when the sensed acceleration signal is of sufficient quality as determined by the signal-to-noise ratio of the signal, then method 700 may ignore the default respiratory phase values at 790. The signal-to-noise ratio may be determined by a comparison with a typical signal morphology, a comparison with a typical signal frequency content, or by other means.

[0247] With further reference to the default respiratory phase values portion 790 in FIG. 18A, in some examples the default respiratory phase values (790) may be determined using at least one of the following: (1 ) mean respiratory phase time values of the overall human population; (2) the patient’s historical or recent mean/median respiratory phase and/or phase time values; and (3) intentionally applying a longer respiratory rate or a shorter respiratory rate to decrease the chance that an appreciable number of consecutive stimulation “off” times may align with inspiration.

[0248] Accordingly, via the default respiratory phase values, some example methods may comprise substituting, upon the sensor obtaining an inadequate signal, stored respiratory information comprising historical respiration information for at least one of: the patient’s respiratory cycle information; and multiple-patient respiratory cycle information. In some such examples, the patient’s respiratory cycle information comprises a respiratory period, and an example method comprises: creating a modified respiratory period by adding a random time value to the respiratory period of the patient’s respiratory cycle information; and implementing the substituting of the stored respiratory information using the modified respiratory period. In some examples, the random time value may comprise about 0 to about 1 second. In some examples, the random time value may comprise other time periods. In some examples, adding the random time value may cause a result similar to that noted above (in regard to the default respiratory phase values) by which the example method may intentionally apply a longer respiratory rate or a shorter respiratory rate to decrease the chance that an appreciable number of consecutive stimulation “off” times may align with inspiration.

[0249] In some examples, the method may comprise substituting, upon the sensor obtaining an inadequate signal, stored respiratory information comprising respiratory cycle information including at least one of: a first respiratory rate substantially faster than the patient’s average respiratory rate; and a second respiratory rate substantially slower than the patient’s average respiratory rate. In some such examples, the terms substantially faster and/or substantially slower may correspond to a difference on the order of 5 percent difference, 10 percent difference, and the like.

[0250] FIG. 18B is a block diagram schematically representing an example confidence factor portion 800, which may be employed at 730 in example method 700 and/or as part of (or via) control portion 900 in FIG. 19A. It will be understood that all or just some of the factors (e.g., different combinations or a single factor) in confidence factor portion 800 may be applied at 730 in method 700 in FIG. 18A. In some examples, a confidence factor may be implemented as an estimated probability of correctness.

[0251] As shown in FIG. 18B, in some examples confidence factor portion 800 comprises a first factor portion 810 comprising a signal-to-noise ratio parameter 812, a threshold parameter 814, and a recent history parameter 816. Accordingly, in some examples, via signal-to-noise ratio information (parameter 812), a confidence level may be determined for each extracted feature (at 740 in FIG. 18A) and/or for each inclination angle signal (at 720 in FIG. 18A). In some examples, at 814 method 700 comprises the confidence comprising an amount by which a value (e.g., of a feature, of the inclination signal, etc.) exceeds a threshold. Stated differently, if a value of the inclination signal such as for a particular axis (e.g., Y-axis in FIG. 3) exceeds a threshold by a significant amount, then the method can apply a high value confidence factor to the Y-axis feature extraction (741 Y in FIG. 18A) such that determination of the respiratory phase information (750 in FIG. 18A) may depend primarily on the Y-axis inclination signal (721 Y in FIG. 18A) as compared to other axes (e.g., X or Z) inclination signals, if present. In some examples, the confidence factor may be applied per recent history parameter 816 according to a difference between a current value of an extracted feature and a mean value of “n” recent extracted features. In some examples, each of the confidence parameters in first factor portion 810 may be applied quantitatively according to a look-up table, multiplication factor (e.g., 1.5, 2x, etc.), and the like.

[0252] In some examples, confidence factor portion 800 may comprise a second factor portion 820 by which confidence in a value of a particular extracted feature (741 X, 741 Y, 741 Z) may be increased or decreased based on posture (822) at the time of sensing, heart rate (824), and/or sleep stage (826). As further shown in third factor portion 830 of FIG. 18B, such confidence factors in second factor portion 820 may be weighted and/or calibrated according to particular patientbased factors, such as patient preferences (e.g., feedback) 832, clinician input 834, and/or other information such sleep study information 836. Further parameters which may comprise part of second confidence factor portion 820 may include sensed body temperature, time of day, etc.

