CN114758786B - Post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters - Google Patents
Post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters Download PDFInfo
- Publication number
- CN114758786B CN114758786B CN202210392852.4A CN202210392852A CN114758786B CN 114758786 B CN114758786 B CN 114758786B CN 202210392852 A CN202210392852 A CN 202210392852A CN 114758786 B CN114758786 B CN 114758786B
- Authority
- CN
- China
- Prior art keywords
- subsystem
- data
- early warning
- post
- patient
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- Theoretical Computer Science (AREA)
- Primary Health Care (AREA)
- Epidemiology (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Pathology (AREA)
- Biophysics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention relates to a post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters, which comprises a physiological parameter sensing subsystem, a warning model updating subsystem, a dynamic warning subsystem, a post-traumatic hemorrhagic shock morbidity judging subsystem and a silencing subsystem, wherein the dynamic warning subsystem is used for calculating the occurrence probability of post-traumatic hemorrhagic shock of a patient, and the post-traumatic hemorrhagic shock morbidity judging subsystem corrects the predicted probability value according to medical environment data and the like of the position of the post-traumatic hemorrhagic shock morbidity judging subsystem. And the silencing subsystem controls the early warning result of the dynamic early warning subsystem according to the difference among the test data, the training data and the tag data. The invention only uses common noninvasive parameters, does not need laboratory data, can be used for scenes such as remote areas, sudden public health events, battlefield first-line situations and the like, eliminates injuries to individuals of patients caused by frequent collection of laboratory parameters, and reduces the use cost and early warning error probability of the system.
Description
Technical Field
The invention relates to the field of artificial intelligence technology and medical health, in particular to a post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters.
Background
Hemorrhagic shock is a hypovolemic shock that is a consequence of severe blood loss resulting in insufficient oxygen delivery at the cellular level. If the blood loss is not controlled, the patient can die rapidly, and the number of the hemorrhagic shock death caused by the wound accounts for 10-40% of the total death number of the wound. Retrospective studies on multiple databases have shown that near 1/4 post-traumatic admitted patients develop symptoms of hemorrhagic shock during admission. For most patients, the impact of shock on the body is initially reversible, but repeated or prolonged hypotension, the use of high doses of vasopressors, may deteriorate prognosis.
Medical personnel intermittently evaluate the monitored patient vital signs and rely on individual physiological parameters to identify patients at risk of deterioration. These early warning systems do not use the full information of the patient and therefore alarms are often inaccurate and can easily lead to alarm fatigue. The low-precision early warning system is not beneficial to the identification and interpretation of relevant information of clinicians, nurses or doctors are difficult to continuously monitor or evaluate ICU patients, and the low-frequency low-precision early warning system is difficult to adapt to patients with acute changes of illness states. Due to lack of sufficiently accurate predictive tools, clinicians may resort to subjective judgment, which increases the risk of the physician taking action based on the relevant information. Preventive transfusion or the use of 100% mechanical ventilation has a significant effect on reducing the mortality rate of hemorrhagic shock. If an early warning and evaluating system can dynamically and early predict the future occurrence probability of hemorrhagic shock of a patient after the trauma, a timely preventive treatment plan can be provided for the high-risk patient, so that the mortality rate and the medical cost are greatly reduced. A doctor in an intensive care unit can obtain a large amount of patient physiological data and measurement indexes from a plurality of monitoring systems, but the early recognition of the patient's condition deterioration is hindered only by the limited capability of human processing complex information, and the patient's condition is difficult to monitor in real time by the conventional recognition early warning method. Deep learning techniques perform well in analyzing complex signals in a data rich environment. The disclosure of large volumes of data collected in the ICU, and large medical critical datasets such as THE MEDICAL Information Mart for INTENSIVE CARE III (MIMIC III) database, eICU, amsterdamUMC, is critical to the development of machine learning use in this environment.
The patent CN201910570791 provides a time sequence prediction method for traumatic hemorrhagic shock injury, which can extract traumatic hemorrhagic shock injury data from a database, and perform data processing on the traumatic hemorrhagic shock injury data, and has the functions of processing data outliers, performing linear and cluster deficiency on the data, designing stepped indexes on the processed data, constructing a prediction model by applying index stepped results and different types of classifiers, and predicting the results after a preset time period through the prediction model. The invention can implement real-time dynamic prediction early warning based on time sequence for traumatic hemorrhagic shock by using indexes capable of being monitored in real time, but does not reserve clinical intervention time and does not carry out interpretable analysis on the model.
In addition, in the field of big medical health data, clinical data of patients can be divided into cross section data with only one cross section and time sequence data with a plurality of cross sections, and the time sequence prediction accuracy is higher than that of cross section prediction due to the characteristics of large information content, trend fluctuation and the like, and the sliding prediction and real-time monitoring and early warning of illness state can be realized. However, because the measurement indexes of wounded persons are different, the measurement time of the wounded persons is different, most of the test indexes cannot be measured for multiple times in a short period, and the like, the problems of sparse medical data and missing medical data are directly caused to be abnormally serious. The prior art has at least the following problems:
Firstly, a mature deficiency tonifying technology system is not adopted in the data deficiency aspect, most of the deficiency tonifying technology system adopts a mean deficiency or linear deficiency, the deficiency tonifying method is single, and the problems of poor data quality, larger difference with real data and the like still exist after deficiency tonifying;
Secondly, the prior art method adopts a section prediction mode, for example, the section is predicted after the data is averaged, the obtained result is 'ending', and rolling prediction and real-time monitoring of illness state cannot be realized;
Thirdly, the existing method utilizes a small amount of time sequences to predict, and only vital sign indexes with low measurement cost and more times, such as heart rate, blood pressure and the like, are selected, so that the predicting effect is poor;
Fourth, the existing early warning method does not adopt related treatment for reducing the false alarm rate, and frequent model false alarms lead to alarm fatigue of medical staff, thereby affecting positive clinical treatment of patients in critical states.
The prior disclosed related technology can be used for predicting the possibility of the onset of the post-traumatic hemorrhagic shock of a patient by using physiological parameters of the patient, but lacks a method for reserving clinical intervention time between an early warning model operation time point and an onset early warning time point and effectively reducing the false alarm rate.
Disclosure of Invention
Aiming at the problems of the existing post-traumatic hemorrhagic shock dynamic early warning system, the invention discloses a post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters, which can be used for predicting the possibility of post-traumatic hemorrhagic shock morbidity in a future multi-time scale of a patient, reducing the false alarm rate and realizing effective early warning.
