CN103476328B - Staged alarming method for patient monitor - Google Patents

Abstract

A kind of system (202), it uses staged alarm schemes to generate patient alarm.Described system (202) includes one or more processor (220), and described processor is programmed to: receive Physiological Score and/or physiologic parameter value；Described Physiological Score and/or described physiologic parameter value are compared with multiple alarm levels；Fall in response to Physiological Score and/or physiologic parameter value in the non-inhibited district of described alarm levels, send alarm；And after sending described alarm, the first suppression phase for described non-inhibited alarm levels is set.

Description

Stepped alarm method for patient monitor

Technical Field

The present application relates generally to patient monitoring. It finds particular application in conjunction with reducing patient alarms and will be described with particular reference thereto. However, it should be understood that the present application also has particular application to other use scenarios and is not necessarily limited to the aforementioned applications.

Background

Patients often have an abnormal vital sign period before deterioration. As such, clinicians often employ predictive systems to assess the likelihood of deterioration. Such systems include anomaly scoring systems such as Early Warning Scoring (EWS) and modified EWS (mews) scoring systems. The abnormality scoring system unifies assessments of multiple physiological parameters, such as vital signs, into a unified unit system and merges individual assessments to determine patient risk, which can cause preventable adverse events, such as cardiac arrest or death.

For the sample EWS scoring system (fig. 1), the EWS is determined using a table based on several vital signs. When vital signs were normal, they were evaluated as scoring zero. As the degree of abnormality of each vital sign increases, more points are evaluated for the vital sign. The total score of all vital signs is an indicator of the degree of abnormality and if it exceeds a preselected threshold, a follow-up action is defined (e.g., a consultation of a clinician or the initiation of a so-called quick response team).

The clinician typically manually implements an anomaly scoring system. However, one challenge with manual scoring is that medical institutions such as hospitals have limited resources. Thus, patients, for example in a general ward, are not evaluated very frequently, usually once every 4 to 8 hours. In such a secondary critical care setting, the patient may not be noticed to deteriorate. Late findings of this exacerbation may lead to unnecessary complications, Intensive Care Unit (ICU), cardiac arrest, death, and the like.

To alleviate this, automated monitoring of patients is becoming more and more common. However, a major challenge of automatic monitoring is alarm fatigue. Alarm fatigue is the case: a condition in which the clinician becomes numb of clinical alerts due to a high probability of having no actually clinically meaningful alerts.

One way to reduce the alarm burden is to raise the alarm threshold, typically manually. However, other nurses in the same shift, and subsequent shifts, may not notice the high threshold and be confused by the illusion that the patient is healthy well. Furthermore, this reduces sensitivity and increases the likelihood that patient deterioration cannot be detected. Another approach is to set a suppression period after the alarm is issued so that similar alarms are not issued until a reset condition is met. In such a method, the reset condition is critical to reducing alarms.

A typical reset condition is the lapse of a predetermined suppression period from the triggering of an alarm. This is based on the following insight: any alarms subsequent to the first alarm may be based on similar physiological data and therefore provide no additional information to the clinician. If the clinician agrees with the alarm, he either has planned to take action to treat the patient, or he suspects the effectiveness of the alarm. In either case, another alarm would not be necessary. It is therefore reasonable to suppress further alarms for a limited period of time.

One disadvantage of this reset condition is that if the patient's condition deteriorates during the alarm suppression period, no additional alarms will occur. Another disadvantage is that the predetermined amount of time is common to the general patient population. As such, the predetermined amount of time is not customized for any particular patient. Furthermore, the predetermined amount of time is not adapted to the dynamics of the individual.

Other challenges of automated monitoring systems stem from the predictive models typically employed by automated monitoring systems. Such predictive models are typically trained on population data of large databases, whereby decisions using such predictive models are based on general characteristics of large populations. Furthermore, differences between individuals and the general training population are not typically taken into account. Training in this manner may result in unnecessary reminders and/or the failure to generate reminders for patients with certain physiological specifications that differ from those of the general training population.

One solution is to adjust the predictive model based on knowledge of the patient's health, or baseline, physiological dynamics. However, baseline data is often not available in practice, especially in the ICU, where it is at no time possible to assume that the data being collected reflects the "normal" physiology of the patient.

Another solution employs direct feedback from the clinician regarding the effectiveness of issued alerts for learning. However, such an approach is not possible for systems that do not have this direct feedback learning. Furthermore, if an alarm is issued in response to a predicted event several hours in advance, immediate feedback from the clinician regarding the effectiveness of the alarm is of no concern.

The present application provides new and improved methods and systems which overcome the above-referenced problems and others.

Disclosure of Invention

According to one aspect, a system for generating a patient alert using a stepped alert scheme is provided. The system includes one or more processors programmed to: receiving a physiological score and/or a physiological parameter value; comparing the physiological score and/or the physiological parameter value to a plurality of alert levels; issuing an alert in response to the physiological score and/or the physiological parameter value falling within a non-suppressed alert level of the alert levels; and setting a first suppression period for the non-suppressed alarm level after issuing the alarm.

According to another aspect, a method of generating a patient alert using a stepped alert scheme is provided. A physiological score and/or a physiological parameter value is received. Comparing the physiological score and/or physiological parameter value to a plurality of alert levels. An alarm is issued in response to the physiological score and/or the physiological parameter falling within a non-suppressed alarm level of the alarm levels. Further, a first suppression period for the non-suppression alarm level is set after the alarm is issued.