[0253] In some examples, the various parameters, etc. of the respective first, second, and third portions of confidence factor portion 800 may be used together in different combinations and/or organized in different groupings (or no groupings) than shown in FIG. 18B.

[0254] FIG. 18C is a block diagram schematically representing an example feature extraction portion 850, which may comprise functions, settings, etc. which may act as part of the implementation of the feature extraction at 740 in method 700 of FIG. 18A. As shown via parameter 852 at 850 in FIG. 18C, in some examples a threshold factor may be applied by a user or clinician to adjust thresholds used in performing feature extraction of the inspiratory phase 752 (e.g., inhalation threshold), of the expiratory active phase 754 (e.g., exhalation threshold), and/or of the expiratory pause phase 756 (e.g., exhalation threshold). As shown via parameter 854 at 850 in FIG. 18C, in some examples a sensitivity factor may be applied by a user or clinician to adjust thresholds used in performing feature extraction of the inspiratory phase 752 (e.g., inhalation sensitivity), of the expiratory active phase 754 (e.g., exhalation sensitivity), and/or of the expiratory pause phase 756 (e.g., exhalation sensitivity).

[0255] In some examples, in determining the respiratory phase information (790) example method 700 also may comprise predicting an inspiratory phase (e.g., 752 in FIG. 18A), as shown at 860 in FIG. 18D. The prediction of the inspiratory phase may be used to increase a likelihood of implementing actions (e.g., start of stimulation, etc.) which are to be synchronized with a start of the inspiratory phase 752. Stated differently, predicting the inspiratory phase 752 as at 860 in FIG. 18D may decrease a chance that detection of a start of the inspiratory phase might be missed. Moreover, in some example methods and/or example devices, electrical stimulation of a nerve (e.g., hypoglossal nerve, infrahyoid muscle (IHM)- innervating nerve) may be initiated prior to a start of inspiration to ensure that the upper airway is open prior to the pressure applied on the upper airway once the actual inspiratory phase commences. In addition, starting electrical stimulation prior to the actual inspiratory phase also may provide some assurance in cases in which prediction of the inspiratory phase may be incorrect or may experience an insufficient signal-to-noise ratio. In some such examples, example methods and/or devices may initiate the stimulation a predetermined period of time prior to an onset of the inspiratory phase. In some examples, the predetermined period of time has a duration less than a duration of the expiratory pause according to an average duration of an expiratory pause phase, according to a duration of the preceding expiratory pause phase, etc. In some examples, the predetermined period of time may comprise an absolute amount of time (e.g., start 0.5 seconds) and in some examples, the predetermined period of time may comprise a relative amount of time, such as 10% of the preceding respiratory period. As mentioned in association with other examples regarding synchronization, in some examples the predetermined period of time may be about 200 milliseconds or 300 milliseconds.

[0256] In some examples, the inspiratory phase prediction function (860) in FIG. 18D may comprise predicting a start of the inspiratory phase via timing based on: (1 ) an expiratory active phase 754 of the most recent (e.g., immediately preceding) respiratory cycle; (2) an expiratory pause phase 756 of the most recent (e.g., immediately preceding) respiratory cycle; and/or (3) an inspiratory phase of one or more previous respiratory cycles and/or the respiratory rate of one or more previous respiratory cycles. In some examples, in determining the timing (of the inspiratory phase and/or respiratory rate of previous respiratory cycles), the method may utilize a mean value, a median value, linear extrapolation, and/or non-linear extrapolation of the respective inspiratory phase or respiratory rate.

[0257] With further reference to inspiratory phase prediction 860 in FIG. 18D, in some examples, determining the timing (of the inspiratory phase and/or respiratory rate of previous respiratory cycles), the use of values from previous respiratory cycles may also enhance an accuracy of feature extraction (740 in FIG. 18A). For instance, accuracy of timing peak detection may be enhanced by using data before and after the peak. In another instance, using values from previous respiratory cycles may make an example method (of detecting respiration) less susceptible to a noisy signal during a particular respiratory cycle, patient limb movements, bed partner movements, etc.

[0258] In some examples, a method may increase accuracy of determining respiration from a sensed acceleration signal (of rotational movement at a portion of a chest wall) by removing noise from the sensed signal according to a noise model, which is shown in association with at least noise model parameter 870 in FIG. 18E.