The invention discloses a post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters, which comprises a physiological parameter sensing subsystem, a warning model updating subsystem, a dynamic warning subsystem, a post-traumatic hemorrhagic shock morbidity judging subsystem and a silencing subsystem, wherein the physiological parameter sensing subsystem is used for sensing the physiological parameters of the post-traumatic hemorrhagic shock;
The physiological parameter sensing subsystem is connected with the dynamic early warning subsystem, the dynamic early warning subsystem is respectively connected with the early warning model updating subsystem and the post-traumatic hemorrhagic shock morbidity judging subsystem, and the silencing subsystem is respectively connected with the post-traumatic hemorrhagic shock morbidity judging subsystem, the dynamic early warning subsystem and the early warning model updating subsystem.
The physiological parameter sensing subsystem is used for monitoring and collecting physiological parameter data information of a patient in real time, preprocessing the collected physiological parameter data of the patient, and sending the preprocessed data to the dynamic early warning subsystem.
The physiological parameter data information of the patient comprises conventional noninvasive physiological parameters of the patient and physiological parameter time sequence data in a learning window of the patient.
The preprocessing of the collected patient physiological parameter data comprises the steps of utilizing the discrete sampling values to form sampling vectors, calculating a cross-correlation matrix of the sampling vectors, utilizing a feature extraction method to process the cross-correlation matrix to obtain a feature value vector, utilizing the feature value vector to weight the sampling vectors to obtain a smooth value of the patient physiological parameter as preprocessed data, and therefore effectively reducing interference caused by error values such as collected wild values to a subsequent early warning judgment process and reducing false warning probability of the early warning judgment process.
The pretreatment of the collected patient physiological parameter data comprises the steps that a sampling vector formed by the patient physiological parameter discrete sampling values collected in a period of time is represented as [ x 1,x2,…,xN ], N is the number of the patient physiological parameter discrete sampling values collected in a period of time, a mutual matrix C of the sampling vector is obtained through calculation, and the characteristic value decomposition is carried out on the mutual matrix C to obtain:
C=VDVH,
And C is a characteristic vector matrix, D is a characteristic value matrix, diagonal elements of the matrix D are normalized and used as weight vectors, and the sampling vectors are weighted and summed to obtain smooth values of physiological parameters of the patient, and the smooth values are used as preprocessed data.
The dynamic early warning subsystem is used for receiving and cleaning the conventional non-invasive physiological parameters of the patient and the physiological parameter time sequence data in the patient learning window, which are acquired by the physiological parameter sensing subsystem, calculating to obtain the original probability of post-traumatic shock incidence of the patient in the prediction time window, calculating the occurrence probability of post-traumatic shock of the patient in the future prediction time window according to the conventional non-invasive physiological parameters acquired by the physiological parameter sensing subsystem and the physiological parameter time sequence data in the patient learning window, obtaining a prediction probability value, transmitting the prediction probability value to the post-traumatic shock incidence judging subsystem, correcting the prediction probability value according to the medical environment data and the doctor treatment experience evaluation value of the post-traumatic shock incidence judging subsystem, judging that the post-traumatic shock incidence of the patient occurs in the future prediction time window by the post-traumatic shock dynamic early warning system based on the non-invasive parameters if the corrected prediction probability value is larger than a first prediction probability threshold, and judging that the post-traumatic shock incidence of the patient does not occur in the future prediction time window if the corrected prediction probability value is smaller than a second prediction probability threshold.
The early warning model updating subsystem monitors the medical environment of the place where the system is located, if the early warning model updating subsystem monitors the change of the medical environment of the place where the system is located, the early warning model updating subsystem updates the weight of the deep learning model in the dynamic early warning subsystem by using an incremental learning method according to the conventional noninvasive physiological parameters of the patient and prognosis information including death condition of the patient, occurrence condition of hemorrhagic shock after trauma and hospitalization time period, which are acquired by the physiological parameter sensing subsystem, so that the early warning result of the system is adapted to the medical environment of the place where the system is located.
The silence subsystem judges the difference among test data, training data and label data of the post-traumatic hemorrhagic shock dynamic early warning of the intelligent dynamic early warning subsystem, if the three types of data have significant differences, the silence subsystem controls the early warning result of the dynamic early warning subsystem not to be output, collects the early warning result data of the dynamic early warning subsystem in a period of time, extracts the early warning result data with the maximum prediction probability in the period of time, and outputs the early warning result data to the post-traumatic hemorrhagic shock morbidity judging subsystem.
The silence subsystem judges the difference of test data, label data and training data of the early warning model updating subsystem, and comprises the following steps:
And regarding the data of the early warning model updating subsystem as a stable random process, respectively establishing corresponding autoregressive-moving average models, namely ARMA models, aiming at test data, label data and training data to respectively obtain a first ARMA model, a second ARMA model and a third ARMA model, calculating a cross-correlation matrix of the coefficients of the three ARMA models, calculating the cross-correlation matrix to obtain a maximum characteristic value, and judging the difference of the test data, the label data and the training data by using the maximum characteristic value.
When the maximum characteristic value is larger than the difference judging threshold value, the significance difference among the test data, the label data and the training data of the early warning model updating subsystem is judged.
The silence subsystem judges the difference of test data, label data and training data of the early warning model updating subsystem, and comprises the following steps:
The silence subsystem predicts and obtains a label data predicted value by utilizing training data and adopting a deep learning model, calculates the credibility among the test data, the training data and the label data predicted value by adopting a credibility evaluation method, and judges that the early warning model updates the early warning data of the subsystem and the obvious difference among the label data and the training data when the credibility is lower than a credibility threshold value.
The silence subsystem judges the difference of test data, label data and training data of the early warning model updating subsystem, and comprises the following steps:
The method comprises the steps of forming a sample set by utilizing training data and a label data predicted value, dividing the sample set into a first reference sample set and a second reference sample set corresponding to the test data according to the label category to which each test data belongs, wherein the first reference sample set is a set formed by samples corresponding to the label category to which the test data belongs, the second reference sample set is a set formed by samples except for the other samples in the first reference sample set in the sample set, calculating the distance between each sample in the first reference sample set and the reference sample set by utilizing a consistency metric calculation function to obtain a first sample set distance value vector, calculating the distance between each test data and the first reference sample set and the second reference sample set corresponding to the test data by utilizing the consistency metric calculation function, correspondingly obtaining a first set distance value and a second set distance value of the test data, sequencing the first set distance value of the test data in the first sample set distance value vector to obtain a sequence number value, calculating the ratio of the sequence number value to the total number of samples of the first reference sample set as the reliability of the test data, and comparing the reliability of the test data with a pre-warning data threshold value when the reliability of the test data is lower than the reliability of the test data is higher than a threshold value, and comparing the reliability of the pre-warning data with the pre-warning data when the reliability of the test data is compared with the pre-warning data is calculated.