According to another aspect, a system for resetting a suppressed alarm level is provided. The system includes one or more processors programmed to receive the physiological scores and/or physiological parameter values. Comparing the physiological score and/or the physiological parameter value to a plurality of alert levels. An alarm is issued in response to the physiological score and/or the physiological parameter value falling within a non-suppressed alarm level of the alarm levels. A suppression period for a non-suppression alarm level is set, and in response to setting the suppression period, the system waits a predetermined amount of time. Determining whether an intervention was administered during the predetermined amount of time. Resetting the suppressed alarm level in response to determining that no intervention is being administered and that the current physiological score and/or physiological parameter value is degraded by a predetermined amount as compared to the physiological score and/or physiological parameter value.

One advantage resides in increased sensitivity to abnormal patient conditions while reducing alarm burden.

Another advantage resides in sensitivity to absolute thresholds.

Another advantage resides in low alarm burden.

Another advantage resides in sensitivity to patient deterioration.

Another advantage resides in applicability to a single physiological parameter as well as multiple physiological parameters.

Another advantage resides in intuitive parameters that can be directly adjusted.

Another advantage resides in minimizing bedside modification to alarm thresholds.

Another advantage resides in adjusting to situations where patients have atypical conditions for the average population.

Still further advantages of the present invention will be appreciated to those of ordinary skill in the art upon reading and understand the following detailed description.

Drawings

The invention may take form in various components and arrangements of components, and in various steps and arrangements of steps. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention.

The table of fig. 1 illustrates the EWS scoring system.

Fig. 2 is a graph illustrating a method of generating a patient alert using measurements for vital signs according to aspects of the present disclosure.

Fig. 3 is a graph illustrating a method of generating a patient alert using an abnormality score according to aspects of the present disclosure.

Fig. 4 is a block diagram of a method of generating a patient alert according to aspects of the present disclosure.

Fig. 5 is a graph illustrating a discrete and piecewise linearized EWS scoring system.

Figure 6 is a graph illustrating a signed EWS scoring system graph (discrete and piecewise linearized).

FIG. 7 is a graph illustrating a sample relationship between alarm thresholds and the number of alarms per patient per day for different periods of suppression.

Fig. 8 is a graph illustrating a sample relationship between changes in EWS versus daily alarms for each patient.

Fig. 9 is a graph illustrating a sample relationship between time between alarms per patient per day and alarms for different Δ EWS.

Fig. 10 is a graphical depiction illustrating an unstable patient condition.

FIG. 11 is a graphical illustration of a method of detecting an unstable patient condition.

Fig. 12 is a graph illustrating a sample relationship between the width of the suppression zone () and the number of alarms per patient per day.

Fig. 13 is a graphical illustration of the unstable patient condition of fig. 10 in EWS-space.

The graph of fig. 14 illustrates the change in vital signs of a sample patient from too high to too low.

Fig. 15 is a graph illustrating the number of alarms per hour per patient versus hours of the day.

Fig. 16 is a block diagram of a method of resetting an alarm level according to aspects of the present disclosure.

Fig. 17 is a graphical illustration of resetting an alarm level.

Fig. 18 is a block diagram of an IT infrastructure in accordance with aspects of the present disclosure.

Detailed Description

Referring to fig. 2 and 3, an illustrative example of a method 100 (see fig. 4) for generating a patient alarm using a stepped alarm scheme is provided. Suitably, patient monitors, such as wearable patient monitors, bedside patient monitors, and central patient monitors, perform the method 100. As discussed in detail later, by using a stepped alert scheme in conjunction with a long suppression period, fewer alerts are generated while still being sensitive to patient deterioration. Figure 2 uses vital sign measurements for a single vital sign (e.g., respiratory rate) to assess patient deterioration; while figure 3 uses an abnormality score (e.g., EWS) that can typically be calculated from vital signs to assess patient deterioration.

Vital sign measurements include measurements of vital signs such as: heart rate, temperature, blood oxygen saturation, level of consciousness, pain, urine output, etc. Anomaly scores such as EWS and MEWS combine vital sign measurements for multiple vital signs into a score that assesses the patient's risk of death. The anomaly scoring system provides non-linear weights to arrive at a "equally severe" scale for each vital sign. In this aspect, all vital signs that entered the abnormality score are assessed using the scale and summed to give the abnormality score. Typically, the vital signs are assumed to be independent when calculating the abnormality score. However, some combinations of vital signs are more abnormal than others, and thus the abnormality score may also include a score for a combination of vital signs. To improve the sensitivity of anomaly scores to vital signs, the vital signs may be given more weight in the determination of the anomaly scores. Additionally or alternatively, to improve sensitivity, the scoring region for the vital signs may be refined. It is contemplated that the anomaly scoring system may be customized for individual patients, medical wards, medical institutions, and the like. In certain embodiments, the clinician manually customizes the anomaly scoring system by using a user input device. In other embodiments, the abnormality scoring system is customized based on patient information from, for example, a patient information system.

In both examples of fig. 2 and 3, there is generally at 12: 30, crossing the initial threshold. In the case of fig. 2, the threshold is 28 breaths per minute, and in the case of fig. 3, the threshold is EWS is 2. Upon exceeding the initial threshold, an alarm is issued, a long suppression period, e.g. 8 hours, is applied for alarms of the same condition, and the threshold is raised. Thereafter, in both examples, approximately at 17: 00, crossing a second higher threshold (i.e., the situation worsens). In the case of fig. 2, the threshold is 30 breaths per minute, and in the case of fig. 3, the threshold is EWS is 3. Upon exceeding the second threshold, another alarm is sounded, a long suppression period, e.g. 8 hours, is applied for the new condition, and the threshold is raised.