[0259] In some such examples, the method comprises constructing the noise model from identifying characteristics (e.g., signal morphology, frequency content, etc.) within the sensed signal which are caused by and/or associated with conditions, phenomenon, etc. other than respiration-related behavior of the patient (and/or cardiac-related behavior, etc.) and which are considered noise relative to the signal of interest regarding patient respiration. In just one example, one source of noise (which may form at least part of a noise model) may comprise movement, behavior, etc. from another person (i.e. , partner) sleeping in the same bed, which may be picked up by the sensed signal for the patient. In some instances, such motion may sometimes be referred to as non-patient-physiologic motion. Other sources of noise, which form at least part of a noise model, may comprise additional/other non-patient-physiologic motion, such as but not limited to motion of a vehicle in which the patient is present such as when the patient is traveling in a car, airplane, spaceship, etc. Other types of non-patient-physiologic motion which may be considered as noise (and which form at least part of a noise model) may comprise movement of a patient support surface, such as a hammock, swings, etc. Another type of noise, which may form at least part of the noise model, may comprise a physical position of the patient such as being in a very tall building in motion due to wind, a location experiencing vibration or movement such that the motion of the patient may affect the sensed acceleration signal and otherwise hinder accurate determination of respiration information per the type of rotational sensing in the examples of the present disclosure.

[0260] By constructing a noise model from these non-patient characteristics, and subtracting the noise model from the sensed acceleration signal of the patient, a more accurate sensed respiration signal may be determined. In some instances, the subtraction may be performed by filtering the noise and/or by excluding sensor element signals including such noise.

[0261] In some examples, such noise may be filtered or excluded from the sensed acceleration signals (of rotational movement of a respiratory body portion, such as a chest wall) without use of a formal noise model.

[0262] In some examples, at least some of the features and attributes of use of a noise model, which may increase a signal-to-noise ratio of the signal of interest (respiration information), may be implemented at least partially within or via filtering 714 in method 700 as shown in FIG. 18A. [0263] In some examples, the prediction of the inspiratory phase (e.g., 860 in FIG. 18D) also may be performed according to cross-referencing (e.g., similarity) the inspiratory phase of a previous respiratory cycle relative to stored reference morphology of the inspiratory phase.

[0264] FIG. 19A is a block diagram schematically representing an example control portion 900. In some examples, control portion 900 provides one example implementation of a control portion forming a part of, implementing, and/or generally managing sensors, sensing element, respiration determination elements, stimulation elements, power/control elements (e.g., pulse generator), elements, devices, user interfaces, instructions, information, engines, elements, functions, actions, and/or methods, as described throughout examples of the present disclosure in association with FIGS. 1A-18E and 19B-20.

[0265] In some examples, control portion 900 includes a controller 902 and a memory 910. In general terms, controller 902 of control portion 900 comprises at least one processor 904 and associated memories. The controller 902 is electrically couplable to, and in communication with, memory 910 to generate control signals to direct operation of at least some of the sensors, sensing elements, respiration determination elements, stimulation elements, power/control elements (e.g., pulse generators), devices, user interfaces, instructions, information, engines, elements, functions, actions, and/or methods, as described throughout examples of the present disclosure. In some examples, these generated control signals include, but are not limited to, employing instructions 911 and/or information 912 stored in memory 910 to at least determine respiration information of a patient. Such determination of respiration information may comprise part of directing and managing treatment of sleep disordered breathing such as obstructive sleep apnea, hypopnea, and/or central sleep apnea. In some instances, the controller 902 or control portion 900 may sometimes be referred to as being programmed to perform the above-identified actions, functions, etc. such that the controller 902, control portion 900 and any associated processors may sometimes be referred to as being a special purpose computer, control portion, controller, or processor. In some examples, at least some of the stored instructions 911 are implemented as, or may be referred to as, a care engine, a sensing engine, a respiration determination engine, a monitoring engine, and/or a treatment engine. In some examples, at least some of the stored instructions 911 and/or information 912 may form at least part of, and/or, may be referred to as a care engine, sensing engine, respiration determination engine, monitoring engine, and/or treatment engine.

[0266] In response to or based upon commands received via a user interface (e.g., user interface 940 in FIG. 20) and/or via machine readable instructions, controller 902 generates control signals as described above in accordance with at least some of the examples of the present disclosure. In some examples, controller 902 is embodied in a general purpose computing device while in some examples, controller 902 is incorporated into or associated with at least some of the sensors, sensing elements, respiration determination elements, stimulation elements, power/control elements (e.g., pulse generators), devices, user interfaces, instructions, information, engines, functions, actions, and/or methods, etc. as described throughout examples of the present disclosure.