The deep learning model can adopt a stack type deep learning model, a convolutional neural network module and the like.
The stack type deep learning model comprises a convolution layer, a two-way long-short-term memory layer and a self-attention layer.
The dynamic early warning subsystem interpolates the missing patient physiological parameter time sequence data in the minimum time unit by adopting a self-adaptive interpolation method, and the specific steps comprise:
s1, calculating a median Midparam and a quartile IQRIDPARAM of sampling frequencies of physiological parameters during the period from the admission to the discharge of each patient;
s2, selecting physiological parameters with missing values of the patient with missing data, recording the parameters as parameters of the param, and selecting the time positions of empty values or abnormal values in the parameters of the param as selected positions according to the ascending order of the acquisition time;
And S3, performing data interpolation on the selected position according to the following priority sequence, wherein firstly, the median of the effective values of the first (Midparam + IQRIDPARAM) sampling time lengths of the selected position is used as the data interpolation value of the selected position, secondly, the median of the effective values of the first (Midparam +2) sampling time lengths of the selected position is used as the data interpolation value of the selected position, thirdly, the median of the effective values of the acquired parameters of the param is used as the data interpolation value of the selected position, and finally, the median of the parameters of the param recorded in the database is used as the data interpolation value of the selected position.
The post-traumatic hemorrhagic shock morbidity judging subsystem receives the original probability of post-traumatic hemorrhagic shock morbidity of a patient within a prediction time window, and the post-traumatic hemorrhagic shock morbidity judging subsystem obtains a classification threshold D of the stack type deep learning model according to the sensitivity and the specificity of the stack type deep learning model in the dynamic early warning subsystem on the test set data, so that when the classification threshold is D, the difference between the sensitivity and the specificity of the stack type deep learning model on the test set data is minimum.
The post-traumatic hemorrhagic shock morbidity judging subsystem corrects the occurrence probability value of post-traumatic hemorrhagic shock according to the medical environment data and the doctor treatment experience evaluation value of the position of the post-traumatic hemorrhagic shock morbidity judging subsystem, and the corrected post-traumatic hemorrhagic shock morbidity probability calculating formula is as follows:
Where C is the probability threshold.
After the early warning model updating subsystem monitors the change of the medical environment of the place, the physiological parameters and prognosis information of the patient are collected, or the input physiological parameters and prognosis information of the patient are used, the model parameters are updated on the basis of not changing the stack type deep learning model structure by combining with an incremental training method, and the physiological parameters of the patient in the range of a learning window are transmitted to the dynamic early warning subsystem after updating the stack type deep learning model parameters in real time.
The beneficial effects of the invention are as follows:
The invention discloses a post-traumatic hemorrhagic shock dynamic early warning deep learning system based on noninvasive parameters, which only uses common noninvasive parameters, does not need laboratory data, expands the application range of the system, enables the system to be used in remote areas, sudden public health events and first-line conditions of battlefields, eliminates injuries to individuals of patients caused by frequent collection of laboratory parameters, and reduces the use cost of the system. Compared with the traditional linear addition regression model, the automatic and flow intelligent dynamic early warning subsystem has higher computational complexity, is more suitable for the common nonlinear problem in the real problem, and can provide better post-traumatic hemorrhagic shock morbidity early warning capability. The pre-warning interval reserved for clinical intervention can provide sufficient time for a doctor to design a patient treatment plan. The analysis result of the system is consistent with the clinical research result, the model performance is excellent, the system can automatically update the model weight according to the environmental change so as to adapt to different clinical environments, and meanwhile, the false probability of the post-traumatic hemorrhagic shock morbidity prediction is effectively reduced.
Drawings
FIG. 1 is a block diagram of a dynamic early warning system for post-traumatic hemorrhagic shock based on non-invasive parameters according to the present invention;
FIG. 2 is a diagram of the relationship between observation window, delay window, and early warning window in the dynamic early warning subsystem of the present invention;
fig. 3 is a flow chart of the silence subsystem operation of the present invention.
Detailed Description
In order to make the present invention better understood by those skilled in the art, the following description will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings, and it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The terms first, second and the like in the description and in the claims and in the above-described figures are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order. Furthermore, the terms "comprise" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, apparatus, article, or device that comprises a list of steps or elements is not limited to the list of steps or elements but may, in the alternative, include other steps or elements not expressly listed or inherent to such process, method, article, or device.
Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the invention. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.
Fig. 1 is a block diagram of a dynamic early warning system for post-traumatic hemorrhagic shock based on noninvasive parameters, and fig. 2 is a diagram of the relationship between an observation window, a delay window and a warning window in the dynamic warning subsystem. In FIG. 2, a learning window, a time range with available data, is used for the input of the intelligent dynamic early warning sub-module. Predictive window, a period of time to determine if post-traumatic hemorrhagic shock has occurred. Delay window-the time difference between the prediction window and the learning window. Wherein T0 is the time point when the patient enters the ICU, T1 is set as the current time point, T1 is more than or equal to T0, x1 is the time length of the learning window, T1-x1 is more than or equal to T0, x2 is the delay window length, the time period of the prediction window is [ T1+x2, T1+x2+x3], and x3 is the prediction window length. Fig. 3 is a flow chart of the silence subsystem operation of the present invention.
The noninvasive parameters described in the present invention refer to parameters that can be measured without causing trauma to the patient's body or laboratory environment, such as heart rate, noninvasive blood pressure, respiration rate, gender, age, etc. Non-invasive parameters are distinguished from laboratory parameters.
Embodiment one:
the invention discloses a post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters, which comprises a physiological parameter sensing subsystem, a warning model updating subsystem, a dynamic warning subsystem, a post-traumatic hemorrhagic shock morbidity judging subsystem and a silencing subsystem, wherein the physiological parameter sensing subsystem is used for sensing the physiological parameters of the post-traumatic hemorrhagic shock;
the physiological parameter sensing subsystem is used for monitoring and collecting physiological parameter data information of a patient in real time, preprocessing the collected physiological parameter data of the patient, and sending the preprocessed data to the dynamic early warning subsystem.