Referring to fig. 4, a block diagram of a method 100 for generating a patient alarm using a stepped alarm scheme is provided. Physiological scores and/or physiological parameter values for one or more patients are received 102. The physiological score is an assessment of a physiological condition (e.g., hemodynamic stability or risk of death) of the patient based on at least one physiological parameter according to a physiological scoring system. The physiological parameter is a measurable or observable patient characteristic. Examples of physiological scores include abnormality scores, and examples of physiological parameters include vital signs.

Typically, the physiological score and/or the physiological parameter value is automatically received from a sensor associated with the patient via, for example, a wired or wireless communication network. However, in other embodiments, the physiological score and/or the physiological parameter value are manually received from a clinician via, for example, a user input device. Furthermore, the physiological score and/or physiological parameter value is typically received continuously. However, the physiological score and/or physiological parameter value may alternatively be received upon the occurrence of an event such as a timer event (e.g., a periodic timer), a patient event, a manual trigger event (e.g., a clinician pressing a button), etc. In certain embodiments, the physiological score is received indirectly from the physiological parameter value. In this regard, the physiological score is automatically calculated from the physiological parameter value. For example, EWS is calculated from vital sign measurements as discussed above.

Anomaly scoring systems typically derive integer values. However, this may lead to discrete jumps in anomaly scores, especially in cases when vital signs fluctuate around the boundary values. To mitigate this, the anomaly scoring system may be piecewise linearized. Referring to fig. 5, an example of an EWS scoring system for respiration rate piecewise linearization is illustrated. The solid line indicates the scoring system in piecewise linearized version and the dashed line indicates the scoring system in discrete version. In certain embodiments, a best fit scheme is applied for linearization, although other schemes are also contemplated.

Furthermore, while the individual scores assigned to vital signs in anomaly scoring systems are typically unsigned, it is contemplated that signed scores may be employed to distinguish between vital sign measurements for vital signs that are too low and too high. This distinction can be made, for example, by adding a positive ("+") and negative ("-") sign to the score for individual vital signs that are too high and too low, respectively. Referring to fig. 6, an example of signed anomaly scores for respiration rates is illustrated. As in fig. 5, the solid line indicates the piecewise linearized version of the scoring system and the dashed line indicates the discrete version of the scoring system.

In calculating a physiological score, situations may arise in which less than all of the physiological parameter values of the physiological parameter needed to calculate the physiological score are received. The physiological parameter values may be missing due to erroneous measurements and/or observations, or may result from differences in measurement and/or observation periodicity. For example, heart rate measured every minute and non-invasive blood pressure (NIBP) measured every 30 minutes. One solution to this situation is to store 104 the most recent physiological parameter value for each physiological parameter needed to calculate the physiological score, e.g., in memory. In this regard, the physiological parameter value is used prior to receiving a new physiological parameter value for the corresponding physiological parameter. Other solutions are from other physiological parameters (e.g., ECG and SpO)₂Both of which may supply heart rate data) or a modeled combination of physiological parameters to derive missing information.

Referring back to fig. 4, the physiological score and/or the physiological parameter value is compared 106 to a plurality of alarm levels. The alarm level may include one or more thresholds, ranges, and the like. Typically, the alarm level is determined by a clinician and/or defined by a medical facility (e.g., hospital) using the method 100. However, in some embodiments, the alarm level may be dynamically generated 108. It is contemplated that the alarm levels may be customized for individual patients, medical wards, medical institutions, and the like. In certain embodiments, the clinician manually customizes parameters of the physiological scoring system, such as thresholds of the abnormality scoring system, by using the user input device. In other embodiments, the physiological scoring system is automatically customized based on patient information from, for example, a patient information system.

Although no particular scheme is required to select the alarm level, the alarm level should be selected to minimize alarms while maximizing sensitivity to patient deterioration. Referring to fig. 7, an example of the relationship between the alarm threshold of the EWS scoring system and the average number of alarms per patient per day for different suppression periods is provided. When a high alarm threshold and/or a long suppression period is selected, the number of alarms is low. Conversely, when a low threshold is selected, the method 100 becomes more sensitive to the detection of abnormal vital signs. However, in combination with a long suppression period, no alarm will occur in case of further deterioration within the suppression period.

One way to dynamically generate 108 the alarm level is through the use of a delta and an initial alarm level. For example, an initial alarm level is defined for a threshold with an anomaly score of 3 and Δ is 0.5, then the initial alarm will sound at 3.0 and another alarm will sound at 3.5 even though still within the suppression period of the alarm level. After another alarm occurs, a new suppression period for a level of 3.5 is set, and an alarm level of 4 will be the next alarm level triggered when the patient further deteriorates. The delta can be customized to an individual patient, medical ward, medical facility, or the like. In certain embodiments, the clinician manually customizes the Δ by using a user input device. In other embodiments, the Δ is customized based on patient medical records from, for example, a patient information system. For more severe patients (e.g., patients with higher abnormality scores), a lower Δ may be selected to increase sensitivity to deterioration. After the suppression period, decreasing the threshold.

Referring to fig. 8, a graph showing an example of the relationship between Δ and the number of alarms per patient per day for the EWS scoring system. An important observation is that by using a, the method 100 is very sensitive to further deterioration of the patient's condition with minimal negative impact on the average alarm burden. Furthermore, setting Δ too high will rarely trigger a new alarm, while setting it too small may cause an alarm burst in a short period of time. One approach is to evaluate a reasonable value of Δ by plotting the time difference between alarms. Referring to fig. 9, an example of such a plot for an EWS scoring system is illustrated. Among these, a large peak at 480 minutes can be seen, which is a logical consequence of the chosen inhibition period of 480 minutes (8 hours). If a small value of Δ (0.5) is chosen, a large peak in the short time between alarms becomes visible, which is caused by the alarms continuing rapidly during the ascending trend of EWS. A reasonable delta value is approximately 1-2 based on visual perception of events generated at different settings.