[0267] For purposes of this application, in reference to the controller 902, the term “processor” shall mean a presently developed or future developed processor (or processing resources) that executes machine readable instructions contained in a memory. In some examples, execution of the machine readable instructions, such as those provided via memory 910 of control portion 900 cause the processor to perform the above-identified actions, such as operating controller 902 to implement the sensing, monitoring, determining respiration information, stimulation, treatment, etc. as generally described in (or consistent with) at least some examples of the present disclosure. The machine readable instructions may be loaded in a random access memory (RAM) for execution by the processor from their stored location in a read only memory (ROM), a mass storage device, or some other persistent storage (e.g., non-transitory tangible medium or nonvolatile tangible medium), as represented by memory 910. In some examples, the machine readable instructions may comprise a sequence of instructions, a processor-executable machine learning model, or the like. In some examples, memory 910 comprises a computer readable tangible medium providing nonvolatile storage of the machine readable instructions executable by a process of controller 902. In some examples, the computer readable tangible medium may sometimes be referred to as, and/or comprise at least a portion of, a computer program product. In some examples, hard wired circuitry may be used in place of or in combination with machine readable instructions to implement the functions described. For example, controller 902 may be embodied as part of at least one application-specific integrated circuit (ASIC), at least one field- programmable gate array (FPGA), and/or the like. In at least some examples, the controller 902 is not limited to any specific combination of hardware circuitry and machine readable instructions, nor limited to any particular source for the machine readable instructions executed by the controller 902.

[0268] In some examples, control portion 900 may be entirely implemented within or by a stand-alone device.

[0269] In some examples, the control portion 900 may be partially implemented in one of the sensors, sensing elements, respiration determination elements, monitoring devices, stimulation devices, apnea treatment devices (or portions thereof), etc. and partially implemented in a computing resource separate from, and independent of, the apnea treatment devices (or portions thereof) but in communication with the apnea treatment devices (or portions thereof). For instance, in some examples control portion 900 may be implemented via a server accessible via the cloud and/or other network pathways. In some examples, the control portion 900 may be distributed or apportioned among multiple devices or resources such as among a server, an apnea treatment device (or portion thereof), and/or a user interface.

[0270] In some examples, control portion 900 includes, and/or is in communication with, a user interface 940 as shown in FIG. 20.

[0271] Figure 19B is a diagram schematically illustrating at least some example arrangements of a control portion 920 by which the control portion 900 (FIG. 19A) can be implemented, according to one example of the present disclosure. In some examples, control portion 920 is entirely implemented within or by an IPG assembly 925, which has at least some of substantially the same features and attributes as a pulse generator (e.g., power/control element) as previously described throughout the present disclosure. In some examples, control portion 920 is entirely implemented within or by a remote control 930 (e.g., a programmer) external to the patient’s body, such as a patient control 932 and/or a physician control 934. In some examples, the control portion 900 is partially implemented in the IPG assembly 925 and partially implemented in the remote control 930 (at least one of patient control 932 and physician control 934).

[0272] FIG. 20 is a block diagram schematically representing user interface 940, according to one example of the present disclosure. In some examples, user interface 940 forms part or and/or is accessible via a device external to the patient and by which the therapy system may be at least partially controlled and/or monitored. The external device which hosts user interface 940 may be a patient remote (e.g., 932 in FIG. 19B), a physician remote (e.g., 934 in FIG. 19B) and/or a clinician portal. In some examples, user interface 940 comprises a user interface or other display that provides for the simultaneous display, activation, and/or operation of at least some of the sensors, sensing elements, respiration determination elements, stimulation elements, power/control elements (e.g., pulse generators), devices, user interfaces, instructions, information, engines, functions, actions, and/or methods, etc., as described in association with FIGS. 1A-20. In some examples, at least some portions or aspects of the user interface 940 are provided via a graphical user interface (GUI), and may comprise a display 944 and input 942.

[0273] Although specific examples have been illustrated and described herein, a variety of alternate and/or equivalent implementations may be substituted for the specific examples shown and described without departing from the scope of the present disclosure. This application is intended to cover any adaptations or variations of the specific examples discussed herein.