The physiological parameter sensing subsystem is connected with the dynamic early warning subsystem, the dynamic early warning subsystem is respectively connected with the early warning model updating subsystem and the post-traumatic hemorrhagic shock morbidity judging subsystem, and the silencing subsystem is respectively connected with the post-traumatic hemorrhagic shock morbidity judging subsystem, the dynamic early warning subsystem and the early warning model updating subsystem.
The physiological parameter data information of the patient comprises conventional noninvasive physiological parameters of the patient and physiological parameter time sequence data in a learning window of the patient.
The preprocessing of the collected patient physiological parameter data comprises the steps of utilizing the discrete sampling values to form sampling vectors, calculating a cross-correlation matrix of the sampling vectors, utilizing a feature extraction method to process the cross-correlation matrix to obtain a feature value vector, utilizing the feature value vector to weight the sampling vectors to obtain a smooth value of the patient physiological parameter as preprocessed data, and therefore effectively reducing interference caused by error values such as collected wild values to a subsequent early warning judgment process and reducing false warning probability of the early warning judgment process.
The pretreatment of the collected patient physiological parameter data comprises the steps that a sampling vector formed by the patient physiological parameter discrete sampling values collected in a period of time is represented as [ x 1,x2,…,xN ], N is the number of the patient physiological parameter discrete sampling values collected in a period of time, a mutual matrix C of the sampling vector is obtained through calculation, and the characteristic value decomposition is carried out on the mutual matrix C to obtain:
C=VDVH,
And C is a characteristic vector matrix, D is a characteristic value matrix, diagonal elements of the matrix D are normalized and used as weight vectors, and the sampling vectors are weighted and summed to obtain smooth values of physiological parameters of the patient, and the smooth values are used as preprocessed data.
The dynamic early warning subsystem is used for receiving and cleaning the conventional non-invasive physiological parameters of the patient and the physiological parameter time sequence data in the patient learning window, which are acquired by the physiological parameter sensing subsystem, calculating to obtain the original probability of post-traumatic shock incidence of the patient in the prediction time window, calculating the occurrence probability of post-traumatic shock of the patient in the future prediction time window according to the conventional non-invasive physiological parameters acquired by the physiological parameter sensing subsystem and the physiological parameter time sequence data in the patient learning window, obtaining a prediction probability value, transmitting the prediction probability value to the post-traumatic shock incidence judging subsystem, correcting the prediction probability value according to the medical environment data and the doctor treatment experience evaluation value of the post-traumatic shock incidence judging subsystem, judging that the post-traumatic shock incidence of the patient occurs in the future prediction time window by the post-traumatic shock dynamic early warning system based on the non-invasive parameters if the corrected prediction probability value is larger than a first prediction probability threshold, and judging that the post-traumatic shock incidence of the patient does not occur in the future prediction time window if the corrected prediction probability value is smaller than a second prediction probability threshold.
The early warning model updating subsystem monitors the medical environment of the place where the system is located, if the early warning model updating subsystem monitors the change of the medical environment of the place where the system is located, the early warning model updating subsystem updates the weight of the deep learning model in the dynamic early warning subsystem by using an incremental learning method according to the conventional noninvasive physiological parameters of the patient and prognosis information including death condition of the patient, occurrence condition of hemorrhagic shock after trauma and hospitalization time period, which are acquired by the physiological parameter sensing subsystem, so that the early warning result of the system is adapted to the medical environment of the place where the system is located.
The silence subsystem judges the difference among test data, training data and label data of the post-traumatic hemorrhagic shock dynamic early warning of the intelligent dynamic early warning subsystem, if the three types of data have significant differences, the silence subsystem controls the early warning result of the dynamic early warning subsystem not to be output, collects the early warning result data of the dynamic early warning subsystem in a period of time, extracts the early warning result data with the maximum prediction probability in the period of time, and outputs the early warning result data to the post-traumatic hemorrhagic shock morbidity judging subsystem.
The silence subsystem judges the difference of test data, label data and training data of the early warning model updating subsystem, and comprises the following steps:
And regarding the data of the early warning model updating subsystem as a stable random process, respectively establishing corresponding autoregressive-moving average models, namely ARMA models, aiming at test data, label data and training data to respectively obtain a first ARMA model, a second ARMA model and a third ARMA model, calculating a cross-correlation matrix of the coefficients of the three ARMA models, calculating the cross-correlation matrix to obtain a maximum characteristic value, and judging the difference of the test data, the label data and the training data by using the maximum characteristic value.
When the maximum characteristic value is larger than the difference judging threshold value, the significance difference among the test data, the label data and the training data of the early warning model updating subsystem is judged.
The silence subsystem judges the difference of test data, label data and training data of the early warning model updating subsystem, and comprises the following steps:
The silence subsystem predicts and obtains a label data predicted value by utilizing training data and adopting a deep learning model, calculates the credibility among the test data, the training data and the label data predicted value by adopting a credibility evaluation method, and judges that the early warning model updates the early warning data of the subsystem and the obvious difference among the label data and the training data when the credibility is lower than a credibility threshold value.
The silence subsystem judges the difference of test data, label data and training data of the early warning model updating subsystem, and comprises the following steps:
The method comprises the steps of forming a sample set by utilizing training data and a label data predicted value, dividing the sample set into a first reference sample set and a second reference sample set corresponding to the test data according to the label category to which each test data belongs, wherein the first reference sample set is a set formed by samples corresponding to the label category to which the test data belongs, the second reference sample set is a set formed by samples except for the other samples in the first reference sample set in the sample set, calculating the distance between each sample in the first reference sample set and the reference sample set by utilizing a consistency metric calculation function to obtain a first sample set distance value vector, calculating the distance between each test data and the first reference sample set and the second reference sample set corresponding to the test data by utilizing the consistency metric calculation function, correspondingly obtaining a first set distance value and a second set distance value of the test data, sequencing the first set distance value of the test data in the first sample set distance value vector to obtain a sequence number value, calculating the ratio of the sequence number value to the total number of samples of the first reference sample set as the reliability of the test data, and comparing the reliability of the test data with a pre-warning data threshold value when the reliability of the test data is lower than the reliability of the test data is higher than a threshold value, and comparing the reliability of the pre-warning data with the pre-warning data when the reliability of the test data is compared with the pre-warning data is calculated.