Referring back to fig. 4, an alarm is generated in response to the physiological score and/or the physiological parameter value falling within the alarm levels 110 to a non-suppressed alarm level. The alarm suitably notifies the clinician to check the patient and, in certain embodiments, to check the severity of the alarm. Thereafter, a long suppression period, for example on the order of a few hours, is set 112 for the triggered alarm level. In certain embodiments, the suppression period also suppresses alarms at lower alarm levels. To ensure that the clinician does not miss an alarm, a suppression period is typically set only after the alarm is acknowledged, e.g., by the clinician. The length of the suppression period may be variable between alarm levels. In certain embodiments, a suppression period comparable to the duration of a nurse shift (e.g., 8 hours) is employed. Advantageously, this ensures that no alarm occurs for patient conditions that have not changed during the shift, while the new nurse in the next shift is again notified of the existing patient conditions. In other embodiments, a suppression period is employed before the next nurse shift. However, this has the disadvantage that a large number of alarms will sound at the beginning of each nurse shift. In other embodiments, a shorter suppression period is employed for higher abnormality scores or vital sign measurements.

In certain embodiments, a short suppression period for all alarms is set in response to the alarm 114 being issued. In other words, after the alarm is issued, a short suppression period is set that defers all alarms. Suitably, the short inhibition period is of the order of a few minutes, for example 5-10 minutes. Advantageously, this reduces the alarm burden without significantly (if at all) compromising the patient's health, as the typical response event for nurses in general wards is on the order of a few minutes.

The anomaly score has the advantage that it is an easily communicated value. For example, 'patient x has an EWS' of 6. However, a disadvantage is that it no longer reveals the contribution of individual vital signs. This results in no risk of detecting changes in the composition of the abnormal score, especially when there are some vital signs that begin to improve and other vital signs worsening. This is undesirable because a constant abnormality score may incorrectly assess a condition as stable despite fluctuations in vital signs. Thus, in certain embodiments, abnormally scored compositions are evaluated 116 for changes in composition, and in response to composition changes 118, an alarm and suppression period is triggered, as described above.

An example of changes in the composition of the anomaly score (EWS in this case) derived from Heart Rate (HR) and Respiratory Rate (RR) is shown in fig. 10. The gray zone is the period of invalid data and the cross (x) indicates that an alarm will occur. The horizontal dotted line is the boundary of the region that scores 'EWS point' as a different value. The EWS point is indicated with a large font. The first alarm occurs at 23 due to HR: around 00, indicated by crosses in both figures. However, the HR improved slightly, while the respiration rate varied from somewhat high to very high. The result is that the EWS remains unchanged. However, even so, there are unstable situations that justify a new alarm.

One method for evaluating 116 (fig. 4) the anomaly scores of the changes in composition according to this example is explained using sub-graphs a to C in fig. 11. As will be seen below, the signed anomaly score of fig. 6 is particularly applicable here. Panels a-C show the case for a two-dimensional anomaly score (i.e., EWS) based on, for example, Respiration Rate (RR) and Heart Rate (HR). The diagonal indicates a constant anomaly score. In addition, the prominence block area 119 represents a minimum alarm level (selected to be a level of 2 in this example) that is set to generate an alarm.

Referring to panel a, point 120 indicates a measurement score of 0.5 point (i.e., -0.5) for abnormally low breaths and 1.7 point (i.e., + 1.7) for abnormally high HRs. Thus, the total abnormality score was 2.2. After the alarm is raised, a higher alarm level is automatically selected, indicated by the highlighted box 122 in sub-graph B. This example assumes that Δ is 1.0. The protruding rectangular area 124 (also called a suppression zone) in sub-figure C indicates an additional limit set to prevent the previously mentioned problem of possible unnoticed changes in composition. The scoring vector 126 of the last alarm that occurred (i.e., the alarm of sub-graph a) forms the axis of the highlighted rectangular region 124. If the current scoring vector 128 deviates outside of the region, a new alarm will sound, even when the next absolute alarm level (3.0 in this example) has not yet occurred. The detection of this deviation is done by continuously checking the vector distance between the current scoring vector 128 and the scoring vector 126 of the previous alarm. If the distance is greater than a predetermined, e.g., user-set level, an alarm will sound and the highlighted rectangular area 124 is redefined. This calculation of the distance can be done using standard vector algebra. In case the method is extended to more vital signs, the algebra remains basically unchanged.

Referring to fig. 12, an example of the number of alarms as a function of distance is shown. The value for the initial EWS was 3, the value for Δ was 1, and the value for the suppression period was 480 minutes. It can be seen that if the distance is too high, the method 100 will be sensitive to changes in composition, and if the distance is too low, the number of alarms will increase as the width of the highlighted rectangular area 124 becomes similar to normal fluctuations in the composition of the scoring vector. Based on fig. 12 and also based on a visual inspection of some of the generated alarms, a reasonable value for the width of the highlighted rectangular area 124 is approximately 1-4.

The example of fig. 10 is also plotted in 'EWS space', as in fig. 13. The measurement results are indicated by interconnected dots, wherein the arrow 130 indicates the direction of time. The two alarms of fig. 10 are indicated in the figure by circles 132. The protruding rectangular region 134 (with 2) is a suppression region, similar to the protruding rectangular region 124 described above. New alerts occur due to large changes in EWS composition.