Claims

1. A device comprising: a sensor securable to a patient to provide a sensor signal, the sensor to sense changes in acceleration indicative of respiration of the patient; a filter to detect zero crossings of the sensor signal; an electrode to deliver electrical stimulation to the patient; and a control portion to apply the electrical stimulation based on the detected zero crossings.

2. The device of claim 1 , wherein the control portion is to ignore a currently detected zero crossing in response to the currently detected zero crossing being within a predetermined period of a previously detected zero crossing.

3. The device of claim 1 , wherein the control portion is to identify each expiratory phase and/or each inspiratory phase of the sensor signal.

4. The device of claim 1 , wherein the control portion is to determine a confidence factor associated with the sensor signal.

5. The device of claim 3, wherein the control portion is to determine a confidence factor associated with each identified expiratory phase and/or each identified inspiratory phase.

6. The device of claim 1 , wherein the control portion is to determine a polarity of the sensor signal.

7. The device of claim 6, wherein the control portion identifies a positive half cycle of the sensor signal as an inspiration half cycle and/or identifies a negative half cycle of the sensor signal as an expiration half cycle of the sensor signal in response to determining the polarity of the sensor signal is positive; and/or wherein the control portion identifies a positive half cycle of the sensor signal as an expiration half cycle and/or identifies a negative half cycle of the sensor signal as an inspiration half cycle of the sensor signal in response to determining the polarity of the sensor signal is negative.

8. The device of claim 6, wherein the control portion is to determine the polarity of the sensor signal by: determining a median of samples of the sensor signal between a current zero crossing and at least two previous zero crossings; identifying the polarity of the sensor signal as positive in response to the median being less than 0; and identifying the polarity of the sensor signal as negative in response to the median being greater than 0.

9. The device of claim 6, wherein the control portion is to determine the polarity of the sensor signal by: determining a number of samples in each of a positive half cycle and a negative half cycle of the sensor signal; identifying the positive half cycle as including a median of samples of the sensor signal between a current zero crossing and at least two previous zero crossings in response to the number of samples of the positive half cycle exceeding the number of samples of the negative half cycle; identifying the negative half cycle as including the median of data points of the sensor signal between the current zero crossing and the at least two previous zero crossings in response to the number of samples of the negative half cycle exceeding the number of samples of the positive half cycle; in response to identifying the positive half cycle as including the median of the samples of the sensor signal, determining the median of the samples of the sensor signal within the positive half cycle; in response to identifying the negative half cycle as including the median of the samples of the sensor signal, determining the median of the samples of the sensor signal within the negative half cycle; identifying the polarity of the sensor signal as positive in response to the median being less than 0; and identifying the polarity of the sensor signal as negative in response to the median being greater than 0.

10. The device of claim 6, wherein the control portion is to determine the polarity of the sensor signal by: determining a number of samples in each of a positive half cycle and a negative half cycle of the sensor signal; identifying the polarity of the sensor signal as positive in response to the negative half cycle including more samples than the positive half cycle; and identifying the polarity of the sensor signal as negative in response to the positive half cycle including more samples than the negative half cycle.

11 . The device of claim 10, wherein the control portion is to: increment a counter each time a respective positive half cycle includes more samples than a respective negative half cycle; and decrement the counter each time a respective negative half cycle includes more samples than a respective positive half cycle.

12. The device of claim 11 , wherein the control portion is to determine a confidence factor based on a current value of the counter.

13. The device of claim 6, wherein the control portion is to: demean low passed samples of the last two half cycles of the sensor signal prior to determining the polarity of the sensor signal such that a high pass filter is excluded.

14. The device of claim 13, wherein the control portion is to: determine a median of data points of the demeaned low passed samples of the last two half cycles; identify the polarity of the sensor signal as positive in response to the median being less than 0; and identify the polarity of the sensor signal as negative in response to the median being greater than 0.

15. The device of claim 6, wherein the control portion determines the polarity of the sensor signal by: determining a median of data points of a first axis of the sensor signal between a current zero crossing and at least two previous zero crossings; identifying the polarity of the first axis of the sensor signal as positive in response to the median being less than 0; identifying the polarity of the first axis of the sensor signal as negative in response to the median being greater than 0; determining a median of data points of a second axis of the sensor signal between a current zero crossing and at least two previous zero crossings; identifying the polarity of the second axis of the sensor signal as positive in response to the median being less than 0; identifying the polarity of the second axis of the sensor signal as negative in response to the median being greater than 0; comparing the identified polarity of the first axis of the sensor signal to the identified polarity of the second axis of the sensor signal; and using the first axis of the sensor signal and/or the second axis of the sensor signal for further calculations in response to the identified polarity of the first axis of the sensor signal matching the identified polarity of the second axis of the sensor signal.