The deep learning model can adopt a stack type deep learning model, a convolutional neural network module and the like.
The stack type deep learning model comprises a convolution layer, a two-way long-short-term memory layer and a self-attention layer.
The dynamic early warning subsystem interpolates the missing patient physiological parameter time sequence data in the minimum time unit by adopting a self-adaptive interpolation method, and the specific steps comprise:
s1, calculating a median Midparam and a quartile IQRIDPARAM of sampling frequencies of physiological parameters during the period from the admission to the discharge of each patient;
s2, selecting physiological parameters with missing values of the patient with missing data, recording the parameters as parameters of the param, and selecting the time positions of empty values or abnormal values in the parameters of the param as selected positions according to the ascending order of the acquisition time;
And S3, performing data interpolation on the selected position according to the following priority sequence, wherein firstly, the median of the effective values of the first (Midparam + IQRIDPARAM) sampling time lengths of the selected position is used as the data interpolation value of the selected position, secondly, the median of the effective values of the first (Midparam +2) sampling time lengths of the selected position is used as the data interpolation value of the selected position, thirdly, the median of the effective values of the acquired parameters of the param is used as the data interpolation value of the selected position, and finally, the median of the parameters of the param recorded in the database is used as the data interpolation value of the selected position.
The post-traumatic hemorrhagic shock morbidity judging subsystem receives the original probability of post-traumatic hemorrhagic shock morbidity of a patient within a prediction time window, and the post-traumatic hemorrhagic shock morbidity judging subsystem obtains a classification threshold D of the stack type deep learning model according to the sensitivity and the specificity of the stack type deep learning model in the dynamic early warning subsystem on the test set data, so that when the classification threshold is D, the difference between the sensitivity and the specificity of the stack type deep learning model on the test set data is minimum.
The post-traumatic hemorrhagic shock morbidity judging subsystem corrects the occurrence probability value of post-traumatic hemorrhagic shock according to the medical environment data and the doctor treatment experience evaluation value of the position of the post-traumatic hemorrhagic shock morbidity judging subsystem, and the corrected post-traumatic hemorrhagic shock morbidity probability calculating formula is as follows:
Where C is the probability threshold.
After the early warning model updating subsystem monitors the change of the medical environment of the place, the physiological parameters and prognosis information of the patient are collected, or the input physiological parameters and prognosis information of the patient are used, the model parameters are updated on the basis of not changing the stack type deep learning model structure by combining with an incremental training method, and the physiological parameters of the patient in the range of a learning window are transmitted to the dynamic early warning subsystem after updating the stack type deep learning model parameters in real time.
The working principle of the post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters is shown in figure 1. The relation of the observation window, the delay window and the early warning window in the dynamic early warning subsystem is shown in figure 2. In FIG. 2, a learning window, a time range with available data, is used for the input of the intelligent dynamic early warning sub-module. Predictive window, a period of time to determine if post-traumatic hemorrhagic shock has occurred. Delay window-the time difference between the prediction window and the learning window. Wherein T0 is the time point when the patient enters the ICU, T1 is set as the current time point, T1 is more than or equal to T0, x1 is the time length of the learning window, T1-x1 is more than or equal to T0, x2 is the delay window length, the time period of the prediction window is [ T1+x2, T1+x2+x3], and x3 is the prediction window length.
The physiological parameter sensing subsystem is used for monitoring conventional noninvasive physiological parameters of a patient in real time, collecting physiological parameter time sequence data in a patient conventional noninvasive physiological parameter and a patient learning window, the dynamic early warning subsystem is used for receiving the physiological parameter time sequence data in the patient conventional noninvasive physiological parameter and the patient learning window collected by the physiological parameter sensing subsystem, calculating the occurrence probability of post-traumatic shock of the patient in a prediction window range according to the conventional noninvasive physiological parameter collected by the physiological parameter sensing subsystem and the physiological parameter time sequence data in the patient learning window to obtain a prediction probability value, sending the prediction probability value to the post-traumatic shock morbidity judging subsystem, correcting the prediction probability value according to medical environment data of the position of the post-traumatic shock morbidity judging subsystem and a doctor treatment experience evaluation value, and judging that the post-traumatic shock dynamic early warning system judges that the patient will occur post-traumatic shock in the prediction window in the future according to the conventional noninvasive physiological parameter and the physiological parameter time sequence data in the patient learning window, and judging that the post-traumatic shock dynamic early warning system does not occur in the future according to the corrected probability value is smaller than 0.5. If the early warning model updating subsystem monitors the change of the medical environment, the subsystem updates the weight of the stack-type deep learning model in the dynamic early warning subsystem by using an incremental learning method according to the conventional non-invasive physiological parameters of the patient, the death condition of the patient, the post-traumatic hemorrhagic shock occurrence condition, the hospitalization duration and other clinical prognosis information acquired by the physiological parameter sensing subsystem, so that the early warning result of the system is adapted to the medical environment of the system. The stack type deep learning model comprises a convolution layer, a two-way long-short-term memory layer and a self-attention layer.
The physiological ventilation parameter sensing subsystem has the function of monitoring the noninvasive physiological parameters of the patient in real time, and can realize the real-time monitoring of parameters such as age、gender、BMI、、the status of Mechanical ventilation、Glasgow Coma Score(gcs)、gcs-verbal、gcs-motor、gcs-eyes、fiO2、PEEP、etco2、tidal volume、urineoutput、Heart rate、Respiration rate、temperature、SpO2、Non-invasive systolic blood pressure、Non-invasive diastolic blood pressure. The physiological ventilation parameter sensing subsystem is connected with a local hospital database or a medical staff to manually input data.
The dynamic early warning subsystem is combined with the conventional noninvasive physiological parameters of the patient in the learning window acquired by the physiological ventilation parameter sensing subsystem to calculate the original probability of the post-traumatic hemorrhagic shock symptom of the patient in the future in the prediction window.
The post-traumatic hemorrhagic shock morbidity judging subsystem corrects the original probability output by the dynamic early warning subsystem according to the local medical environment and doctor treatment experience, and feeds back to the medical staff early dynamic early warning system to judge whether the post-traumatic hemorrhagic shock symptom can appear in a prediction window in the future of the patient.
After the ICU patient is connected with the post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters, the physiological ventilation parameter sensing subsystem monitors physiological parameters of the patient in real time, and the parameters of the patient in the range of a learning window are transmitted to the dynamic early warning subsystem in real time.