Another example of a change in composition of an abnormality score (EWS in this case) derived from Heart Rate (HR) and Respiratory Rate (RR) is shown in fig. 14. The gray zone is the period of invalid data, and the cross (x) indicates the time at which an alarm will occur. The horizontal dotted line is the boundary of the region scored as different values for 'EWS point'. The EWS point is indicated with a large font. The first alarm occurs at 12 due to an abnormally low breathing rate: around 15 hours, indicated by a cross (score 3EWS point). At 17: around 15 hours, the absolute EWS still scored less than 3 points, but the situation was unstable due to a rapid increase in respiration. If no technical measures are taken, no alarm is issued because: 1) the previous alarm was less than 8 hours ago; 2) EWS for breathing is still less than at 12: value scored at 15 hours; and 3) the perpendicular distance φ to the previous alarm vector is small.

The foregoing is schematically depicted in sub-diagram D of fig. 11. Sub-graph D shows the case of a two-dimensional anomaly score (i.e., EWS) based on, for example, Respiration Rate (RR) and Heart Rate (HR). The diagonal indicates a constant anomaly score. Further, highlighted box area 136 represents a level set to generate an alarm minimum alarm level (in this example, a level of 2 is selected), and a higher alarm level is indicated by highlighted box 138. Even a scoring vector 142 for an alarm preceding the current scoring vector 140 is shown.

One method for evaluating the anomaly scores for compositional changes according to this example includes calculating the inner product between the current score vector 140 and the score vector 142 of the previous alarm. For inner products <0 (rotation greater than 90 °), the suppression period should be turned off, allowing an alarm to be issued.

Referring to fig. 15, a low alarm burden resulting from the proposed method is shown. The initial alarm level was 3.0, Δ was 1.0 and the suppression period was 8 hours. This results in an average alarm rate of approximately 0.02 alarms per hour per patient on that day. For a 25 bed ward, this would be approximately one alarm every 2 hours.

The process discussed hereinafter resets the alarm level after a predetermined amount of time has elapsed. For example, when the physiological score and/or the physiological parameter value falls within an alarm level, an alarm is triggered, and a suppression period of a predetermined length is set to suppress the same alarm level from triggering more alarms until the suppression period ends. However, other methods of resetting are contemplated. Referring to fig. 16, a block diagram of an adaptive method 150 of resetting is provided.

The adaptive method 150 assumes familiarity with typical intervention measures taken by clinicians in response to alarms. For example, typical interventions taken in response to an alarm for hemodynamic stability include administration of fluids, vasopressors, concentrated red blood cells. Further, the adaptive method 150 assumes at least one clinical data source that prescribes an intervention taken by a clinician in response to an alert. For example, such a source may be a patient information system to which a clinician provides data regarding an intervention taken via a user input device. Even more, the adaptive method 150 assumes that the alert issued by the physiological score and/or the physiological parameter value reflects the stability of the patient in terms of the physiological condition (e.g., hemodynamic stability or nutritional stability). The adaptive method 150 may be employed for resetting when the foregoing is available.

Triggering an alarm and suppressing the alarm level when the physiological score and/or the physiological parameter value of the physiological scoring system and/or the physiological parameter falls into a non-suppressed alarm level. In response to suppressing the alarm level, the adaptive method 150 waits 152 for a predetermined amount of time, such as three hours. Typically, the physiological scoring system and/or physiological parameter is predictive, such that an alert generated therefrom is triggered before the patient deteriorates. The predetermined amount of time generally corresponds to the lead period and generally varies depending on the physiological condition of the physiological score and/or physiological parameter value. After the predetermined amount of time has elapsed, a determination 154 is made as to whether an intervention was given during the predetermined amount of time to resolve the alert based on the received clinical data. This may be based on the physiological condition for the physiological score and/or physiological parameter value, the actual situation in which the intervention occurred, or typical past intervention measures.

If no intervention is given, the adaptive method 150 waits 156 until the current physiological score and/or physiological parameter value of the physiological scoring system and/or physiological parameter has been degraded 156 by a threshold amount compared to the physiological score and/or physiological parameter value at the time of the alert. The threshold amount may be fixed or variable, such as a sum of distances to previous physiological scores and/or physiological parameter values of the physiological scoring system and/or physiological parameters. The alarm level is reset 158 once the current physiological score and/or physiological parameter value has deteriorated. By resetting in this manner, the lack of intervention by the clinician after a significant amount of time has elapsed is interpreted as an indication that the condition of the patient at the time of the first alarm is acceptable, normal, or stable for that particular patient. Thus, even if the current physiological score and/or physiological parameter value is abnormal, or unstable, to the population criteria, the adaptive method learns to be normal for that patient.

If an intervention is given, this is considered confirmation by the clinician and no further alarm is necessary. The adaptive method 150 waits 160 until at least one reset condition is satisfied. Reset conditions include that a fixed period of time has elapsed, that a current physiological score and/or physiological parameter value has deteriorated by a predetermined amount compared to the physiological score and/or physiological parameter value at the time of the alarm, that the current physiological score and/or physiological parameter value has deteriorated by more than half the distance from a previous physiological score and/or physiological parameter value (e.g., the physiological score and/or physiological parameter value at the time of the alarm), the physiological score and/or physiological parameter value of the physiological scoring system and/or physiological parameter has fallen below a fixed threshold at least once since the alarm, the physiological score and/or physiological parameter value of the physiological scoring system and/or physiological parameter has fallen below a threshold determined by a current limit of the alarm level, and a threshold that varies based on a typical of the intervention applied. The reset conditions may be employed alone or in combination with one another. Once the intervention is complete and at least one reset condition is met, the alarm level is reset 158.