16. The device of claim 6, wherein the control portion is to determ ine a confidence factor associated with the identified polarity of the sensor signal using a statistical test.

17. The device of claim 1 , wherein the control portion is to identify inspiration half cycles and expiration half cycles of the sensor signal.

18. The device of claim 17, wherein the control portion identifies inspiration half cycles and expiration half cycles of the sensor signal by: determining a number of samples in each of a positive half cycle and a negative half cycle of the sensor signal; identifying the positive half cycle as an inspiration half cycle and/or the negative half cycle as an expiration half cycle in response to the number of the samples in the positive half cycle being less than the number of samples in the negative half cycle; and identifying the negative half cycle as an inspiration half cycle and/or the positive half cycle as an expiration half cycle in response to the number of the samples in the positive half cycle being greater than the number of samples in the negative half cycle.

19. The device of claim 1 , wherein the filter comprises a bandpass filter.

20. The device of claim 19, wherein a lower corner frequency of the bandpass filter is within a range between 0.01 Hz and 1 Hz, and an upper corner frequency of the bandpass filter is within a range between 0.25 Hz and 10 Hz.

21. The device of claim 20, wherein the corner frequencies are adjusted based on frequency domain analysis of the sensor signal to remove noise from the sensor signal.

22. The device of claim 21 , wherein the frequency domain analysis comprises power spectral density (PSD) analysis.

23. The device of claim 20, wherein the control portion is to determine a respiration period based on the detected zero crossings, and wherein the corner frequencies are adjusted based on the respiration period.

24. The device of claim 1 , further comprising: a decimation filter to reduce a sampling rate of the sensor signal.

25. The device of claim 1 , further comprising: a low pass decimation filter to reduce a sampling rate of the sensor signal and increase an amplitude resolution of the sensor signal.

26. The device of claim 1 , wherein the control portion is to determine an average respiration rate based on the detected zero crossings.

27. The device of claim 1 , wherein the control portion is to identify each positive half cycle and each negative half cycle of the sensor signal based on the detected zero crossings.

28. The device of claim 1 , wherein the control portion is to determine a confidence factor associated with the sensor signal, and wherein the control portion is to apply the electrical stimulation in an open loop mode in response to the confidence factor being less than a threshold value and in a closed loop mode in response to the confidence factor being greater than the threshold value.

29. The device of claim 1 , wherein the control portion is to identify each expiratory phase and/or each inspiratory phase of the sensor signal, and wherein the control portion is to determine a current respiration period based on the detected zero crossings.

30. The device of claim 1 , wherein the control portion is to identify each expiratory phase and/or each inspiratory phase of the sensor signal, and wherein the control portion is to determine whether a current zero crossing is a midpoint of a slope of an inspiratory phase (TI.MID) or a midpoint of a slope of an expiratory phase (TE,MID).

31 . The device of claim 30, wherein the control portion is to determine a current respiration period based on a current midpoint of a slope of an inspiratory phase and a previous midpoint of a slope of an inspiratory phase

32. The device of claim 30, wherein the control portion is to determine a current respiration period based on a current midpoint of a slope of an expiratory phase and a previous midpoint of a slope of an expiratory phase.

33. The device of claim 30, wherein the control portion is to determine a current respiration period based on a first period between a current midpoint of a slope of an inspiratory phase and a current midpoint of a slope of an expiratory phase plus a second period between the current midpoint of the slope of the expiratory phase and a previous midpoint of a slope of an inspiratory phase.

34. The device of claim 30, wherein the control portion is to determine a current respiration period based on a first period between a current midpoint of a slope of an expiratory phase and a current midpoint of a slope of an inspiratory phase plus a second period between the current midpoint of the slope of the inspiratory phase and a previous midpoint of a slope of an expiratory phase.

35. The device of claim 29, wherein the control portion is to determine an average respiration period for a predetermined number of respiration periods.

36. The device of claim 35, wherein the control portion is to exclude, from the determination of the average respiration period, respiration periods outside a predetermined range of threshold values.

37. The device of claim 29, wherein the control portion is to determine a median respiration period over a predetermined number of respiration periods.