Taking 1 hour as a minimum time unit, a plurality of effective sampling points exist in the span of one hour in the original data, the dynamic early warning subsystem calculates the total urination amount of patients in the hour for urination amount, calculates the median for other parameters, and then interpolates the missing data value.
The dynamic early warning subsystem interpolates the missing physiological parameter time sequence data in the minimum time unit by adopting a self-adaptive interpolation method, and the specific steps comprise:
s1, calculating a median Midparam and a quartile IQRIDPARAM of sampling frequencies of physiological parameters during the period from the admission to the discharge of each patient;
S2, selecting a physiological parameter with a missing value of a patient with missing acquired data, recording the physiological parameter as a parameter of the parameter, and selecting a time position of a null value or an abnormal value in the acquired parameter as a selected position according to a ascending sequence of recording time, wherein the abnormal value is data except 95% CI of the parameter in a database;
And S3, performing data interpolation on the selected position according to the following priority sequence, wherein firstly, the median of the effective values of the first (Midparam + IQRIDPARAM) sampling time lengths of the selected position is used as the data interpolation value of the selected position, secondly, the median of the effective values of the first (Midparam +2) sampling time lengths of the selected position is used as the data interpolation value of the selected position, thirdly, the median of the effective values of the parameters of the param is used as the data interpolation value of the selected position, and finally, the median of the parameters of the param recorded in the database is used as the data interpolation value of the selected position. For example, if a patient with ID 32572156 has a missing value at 8 hours after admission, the heart rate parameter of the patient is calculated to have a median of 2 hours as an effective value and an upper quartile as an effective value, and the median of the heart rate parameter effective values of the patient from (8-2-1) to 8 hours after admission is calculated to have an interpolation value of the missing value at 8 hours.
The interpolation data is not regarded as a valid point.
And the stack type deep learning model in the dynamic early warning subsystem obtains the original probability of the onset of hemorrhagic shock after the trauma of the patient in the prediction window by using the time sequence data of the physiological parameters of the patient in the learning window after cleaning.
The original probability of the post-traumatic hemorrhagic shock morbidity of the patient in the prediction window is led into a post-traumatic hemorrhagic shock morbidity judging subsystem, and the post-traumatic hemorrhagic shock morbidity judging subsystem obtains a classification threshold D of the stack type deep learning model according to the sensitivity and the specificity of the stack type deep learning model in the dynamic early warning subsystem on the test set data, so that when the classification threshold is D, the difference between the sensitivity and the specificity of the stack type deep learning model on the test set data is minimum.
If the corrected probability value of the incidence probability of post-traumatic hemorrhagic shock is larger than 0.5, the early dynamic early warning system judges that the post-traumatic hemorrhagic shock symptoms of the patient will appear in the range of the prediction window in the future, and otherwise judges that the post-traumatic hemorrhagic shock symptoms of the patient will not appear.
In the machine learning classification problem, the silence subsystem assumes that training data in a sample set is X i e X, i=1, 2, & n, label data in the sample set is Y i e Y, i=1, 2, & n, and defines a sample space Z i＝(xi,yi), i=1, 2, & n is an element in the sample space z=x×y. Generally, when a new object X n+1 is given, the machine learning model predicts that the label of X n+1 is Y n+1＝F(x1,x2,…,xn,xn+1,y1,y2,…,yn according to the rule between the sample X and the label Y in the space z=x×y).
The silence subsystem aims at evaluating test data x n+1 and early warning label dataFor the label with lower credibility, the silencing system inhibits the post-traumatic hemorrhagic shock onset judgment subsystem from outputting an early warning result. The parameter epsilon e (0, 1) reflecting the level of salience is set. The method for calculating the credibility between the object and the early warning label by the silencing subsystem is a consistency measurement method, namely the similarity degree of the test data x n+1 and the existing sample. At the silence subsystem, the degree of similarity is the minimum Euclidean distance of the test data from the reference sample set.
Taking a classification task as an example, first, a positive reference sample set in a reference sample set is definedAnd negative reference sample setAnd a consistency metric calculation function is set as F α, which is used for calculating the minimum Euclidean distance between the test data and the reference sample set. The minimum euclidean distance between the test data and the reference sample set refers to the minimum euclidean distance between the test data and the elements of the reference sample set.
The minimum distance of the reference sample set of test data x n+1 opposite to its corresponding predictive label is calculated and labeled as a n+1. The distance from the sample to the set is obtained by a Euclidean distance calculation method. Similarly calculating measurement values of a positive reference sample set and a negative reference sample set respectively as followsAnd (3) withThe metric value of the reference sample set is a set of distances from each element in the reference sample set to the set. If the test data x n+1 is early-warned as a positive sample by a model, sorting alpha i+1 in an ascending order in a positive sample measurement value set alpha pos to obtain a sorting sequence number result, and calculating the ratio of the sorting sequence number result to the total number of positive samples, wherein the calculation formula is as follows:
If p is smaller than the set threshold (e.g. 0.05), the silencing subsystem controls the model early warning result not to be output temporarily, and collects the model predicted probability output result within 15 minutes later to the silencing subsystem, and the silencing subsystem returns the maximum predicted probability output result within 15 minutes to the post-traumatic hemorrhagic shock morbidity judging subsystem.
For the situation that the clinical medical environment is changed, the ICU patient is connected with an interpretable noninvasive parameter-based noninvasive parameter post-traumatic hemorrhagic shock dynamic early warning system. The physiological ventilation parameter sensing subsystem monitors a physiological parameter of a patient in real time.
Taking an hour as a minimum time unit, a plurality of effective sampling points exist in the span of one hour in the original data, the dynamic early warning subsystem calculates the total urination amount of patients in the hour for urination amount, calculates the median for other parameters, and then interpolates the missing data value. The early warning model updating subsystem collects physiological parameters and prognosis information of a patient, or uses the input physiological parameters and prognosis information and combines an incremental training method to update the model parameters on the basis of not changing the structure of the stack type deep learning model, and transmits the physiological parameters of the patient in the range of a learning window to the dynamic early warning subsystem after updating the parameters of the stack type deep learning model in real time.
The dynamic early warning subsystem transmits patient data in the cleaned learning window to a trained machine learning model, and the original probability of hemorrhagic shock incidence after the patient is wounded in the prediction window can be obtained through model processing.