In some embodiments, the physiological scoring system and/or physiological parameter is a vital signs index (VIX). VIX is a physiological scoring system that typically combines low latency data, such as current physiological parameter values, and optionally high latency data, such as laboratory test results, and/or statistical data, such as demographics, into a single value that reflects the stability of a patient's physiological condition, e.g., the patient's hemodynamic status, stability of the lungs, nutritional stability, and the like. The VIX value may be calculated continuously and/or upon occurrence of an event, such as a timer event, a user input event, availability of new data, and so forth. Further, in some embodiments, the VIX values are saved for historical analysis.

Calculating a VIX value for stability of a physiological state by providing a value for a predictor variable to a VIX model that generates the VIX value based on the predictor variable. The predictor variables are one or more of vital signs, features extracted from statistics (e.g., ethnicity) related to determining the stability of the physiological condition, and the like. The VIX values generated by the model are typically probabilities. For example, VIX values typically range from 0 to 1, with the closer the value is to 1, the more likely the patient is unstable. The VIX model may employ any predictive model approach, such as logistic regression, polynomial logistic regression, linear regression, and vector support machine learning.

In some embodiments, the VIX model comprises a logistic regression model for hemodynamic stability, having the form:

$\mathrm{VIX}=\frac{1}{1+{\mathrm{\&beta;}}^{-B^2}}$

wherein,

Z＝γ+β₁*SBP+β₂*SI+…

the model takes into account SBP and SI, which are highly significant predictive variables in determining hemodynamic stability, resulting in a VIX between zero and one₁The coefficient of SBP is negative, as SBP becomes lower, VIX tends to increase, reflecting that the patient is approaching a less stable state, moreover, β₂The SI coefficient is positive. As SI becomes higher, VIX also tends to increase, again decreasing the stability of the reaction.

Referring to fig. 17, a diagram of VIX and corresponding inputs, SBP and HR over a several hour time course for a patient is illustrated. Generating a reminder at 214.5 hours because the VIX value of the patient crossed the reminder threshold. During the three hour time course following the first reminder, no intervention was taken. This is interpreted to mean that the patient's dynamics at the time of the first reminder are normal, or acceptable, for that particular patient. Thus, after three hours have elapsed, no reminder is generated because the patient's VIX is not much higher than when the first reminder was made. However, by the time of 222.5 hours, VIX has increased significantly, so another reminder is issued for the patient. It should be noted that at 226 hours, the clinician administered the vasopressor did indicate that the patient experienced a clinically significant hemodynamic instability event.

In some embodiments, the physiological scoring system and/or the physiological parameter is baseline vix (bvix). bVIX is a physiological scoring system that indicates what behavior VIX has performed over a predetermined amount of time (e.g., three hours) in the past. A number of methods may be used to evaluate the trend of a series of VIX values. Some methods are more elaborate than others. In some embodiments, the bVIX value is the maximum VIX value or a 90 percentile VIX value over a predetermined amount of time in the past.

In view of the foregoing, a method of sensitively resetting individual physiological differences in a patient is provided by interpreting clinical data regarding an intervention, or lack thereof, performed by an attending clinician in the context of the condition of the patient at the time of the first alarm. The method may suitably be used to create a predictive alert system that can learn and adapt to the dynamics of an individual in the absence of direct clinical feedback.

Referring to FIG. 18, a block diagram illustrates one embodiment of an IT infrastructure 200 for a medical facility, such as a hospital. The IT infrastructure 200 includes one or more bedside or point of care patient monitoring systems 202, a patient information system 204, one or more patient information display systems 206, and the like interconnected via a communication network 208. It is contemplated that communication network 208 includes one or more of the internet, a local area network, a wide area network, a wireless network, a wired network, a cellular network, a data bus, and the like.

The patient monitoring system 202 receives physiological scores and/or physiological parameter values for patients (not shown) attended by a medical facility. In general, the patient monitoring system 202 receives physiological scores and/or physiological parameter values automatically collected via one or more sensors 210, such as Electrocardiogram (ECG) electrodes, blood pressure sensors, SpO, and/or from other components of the IT infrastructure 200₂Sensors, pulse sensors, thermometers, respiration sensors, exhaled gas sensors, non-invasive blood pressure (NBP) sensors, etc., said other components of the IT infrastructure 200 such as laboratory equipment or other patient monitoring systems. However, the patient monitoring system 202 may receive the physiological scores and/or physiological parameter values manually collected by the clinician, for example, via the user input device 212. In some embodiments, when the physiological score and/or physiological parameter value is accepted from a user input device, such user input may be facilitated using display 214. The physiological score and/or physiological parameter value is typically received continuously, but may alternatively be received upon the occurrence of an event, such as a timer event.

When the patient monitoring system receives the physiological score and/or the physiological parameter value, the corresponding deterioration detection module 216 is used to apply the method 100 for generating a patient alarm using a stepped alarm scheme to detect patient deterioration. In certain embodiments, the physiological score is customized for the patient based on patient information in the patient information system 204. The patient monitoring system generates an alert whenever a deterioration is detected. In certain embodiments, the alert is generated as an audible and/or visual alert via, for example, a respective display. In other embodiments, the notification of patient deterioration is provided to another component of the IT infrastructure, such as one of the patient information display systems 206. Further, in some embodiments, rather than a predetermined amount of time elapsing, the method 150 of FIG. 16 is used for a reset.