38. The device of claim 29, wherein the control portion is to: determine a standard deviation over a predetermined number of respiration periods; and in response to the standard deviation exceeding a threshold value, resetting a confidence factor to a predetermined value.

39. The device of claim 28, wherein the sensor is to sense a posture of the patient.

40. The device of claim 39, wherein the control portion is to: detect a change in the posture of the patient; and in response to detecting the change in the posture of the patient, reset the filter, and reset the confidence factor to a predetermined value less than the threshold value.

41 . The device of claim 39, wherein the control portion is to: detect a change in the posture of the patient; and in response to detecting the change in the posture of the patient, reset an average respiration period and an identified polarity of the sensor signal.

42. The device of claim 39, wherein the control portion is to detect the change in the posture of the patient in response to a maximum value of sensor signal samples minus a minimum value of the sensor signal samples exceeding a threshold value.

43. The device of claim 30, wherein the control portion, with the current zero crossing being a midpoint of a slope of the inspiratory phase, is to: schedule a start of electrical stimulation based on the current midpoint of the slope of the inspiratory phase; and schedule a stop of electrical stimulation based on the current midpoint of the slope of the inspiratory phase.

44. The device of claim 43, wherein the control portion, with the current zero crossing being a midpoint of a slope of the inspiratory phase, is to start electrical stimulation immediately in response to electrical stimulation not already being started.

45. The device of claim 43, wherein the control portion is to schedule the stop of electrical stimulation one half average respiration period in the future.

46. The device of claim 43, wherein the control portion is to schedule the stop of electrical stimulation at the end of the current inspiration half cycle.

47. The device of claim 43, wherein the control portion is to schedule the stop of electrical stimulation 1.5 times the average respiration period in the future.

48. The device of claim 43, wherein the control portion is to schedule the stop of electrical stimulation within a range between 0.6 and 2 times the average respiration period in the future.

49. The device of claim 30, wherein the control portion, with the current zero crossing being a midpoint of a slope of the inspiratory phase, is to schedule a stop of electrical stimulation based on the current midpoint of the slope of the inspiratory phase.

50. The device of claim 49, wherein the control portion, with the current zero crossing being a midpoint of a slope of the expiratory phase, is to schedule a start of electrical stimulation based on the current midpoint of the slope of the expiratory phase.

51 . The device of claim 50, wherein the control portion, with the current zero crossing being a midpoint of a slope of the inspiratory phase, is to start electrical stimulation immediately in response to electrical stimulation not already being started.

52. The device of claim 49, wherein the control portion is to schedule the stop of electrical stimulation one half average respiration period in the future.

53. The device of claim 49, wherein the control portion is to schedule the stop of electrical stimulation at the end of the current inspiration half cycle.

54. The device of claim 49, wherein the control portion is to schedule the stop of electrical stimulation 1.5 times the average respiration period in the future.

55. The device of claim 49, wherein the control portion is to schedule the stop of electrical stimulation within a range between 0.6 and 2 times the average respiration period in the future.

56. The device of claim 50, wherein the control portion is to schedule the start of electrical stimulation one average respiration period in the future.

57. The device of claim 50, wherein the control portion is to schedule the start of electrical stimulation at the beginning of the next inspiration half cycle.

58. The device of claim 50, wherein the control portion is to schedule the start of electrical stimulation within a selectable range after a previous electrical stimulation stops.

59. The device of claim 30, wherein the control portion, with the current zero crossing being a midpoint of a slope of an inspiratory phase, is to: schedule a future start of electrical stimulation based on the current midpoint of the slope of the inspiratory phase; and schedule a future stop of electrical stimulation based on the current midpoint of the slope of the inspiratory phase.

60. The device of claim 59, wherein the control portion is to schedule the future start of electrical stimulation one average respiration period in the future.

61 . The device of claim 59, wherein the control portion is to schedule the future start of electrical stimulation at the beginning of the next inspiration half cycle.

62. The device of claim 59, wherein the control portion is to schedule the future start of electrical stimulation within a selectable range after a previous electrical stimulation stops.

63. The device of claim 1 , wherein the control portion is to determine a current respiration period based on the detected zero crossings, and wherein the control portion is to switch from a closed loop mode to an open loop mode in response to a standard deviation of a predetermined number of respiration periods exceeding a threshold value.