The original probability of post-traumatic hemorrhagic shock morbidity in the prediction window is led into a post-traumatic hemorrhagic shock morbidity judging subsystem, and the early warning performance of a machine learning model on test set data in the dynamic early warning subsystem and doctor experience are combined to obtain a classification threshold D, so that when the classification threshold is D, the sensitivity or the specificity of the model on the test set data is an expert advice value.
Therefore, the invention discloses a post-traumatic hemorrhagic shock dynamic early warning depth system based on noninvasive parameters, which only uses common noninvasive parameters and does not need laboratory data, so that the application range of the system is enlarged, the system can be used in remote areas, sudden public health events and first-line conditions of battlefields, the damage to individuals of patients caused by frequent collection of laboratory parameters is eliminated, and the use cost of the system is reduced. Compared with the traditional linear addition regression model, the automatic and flow intelligent dynamic early warning subsystem has higher computational complexity, is more suitable for the common nonlinear problem in the real problem, and can provide better post-traumatic hemorrhagic shock morbidity early warning capability. The pre-warning interval reserved for clinical intervention can provide sufficient time for a doctor to design a patient treatment plan. The analysis result of the system is consistent with the clinical research result, the model performance is excellent, the system can automatically update the model weight according to the environmental change so as to adapt to different clinical environments, and meanwhile, the false probability of the post-traumatic hemorrhagic shock morbidity prediction is effectively reduced.
The foregoing is merely exemplary of the present application and is not intended to limit the present application. Various modifications and variations of the present application will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. which come within the spirit and principles of the application are to be included in the scope of the claims of the present application.
Claims (5)
1. A dynamic early warning system for post-traumatic hemorrhagic shock based on noninvasive parameters comprises a physiological parameter sensing subsystem, a warning model updating subsystem, a dynamic warning subsystem, a post-traumatic hemorrhagic shock morbidity judging subsystem and a silencing subsystem;
The physiological parameter sensing subsystem is used for monitoring and collecting physiological parameter data information of a patient in real time, preprocessing the collected physiological parameter data of the patient, and sending the preprocessed data to the dynamic early warning subsystem;
The physiological parameter sensing subsystem is connected with the dynamic early warning subsystem, the dynamic early warning subsystem is respectively connected with the early warning model updating subsystem and the post-traumatic hemorrhagic shock morbidity judging subsystem, and the silencing subsystem is respectively connected with the post-traumatic hemorrhagic shock morbidity judging subsystem, the dynamic early warning subsystem and the early warning model updating subsystem;
The dynamic early warning subsystem is used for receiving and cleaning the conventional non-invasive physiological parameters of the patient and the physiological parameter time sequence data in the patient learning window, which are acquired by the physiological parameter sensing subsystem, calculating to obtain the original probability of post-traumatic shock incidence of the patient in the prediction time window, calculating the occurrence probability of post-traumatic shock of the patient in the future prediction time window according to the conventional non-invasive physiological parameters acquired by the physiological parameter sensing subsystem and the physiological parameter time sequence data in the patient learning window, obtaining a prediction probability value, transmitting the prediction probability value to the post-traumatic shock incidence judging subsystem, correcting the prediction probability value according to the medical environment data and the doctor treatment experience evaluation value of the post-traumatic shock incidence judging subsystem, judging that the post-traumatic shock incidence of the patient in the future prediction time window occurs according to the post-traumatic shock dynamic early warning system based on the non-invasive parameters if the corrected prediction probability value is larger than a first prediction probability threshold, and judging that the post-traumatic shock incidence of the patient does not occur in the future prediction time window if the corrected prediction probability value is smaller than a second prediction probability threshold;
The silence subsystem judges the difference among test data, training data and label data of the post-traumatic hemorrhagic shock dynamic early warning of the intelligent dynamic early warning subsystem, if the three types of data have significant differences, the silence subsystem controls the early warning result of the dynamic early warning subsystem not to be output, collects the early warning result data of the dynamic early warning subsystem in a period of time, extracts the early warning result data with the maximum prediction probability in the period of time, and outputs the early warning result data to the post-traumatic hemorrhagic shock morbidity judging subsystem;
The silence subsystem judges the difference of test data, label data and training data of the early warning model updating subsystem, and comprises the following steps:
Forming a sample set by using training data and a label data predicted value, dividing the sample set into a first reference sample set and a second reference sample set corresponding to the test data according to the label category to which each test data belongs, wherein the first reference sample set is a set formed by samples corresponding to the label category to which the test data belongs, the second reference sample set is a set formed by other samples except the first reference sample set in the sample set, calculating the distance between each sample in the first reference sample set and the reference sample set by using a consistency metric calculation function to obtain a first sample set distance value vector, calculating the distance between each test data and the first reference sample set and the second reference sample set corresponding to the test data by using a consistency metric calculation function, and correspondingly obtaining a first set distance value and a second set distance value of the test data;
The physiological parameter data information of the patient comprises conventional noninvasive physiological parameters of the patient and physiological parameter time sequence data in a learning window of the patient;
The preprocessing of the acquired patient physiological parameter data comprises the steps of constructing a sampling vector by using a discrete sampling value of the acquired patient physiological parameter, calculating a cross-correlation matrix of the sampling vector, processing the cross-correlation matrix by using a characteristic extraction method to obtain a characteristic value vector, and weighting the sampling vector by using the characteristic value vector to obtain a smooth value of the patient physiological parameter as preprocessed data.
2. The post-traumatic hemorrhagic shock dynamic early warning system based on non-invasive parameters according to claim 1,
The pretreatment of the collected patient physiological parameter data comprises the steps that a sampling vector formed by the patient physiological parameter discrete sampling values collected in a period of time is represented as [ x 1,x2,…,xN ], N is the number of the patient physiological parameter discrete sampling values collected in a period of time, a mutual matrix C of the sampling vector is obtained through calculation, and the characteristic value decomposition is carried out on the mutual matrix C to obtain:
C=VDVH,
And C is a characteristic vector matrix, D is a characteristic value matrix, diagonal elements of the matrix D are normalized and used as weight vectors, and the sampling vectors are weighted and summed to obtain smooth values of physiological parameters of the patient, and the smooth values are used as preprocessed data.