To perform the above-mentioned functions, the patient monitoring system 202 suitably includes one or more memories 218 and one or more processors 220. Common examples of patient monitoring systems include patient wearable patient monitors, bedside patient monitors, spot check patient monitors, and central patient monitors. The memory 218 stores executable instructions for performing one or more of the above-described functions of the patient monitoring system, and implements the deterioration detection module 216 as appropriate. The processor 220 executes executable instructions stored on the memory 218 to perform functions associated with the patient monitoring system 202. When the patient monitoring system 202 is operative to communicate over the communication network 208, the patient monitoring system 202 also includes one or more communication units 222, the communication units 222 facilitating communication between the processor 220 and the communication network 208.

The patient information system 204, such as a central recorded medical database, typically serves as a central repository of patient information, including, for example, Electronic Medical Records (EMRs). Additionally or alternatively, the patient information system 204 receives and stores in one or more memories 224 thereof one or more of the physiological scores, physiological parameter values, and clinical data for the patient. Typically, the physiological parameter values and/or physiological scores are received from components of the IT infrastructure 200 via, for example, the communication network 208, but the measurements may be manually input via one or more user input devices 212, 216. For the latter, a user interface presented via display 228 may facilitate such manual input. The patient information system 204 also allows components of the IT infrastructure 200 to access stored data, such as EMR and/or physiological parameter values for a patient, via the communication network 208.

To perform the above-mentioned functions, the patient information system 204 suitably includes one or more communication units 230, memory 224, and one or more processors 232. The communication unit 230 facilitates communication between the processor 232 and the communication network 208. The memory 224 stores executable instructions for controlling the processor 232 to perform one or more of the above-described functions of the patient information system 204. The processor 232 executes executable instructions stored on the memory 224.

The patient information display system 206 receives physiological scores and/or physiological parameter values for patients attended by the medical facility from components of the IT infrastructure 200 over the communication network 208. Additionally or alternatively, the patient information display system 206 receives an alert for a patient attended by a medical facility. Using the received data, the patient information display system 206 updates the associated display 234, graphically presents the data to the clinician, and/or generates an alert. For the latter, an audible and/or visual alarm, e.g. via the display 234, is contemplated. Further, in certain embodiments, the user input device 236 of the patient information display system 206 is employed to acknowledge the alert to the component of the IT infrastructure 200 that generated the alert.

To implement the above functionality, the patient information display system 206 suitably includes one or more communication units 238, one or more memories 240, and one or more processors 242. The communication unit 238 facilitates communication between the processor 242 and the communication network 208. The memory 240 stores executable instructions for controlling the processor 242 to perform one or more of the above-described functions of the patient information display system 206. Processor 242 executes executable instructions stored on memory 240.

Memory as used herein includes one or more of the following: a non-transitory computer readable medium; magnetic disks or other magnetic storage media; an optical disc or other optical storage medium; random Access Memory (RAM), Read Only Memory (ROM), or other electronic storage devices or chips or collections of operatively interconnected chips; an internet/intranet server from which the stored instructions may be retrieved via an internet/intranet or local area network; and so on. Further, a processor as used herein includes one or more of the following: microprocessors, microcontrollers, Graphics Processing Units (GPUs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs), and the like; the user input device comprises one or more of: a mouse, keyboard, touch screen display, one or more buttons, one or more switches, one or more triggers, etc.; and the display comprises one or more of: LCD displays, LED displays, plasma displays, projection displays, touch screen displays, and the like.

The invention has been described with reference to the preferred embodiments. Modifications and alterations may occur to others upon reading and understanding the preceding detailed description. For example, although the methods and systems disclosed herein are made using a general ward population in question, the alarm staging may be applied with other healthcare settings, such as ICU, emergency or home monitoring. It is intended that the invention be construed as including all such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims (19)

1. A system (200) for generating a patient alarm using a stepped alarm scheme, the system (202) comprising:

one or more processors (220) programmed to:

receiving a physiological score and/or a physiological parameter value comprising a vital sign index (VIX) value calculated from vital sign measurements with low latency and data with high latency compared to the vital sign measurements using a predictive model of hemodynamic instability;

comparing the physiological score and/or the physiological parameter value to a plurality of stepped alert levels;

issuing an alert in response to the physiological score and/or the physiological parameter value falling within a non-suppressed alert level of the alert levels; and,

setting a first suppression period for the non-suppression alarm level after issuing the alarm;

wherein the inhibition period is a period of: during the suppression period, the respective alarm level is suppressed so as not to trigger an alarm.

2. The system (200) according to claim 1, wherein receiving a physiological score and/or a physiological parameter value comprises calculating the physiological score from a physiological parameter value.

3. The system (200) according to either one of claims 1 and 2, wherein the physiological score is calculated using an abnormality scoring system, the abnormality score based on a plurality of vital sign measurements of the physiological parameter value.

4. The system (200) according to either one of claims 1 and 2, wherein the first suppression period is set in response to an acknowledgement of the alarm.

5. The system (200) according to either one of claims 1 and 2, wherein the processor (220) is further programmed to:

generating the plurality of stepped alarm levels from an initial alarm level and an increment Δ, wherein the plurality of stepped alarm levels includes the initial alarm level and one or more alarm levels of increasing severity that are separated from other of the initial alarm level and the alarm level by the increment Δ.

6. The system (200) according to either one of claims 1 and 2, wherein the processor (220) is further programmed to:

in response to issuing an alarm, setting a second suppression period on all of the plurality of stepped alarm levels, the second suppression period being shorter than the first suppression period.

7. The system (200) according to either one of claims 1 and 2, wherein the processor (220) is further programmed to:

evaluating an abnormality score of the physiological score for changes in composition; and,

issuing a second alert in response to the change in the composition of the anomaly score; and,

after issuing the second alarm, setting a suppression period for the non-suppression alarm level.

8. The system (200) according to either one of claims 1 and 2, wherein the processor (220) is further programmed to:

resetting the non-suppressed alarm level in response to at least one of a plurality of reset conditions being met, the reset conditions including:

a first reset condition comprising:

a predetermined amount of time has elapsed;

no intervention is given during the predetermined amount of time; and

the current physiological score and/or physiological parameter value is worse by a predetermined amount compared to the physiological score and/or physiological parameter value; and the number of the first and second groups,

a second reset condition comprising:

the predetermined amount of time has elapsed;

intervention is given during the predetermined amount of time; and

the reset condition is satisfied.

9. The system (200) according to either one of claims 1 and 2, further including at least one of:

one or more sensors (210) that measure one or more vital signs of the physiological parameter values;

one or more user input devices (212) that receive values of the physiological parameter values and/or the physiological scores; and the number of the first and second groups,

a communication network (208) that exchanges physiological scores and/or physiological parameter values between the system (200) and other components (202, 204, 206) connected to the communication network (208);

wherein the physiological score and/or the physiological parameter value is received from at least one of the sensor (210), the user input device (212), and the communication network (208).

10. A method (100) for generating a patient alert using a stepped alert scheme, the method (100) comprising:

receiving (102) a physiological score and/or a physiological parameter value comprising a vital sign index (VIX) value calculated from low latency vital sign measurements and data with a high latency compared to the vital sign measurements using a predictive model of hemodynamic instability;

comparing (106) the physiological score and/or the physiological parameter value to a plurality of stepped alert levels; and the number of the first and second groups,

issuing (110) an alarm in response to the physiological score and/or the physiological parameter value falling within a non-suppressed alarm level of the alarm levels; and the number of the first and second groups,

setting (112) a first suppression period for the non-suppression alarm level after issuing the alarm;

wherein the inhibition period is a period of: during the suppression period, the respective alarm level is suppressed so as not to trigger an alarm.

11. The method (100) according to claim 10, wherein receiving (102) a physiological score and/or a physiological parameter value comprises calculating the physiological score from the physiological parameter value.

12. The method (100) according to either one of claims 10 and 11, further including:

evaluating (116) an abnormality score of the physiological score for a change in composition; and the number of the first and second groups,

issuing (188) a second alert in response to a change in the composition of the anomaly score; and the number of the first and second groups,

after issuing the second alarm, setting (118) a suppression period for the non-suppression zone for the alarm level.

13. The method (100) according to either one of claims 10 and 11, further including:

generating (108) the plurality of stepped alarm levels from an initial alarm level and an incremental delta, wherein the plurality of stepped alarm levels includes the initial alarm level and one or more alarm levels having progressively increasing severity, the alarm levels being separated from other of the initial alarm level and the alarm levels by one or more of the incremental deltas.

14. The method (100) according to either one of claims 10 and 11, further including:

resetting the non-suppressed alarm level in response to at least one of a plurality of reset conditions being met, the reset conditions including:

a first reset condition comprising:

a predetermined amount of time has elapsed;

no intervention is given during the predetermined amount of time; and the number of the first and second groups,

the current physiological score and/or physiological parameter value is worse by a predetermined amount compared to the physiological score and/or physiological parameter value; and the number of the first and second groups,

a second reset condition comprising:

the predetermined amount of time has elapsed;

intervention is given during the predetermined amount of time; and the number of the first and second groups,

the reset condition is satisfied.

15. An apparatus for generating a patient alert using a stepped alert scheme, the apparatus comprising:

means for receiving (102) a physiological score and/or a physiological parameter value comprising a vital sign index (VIX) value calculated from low latency vital sign measurements and data with high latency compared to the vital sign measurements using a predictive model of hemodynamic instability;

means for comparing (106) the physiological score and/or the physiological parameter value to a plurality of stepped alert levels; and the number of the first and second groups,

means for issuing (110) an alarm in response to the physiological score and/or the physiological parameter value falling within a non-suppressed alarm level of the alarm levels; and the number of the first and second groups,

means for setting (112) a first suppression period for the non-suppressed alarm level after issuing the alarm;

wherein the inhibition period is a period of: during the suppression period, the respective alarm level is suppressed so as not to trigger an alarm.

16. The apparatus as recited in claim 15, wherein receiving (102) a physiological score and/or a physiological parameter value includes calculating the physiological score from the physiological parameter value.

17. The apparatus according to any one of claims 15 and 16, further comprising:

means for evaluating (116) an abnormality score of the physiological score for a change in composition; and the number of the first and second groups,

means for issuing (188) a second alert in response to a change in the composition of the anomaly score; and the number of the first and second groups,

means for setting (118) a suppression period for the non-suppression zone for the alarm level after issuing the second alarm.

18. The apparatus according to any one of claims 15 and 16, further comprising:

means for generating (108) the plurality of stepped alarm levels from an initial alarm level and an incremental delta, wherein the plurality of stepped alarm levels includes the initial alarm level and one or more alarm levels having progressively increasing severity that are separated from other of the initial alarm level and the alarm level by one or more of the incremental deltas.

19. The apparatus according to any one of claims 15 and 16, further comprising:

means for resetting the non-suppressed alarm level in response to at least one of a plurality of reset conditions being met, the reset conditions comprising:

a first reset condition comprising:

a predetermined amount of time has elapsed;

no intervention is given during the predetermined amount of time; and the number of the first and second groups,

the current physiological score and/or physiological parameter value is worse by a predetermined amount compared to the physiological score and/or physiological parameter value; and the number of the first and second groups,

a second reset condition comprising:

the predetermined amount of time has elapsed;

intervention is given during the predetermined amount of time; and the number of the first and second groups,

the reset condition is satisfied.