64. The device of claim 1 , wherein the control portion is to identify each expiratory phase and/or each inspiratory phase of the sensor signal, wherein the control portion is to determine whether a current zero crossing is a midpoint of a slope of an inspiratory phase (TI.MID) or a midpoint of a slope of an expiratory phase (TE.MID), and wherein the control portion, with the current zero crossing being a midpoint of a slope of an expiratory phase, is to: schedule a start of electrical stimulation based on the current midpoint of the slope of the expiratory phase; and stop electrical stimulation or schedule a stop of electrical stimulation based on the current midpoint of the slope of the expiratory phase.

65. The device of claim 64, wherein the control portion is to schedule the start of electrical stimulation one half average respiration period in the future.

66. The device of claim 64, wherein the control portion is to schedule the start of electrical stimulation at the beginning of the next inspiration half cycle.

67. The device of claim 64, wherein the control portion is to schedule the start of electrical stimulation within a range between -2 and 2 times the average respiration period in the future relative to a midpoint of a slope of an expiratory phase.

68. The device of claim 64, wherein the control portion is to schedule the stop of electrical stimulation one average respiration period in the future.

69. The device of claim 64, wherein the control portion is to schedule the stop of electrical stimulation at the start of the next expiration half cycle.

70. The device of claim 64, wherein the control portion is to replace a previously scheduled stop of electrical stimulation in response to a midpoint of a slope of an inspiratory phase occurring.

71 . The device of claim 1 , wherein the control portion comprises a processor.

72. The device of claim 1 , wherein the control portion comprises an Application Specific Integrated Circuit (ASIC).

73. The device of claim 1 , wherein the sensor comprises an at least two axis accelerometer.

74. The device of claim 73, wherein the at least two axis accelerometer provides at least two respective sensor signals, and wherein the control portion is to: determine a confidence factor associated with each of the at least two respective sensor signals; identify which respective sensor signal exhibits the highest confidence factor; and detect zero crossing of the sensor signal exhibiting the highest confidence factor.

75. The device of claim 73, wherein the at least two axis accelerometer provides at least two respective sensor signals, and wherein the control portion is to detect zero crossings of a predetermined sensor signal.

76. The device of claim 73, wherein the at least two axis accelerometer provides at least two respective sensor signals, and wherein the control portion is to detect zero crossings of the sensor signal selected by a clinician.

77. The device of claim 74, wherein the control portion is to update the confidence factor associated with each of the at least two respective sensor signals in response to each detected zero crossing.

78. The device of claim 74, wherein the control portion is to apply the electrical stimulation in an open loop mode or a closed loop mode, and wherein the open loop mode is assigned a predetermined open loop mode confidence factor.

79. The device of claim 78, wherein the control portion is to: apply the electrical stimulation in the closed loop mode in response to the respective sensor signal exhibiting the highest confidence factor comprising a confidence factor greater than the predetermined open loop mode confidence factor; and apply the electrical stimulation in the open loop mode in response to the respective sensor signal exhibiting the highest confidence factor comprising a confidence factor less than the predetermined open loop mode confidence factor.

80. The device of claim 1 , wherein an electrical stimulation duty cycle is less than or equal to 75%.

81. The device of claim 80, wherein the control portion is to adjust stimulation start times and/or stimulation stop times based on the electrical stimulation duty cycle.

82. The device of claim 80, wherein the electrical stimulation duty cycle is greater than 25%.

83. A method comprising: sensing changes in acceleration indicative of respiration of the patient to provide a sensor signal; filtering the sensor signal to detect zero crossings of the sensor signal; and applying electrical stimulation to the patient based on the detected zero crossings.

84. A method of stimulating an upper airway patency-related nerve comprising: chronically implanting a stimulation element in or near an upper airway; sensing a signal corresponding to respiration; detecting one or more zero crossings of the sensor signal; calculating a future onset of an inspiratory phase from the detected one or more zero crossings; and delivering stimulation to the upper airway patency-related nerve via the stimulation element, wherein the stimulation starts a predetermined amount of time prior to the calculated onset of the inspiratory phase.

85. A method of treating obstructive sleep apnea, the method comprising: chronically implanting an electrode on a nerve innervating an upper airway muscle; sensing a measure of respiration; analyzing the measure of respiration to identify midpoints of expiration; calculating a respiratory period from the midpoints of expiration; predicting the midpoint of a future expiratory phase; and beginning stimulation of the nerve a fraction of the calculated respiratory period before the midpoint of the future expiratory phase, and continuing stimulation of the nerve during an entire inspiratory phase, wherein the method is performed without identifying an onset of the inspiratory phase.