3. The early warning system for post-traumatic hemorrhagic shock dynamic early warning based on noninvasive parameters according to claim 1, wherein the early warning model updating subsystem monitors the medical environment of the system, and if the early warning model updating subsystem monitors the change of the medical environment of the system, the early warning model updating subsystem updates the weight of the deep learning model in the dynamic early warning subsystem by using an incremental learning method according to the conventional noninvasive physiological parameters of the patient and prognosis information including the death condition of the patient, the occurrence condition of post-traumatic hemorrhagic shock and the hospitalization duration acquired by the physiological parameter sensing subsystem, so that the early warning result of the system is adapted to the medical environment of the system.
4. The post-traumatic hemorrhagic shock dynamic early warning system based on non-invasive parameters according to claim 1,
The silence subsystem judges the difference of test data, label data and training data of the early warning model updating subsystem, and comprises the following steps:
The silence subsystem predicts and obtains a label data predicted value by utilizing training data and adopting a deep learning model, calculates the credibility among the test data, the training data and the label data predicted value by adopting a credibility evaluation method, and judges that the early warning model updates the early warning data of the subsystem and the obvious difference among the label data and the training data when the credibility is lower than a credibility threshold value.
5. The post-traumatic hemorrhagic shock dynamic early warning system based on non-invasive parameters according to claim 1,
The silence subsystem judges the difference of test data, label data and training data of the early warning model updating subsystem, and comprises the following steps:
regarding the data of an early warning model updating subsystem as a stable random process, respectively establishing corresponding autoregressive-moving average models, namely ARMA models, aiming at test data, label data and training data to respectively obtain a first ARMA model, a second ARMA model and a third ARMA model, calculating a cross-correlation matrix of the coefficients of the three ARMA models, calculating the cross-correlation matrix to obtain a maximum characteristic value, and judging the difference of the test data, the label data and the training data by using the maximum characteristic value;
When the maximum characteristic value is larger than the difference judging threshold value, the significance difference among the test data, the label data and the training data of the early warning model updating subsystem is judged.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210392852.4A CN114758786B (en) | 2022-04-14 | 2022-04-14 | Post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210392852.4A CN114758786B (en) | 2022-04-14 | 2022-04-14 | Post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114758786A CN114758786A (en) | 2022-07-15 |
| CN114758786B true CN114758786B (en) | 2025-01-17 |
Family
ID=82331306
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210392852.4A Active CN114758786B (en) | 2022-04-14 | 2022-04-14 | Post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114758786B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116258276A (en) * | 2023-05-06 | 2023-06-13 | 中国能源建设集团湖南省电力设计院有限公司 | Carbon emission total amount prediction method based on saturated carbon emission density |
| CN116978570B (en) * | 2023-09-25 | 2024-02-06 | 之江实验室 | An online real-time patient criticality assessment and vital sign parameter prediction system |
| CN119541859A (en) * | 2024-11-15 | 2025-02-28 | 中国人民解放军总医院第二医学中心 | A tracking and analysis system for disabled patients based on smart chips |
| CN120473065B (en) * | 2025-07-15 | 2025-10-31 | 天津工业大学 | Methods, devices, equipment, storage media and products for improving alarm accuracy |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104755019A (en) * | 2012-10-30 | 2015-07-01 | 德克斯康公司 | Systems and methods for providing sensitive and specific alerts |
| CN113421654A (en) * | 2021-07-12 | 2021-09-21 | 军事科学院系统工程研究院卫勤保障技术研究所 | Dynamic early warning and deep learning system for post-traumatic hemorrhagic shock |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9750463B2 (en) * | 2013-12-10 | 2017-09-05 | General Electric Company | Respiratory stress detection |
| CN111063162A (en) * | 2019-12-05 | 2020-04-24 | 恒大新能源汽车科技(广东)有限公司 | Silent alarm method and device, computer equipment and storage medium |
-
2022
- 2022-04-14 CN CN202210392852.4A patent/CN114758786B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104755019A (en) * | 2012-10-30 | 2015-07-01 | 德克斯康公司 | Systems and methods for providing sensitive and specific alerts |
| CN113421654A (en) * | 2021-07-12 | 2021-09-21 | 军事科学院系统工程研究院卫勤保障技术研究所 | Dynamic early warning and deep learning system for post-traumatic hemorrhagic shock |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114758786A (en) | 2022-07-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114758786B (en) | Post-traumatic hemorrhagic shock dynamic early warning system based on noninvasive parameters | |
| US11217349B2 (en) | System and method for processing human related data including physiological signals to make context aware decisions with distributed machine learning at edge and cloud | |
| CN112365978B (en) | Method and device for establishing a model for early risk assessment of tachycardia events | |
| JP5841196B2 (en) | Residue-based management of human health | |
| CN113421654A (en) | Dynamic early warning and deep learning system for post-traumatic hemorrhagic shock | |
| CN118039166B (en) | Health monitoring management method based on medical big data | |
| CN118969295B (en) | Early warning system and method for high-risk pregnant women | |
| US20210345915A1 (en) | Methods Circuits Devices Systems and Machine Executable Code for Glucose Monitoring Analysis and Remedy | |
| CN119442124B (en) | A method, apparatus, device, and storage medium for identifying abnormal electrocardiogram (ECG) data based on deep learning. | |
| CN119028512A (en) | A real-time obstetric nursing optimization method based on big data | |
| CN118280607A (en) | Intelligent critical index monitoring method and system based on clinical data | |
| CN118072971B (en) | Smart elderly care service management method based on deep information integration | |
| CN119791621A (en) | A multifunctional vital signs monitoring method and system | |
| CN119763828A (en) | Digital management method of graded nursing care in intensive care units based on the Internet of Things | |
| Kannathal et al. | Cardiac health diagnosis using data fusion of cardiovascular and haemodynamic signals | |
| CN116098595B (en) | System and method for monitoring and preventing sudden cardiac death and sudden cerebral death | |
| CN117653053A (en) | A way to predict health risks with smartwatches | |
| CN117037989A (en) | A health data management method and system for smart communities | |
| CN115775627A (en) | Diabetes early warning method, equipment and system | |
| CN120032885A (en) | Intelligent nursing risk assessment and early warning system and method | |
| CN119969970A (en) | Pregnancy health care control system based on data feedback analysis | |
| CN120340845A (en) | An intelligent health management system | |
| CN119517441A (en) | Emergency and critical care auxiliary assessment system based on machine learning | |
| CN121011359B (en) | A Multidimensional Prediction Method for Animal Health Status in Smart Ranches | |
| CN120526958B (en) | Medical monitoring methods, systems, storage media and medical devices |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |