A method, apparatus, storage medium, and electronic device for identifying chronic stress.

Abstract

Some embodiments of this application provide a method, apparatus, storage medium, and electronic device for identifying chronic stress. The method includes: acquiring multiple sleep stages of a target object during the night, wherein each stage of sleep data includes: dynamic electrocardiogram data and human three-dimensional acceleration data; inputting the multiple sleep stages of sleep data into multiple target classifiers to obtain multiple output results, wherein one stage of sleep data corresponds to one target classifier, and each output result indicates either the presence or absence of chronic stress; and statistically analyzing the multiple output results to obtain a chronic stress identification result for the target object. Some embodiments of this application can detect whether a target object is experiencing chronic stress with high accuracy.

Description

Technical Field

[0001] This application relates to the field of pressure detection technology, and more specifically, to a method, apparatus, storage medium, and electronic device for identifying chronic pressure. Background Technology

[0002] As society continues to develop, people face pressure from various sources. Stress, especially chronic stress, is a high-risk factor for many physical and mental health problems.

[0003] Currently, stress levels can be identified to prevent mental and physical health problems. Typically, stress detection involves collecting physical data from the individual during the day and analyzing this data to determine if chronic stress is present. However, because daily activities vary and can influence physical data, the accuracy of the collected data is low, thus affecting the accuracy of chronic stress identification.

[0004] Therefore, how to provide a technical solution for identifying chronic stress with high accuracy has become an urgent technical problem to be solved. Summary of the Invention

[0005] The purpose of some embodiments of this application is to provide a method, apparatus, storage medium and electronic device for identifying chronic stress. The technical solutions of the embodiments of this application can effectively identify whether a target object has chronic stress, with high identification efficiency and accurate identification results.

[0006] In a first aspect, some embodiments of this application provide a method for identifying chronic stress, comprising: acquiring multiple sleep stages of a target object at night, wherein each stage of sleep data includes: dynamic electrocardiogram data and human three-dimensional acceleration data; inputting the multiple stages of sleep data into multiple target classifiers to obtain multiple output results, wherein one stage of sleep data corresponds to one target classifier, and each output result indicates whether chronic stress is present or not; and statistically analyzing the multiple output results to obtain a chronic stress identification result for the target object.

[0007] Some embodiments of this application input sleep data of the target object at multiple nighttime stages into a target classifier corresponding to each stage, obtaining multiple output results. Finally, the multiple output results are statistically analyzed to obtain the chronic stress identification result. Embodiments of this application can collect sleep data from multiple nighttime stages with minimal disruption to the body, achieving effective identification of whether the target object experiences chronic stress, with high identification efficiency and accurate results.

[0008] In some embodiments, obtaining multiple sleep stages of a target object at night includes: obtaining the target object's nighttime sleep data; and dividing the nighttime sleep data according to a preset sleep time threshold to obtain the multiple sleep stages.

[0009] Some embodiments of this application divide the nighttime sleep data of the target object into multiple stages of sleep data, which can provide effective data support for subsequent accurate identification of whether the target object has chronic stress.

[0010] In some embodiments, obtaining the nighttime sleep data of the target object includes: obtaining the target object's raw nighttime electrocardiogram data and raw human three-dimensional acceleration data; and filtering the raw nighttime electrocardiogram data and the raw human three-dimensional acceleration data to obtain the nighttime sleep data.

[0011] Some embodiments of this application obtain nighttime sleep data by filtering the original nighttime electrocardiogram data and original human three-dimensional acceleration data of the target object. This can effectively remove noise interference in the original data and ensure the accuracy of identifying whether the target object has chronic stress.

[0012] In some embodiments, the step of statistically analyzing the multiple output results to obtain the chronic stress identification result of the target object includes: counting the number of results indicating chronic stress and the number of results indicating no chronic stress among the multiple output results; and obtaining the chronic stress identification result based on the number of results indicating chronic stress and the number of results indicating no chronic stress.

[0013] Some embodiments of this application obtain chronic stress identification results by statistically analyzing the number of results with or without chronic stress in multiple output results, which can ensure the accuracy of the identification results.

[0014] In some embodiments, obtaining the chronic stress identification result based on the number of results with chronic stress and the number of results without chronic stress includes: if the number of results with chronic stress is not equal to the number of results without chronic stress, then the output result corresponding to the larger value of the number of results with chronic stress and the number of results without chronic stress is taken as the chronic stress identification result; or, if the number of results with chronic stress is equal to the number of results without chronic stress, then the last output result among the plurality of output results is removed to obtain the removed chronic stress result data and the non-chronic stress result data; the output result corresponding to the larger value of the removed chronic stress result data and the non-chronic stress result data is taken as the chronic stress identification result.

[0015] Some embodiments of this application select or eliminate results with low impact by considering the number of results with or without chronic stress, thereby obtaining the final chronic stress identification result and effectively ensuring the accuracy of the identification result.

[0016] In some embodiments, before inputting the multiple sleep stages into multiple target classifiers, the method further includes: acquiring training datasets and validation datasets for multiple stages, wherein both the training dataset and the validation dataset include: heart rate data of multiple sample objects at night, human body three-dimensional acceleration high-pass filtered data, and stress labels, the stress labels including: chronic stress labels and no chronic stress labels; training an initial classifier using the training datasets of each stage in the training dataset of the multiple stages to obtain a classifier to be validated for each stage; and validating the classifier to be validated for each stage using the validation datasets of each stage in the validation dataset of the multiple stages to obtain the multiple target classifiers.

[0017] Some embodiments of this application train and validate an initial classifier using a training dataset and a validation dataset composed of nighttime sleep data from multiple sample objects, resulting in multiple target classifiers corresponding to multiple stages. These embodiments can obtain target classifiers with high accuracy, providing a model foundation for subsequently obtaining accurate chronic stress identification results.

[0018] In some embodiments, before acquiring the training and validation datasets for multiple stages, the method further includes: during the nighttime stage, collecting raw heartbeat data, raw human acceleration data, and the pressure labels of the multiple sample objects; preprocessing the raw heartbeat data and the raw human acceleration data to obtain heartbeat data and human three-dimensional acceleration high-pass filtered data of the multiple sample objects; and constructing a total training sample dataset and a total validation sample dataset based on the heartbeat data, human three-dimensional acceleration high-pass filtered data, and the pressure labels of the multiple sample objects.

[0019] Some embodiments of this application construct a total training sample dataset and a total validation sample dataset by collecting raw data and stress labels from multiple sample objects at night, and then preprocessing the raw data to obtain heart rate data and human body 3D acceleration high-pass filtered data from the multiple sample objects. These embodiments can effectively remove interfering noise data, obtaining highly accurate training and validation sample data, providing rich and accurate data support for training the classifier.

[0020] In some embodiments, obtaining training and validation datasets for multiple phases includes: dividing the nighttime phase into multiple time phases in chronological order; and dividing the total training sample dataset and the total validation sample dataset into the multiple time phases respectively to obtain training and validation datasets for the multiple phases.

[0021] Some embodiments of this application divide the total training sample dataset and the total validation sample dataset according to time stages, thereby obtaining sleep data under different sleep stages, providing rich and accurate data support for training classifiers for multiple stages.

[0022] Secondly, some embodiments of this application provide an apparatus for identifying chronic stress, comprising: a data acquisition module for acquiring multiple sleep stages of a target object at night, wherein each stage of sleep data includes: dynamic electrocardiogram data and human three-dimensional acceleration data; a data processing module for inputting the multiple sleep stages of sleep data into multiple target classifiers to obtain multiple output results, wherein one stage of sleep data corresponds to one target classifier, and each output result indicates whether chronic stress is present or not; and a result output module for statistically analyzing the multiple output results to obtain the chronic stress identification result of the target object.

[0023] Thirdly, some embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can implement the method described in any embodiment of the first aspect.

[0024] Fourthly, some embodiments of this application provide an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, can implement the method as described in any embodiment of the first aspect.

[0025] Fifthly, some embodiments of this application provide a computer program product, the computer program product including a computer program, wherein the computer program, when executed by a processor, can implement the method described in any embodiment of the first aspect. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of some embodiments of this application, the accompanying drawings used in some embodiments of this application will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 A system diagram for identifying chronic stress is provided for some embodiments of this application;

[0028] Figure 2 Flowcharts of methods for obtaining target classifiers provided for some embodiments of this application;

[0029] Figure 3 Raw ECG data plots and raw T-ACC data plots are provided for some embodiments of this application;

[0030] Figure 4 ECG plots and AT-ACC time series plots of processing results provided for some embodiments of this application;

[0031] Figure 5 RR interval sequence diagrams provided for some embodiments of this application;

[0032] Figure 6 Processed RR interval sequence diagrams provided for some embodiments of this application;

[0033] Figure 7 ECG data graphs and AT-ACC filtered data diagrams provided for some embodiments of this application;

[0034] Figure 8 One of the schematic diagrams illustrating the changes in accuracy and loss rate of the target classifier provided for some embodiments of this application;

[0035] Figure 9 A second schematic diagram illustrating the changes in accuracy and loss rate of the target classifier provided for some embodiments of this application;

[0036] Figure 10 The third schematic diagram illustrating the changes in accuracy and loss rate of the target classifier provided for some embodiments of this application;

[0037] Figure 11 A flowchart of a method for identifying chronic stress is provided for some embodiments of this application;

[0038] Figure 12 One of the state diagrams of the data acquisition object provided for some embodiments of this application;

[0039] Figure 13 A second schematic diagram illustrating the state of a data acquisition object provided for some embodiments of this application;

[0040] Figure 14 The third schematic diagram of the state of the data acquisition object provided for some embodiments of this application;

[0041] Figure 15A block diagram of a device for identifying chronic stress is provided for some embodiments of this application;

[0042] Figure 16 A schematic diagram of an electronic device provided for some embodiments of this application. Detailed Implementation

[0043] The technical solutions of some embodiments of this application will now be described with reference to the accompanying drawings.

[0044] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0045] To facilitate understanding, let's first explain the professional terms used in the text:

[0046] POMS, or Profile of Mood States, developed by McNair, is a scale for rating mood states.

[0047] ECG, Electrocardiogram, is a technique that uses an electrocardiograph to record the changes in electrical activity of the heart during each cardiac cycle from the body surface.

[0048] T-ACC, triple acceleration data, can reflect the body's physical condition.

[0049] In related technologies, with the increasingly fast pace of social development, people experience various pressures in their daily work and life. To protect individual physical and mental health, existing technologies can identify chronic stress states. Once chronic stress is detected, timely guidance and stress relief can be provided to the target individual. In existing technologies, when identifying whether an individual has chronic stress, data collection is typically done by collecting a person's physical data during daytime activities for convenience. Analysis of this daytime data is then used to confirm the presence of chronic stress. However, because daytime activities are complex, and the impact of each activity on physical data varies, analyzing this data alone cannot yield accurate results and has poor reliability. Therefore, as can be seen from the above related technologies, the accuracy of existing technologies in identifying chronic stress in target individuals is low.

[0050] In view of this, some embodiments of this application provide a method for identifying chronic stress. This method involves inputting sleep data of a target object at multiple stages during the night into multiple trained target classifiers to obtain multiple output results. Finally, a chronic stress identification result for the target object is obtained based on these multiple output results. Some embodiments of this application can effectively eliminate the influence of various activities on sleep data by acquiring the target object's sleep data at night. Dividing the sleep data into three stages and inputting them into the target classifier corresponding to each stage allows for effective identification of the target object at different stages. Finally, obtaining the final chronic stress identification result based on multiple output results ensures the accuracy and reliability of the identification result, thereby achieving accurate identification of the target object's health status.

[0051] The following is in conjunction with the appendix Figure 1 This application provides an exemplary embodiment of a system structure for identifying chronic stress.

[0052] like Figure 1 As shown, some embodiments of this application provide a system for identifying chronic stress, which includes a terminal 100 and a server 200. The terminal 100 can collect raw nighttime electrocardiogram (ECG) data and raw human 3D acceleration data of a target object at night (e.g., nighttime can be from 0:00 AM to 7:00 AM daily). The server 200 can then process the raw nighttime ECG and raw human 3D acceleration data to obtain nighttime sleep data. The server 200 then divides the nighttime sleep data into three stages of sleep data (as a specific example of multiple stages of sleep data). Each stage of sleep data can include multiple sets of data. Finally, the three stages of sleep data are input into three pre-trained target classifiers (as a specific example of multiple target classifiers), obtaining multiple output results corresponding to the multiple sets of data. Finally, the multiple output results are statistically analyzed to obtain the chronic stress identification result of the target object.

[0053] In some embodiments of this application, terminal 100 may be a mobile terminal or a non-portable computer terminal. This application does not impose specific limitations on it.

[0054] In some other embodiments of this application, if the terminal 100 has the function of collecting raw nighttime data of the target object, and can process the raw data to obtain multiple stages of sleep data, and input the multiple stages of sleep data into multiple target classifiers to obtain the chronic stress identification result of the target object, then the server 200 may not be set up.

[0055] It should be noted that in some embodiments of this application, before performing chronic stress identification on the target object, an initial classifier needs to be trained to obtain a target classifier. The following is a detailed explanation in conjunction with the appendix. Figure 2 The implementation process of obtaining a target classifier provided by some embodiments of this application is illustrated by way of example.

[0056] Please see the appendix Figure 2 , Figure 2 A flowchart of a method for obtaining a target classifier is provided for some embodiments of this application. The method includes:

[0057] S210, acquire training datasets and validation datasets for multiple stages, wherein the training dataset and the validation dataset each include: heart rate data of multiple sample objects at night, human body three-dimensional acceleration high-pass filtered data, and stress labels, wherein the stress labels include: chronic stress labels and no chronic stress labels.

[0058] The implementation process of S210 is illustrated below.

[0059] In some embodiments of this application, S210 may include:

[0060] S211, During the nighttime phase, raw heart rate data, raw human acceleration data, and pressure labels of the multiple sample objects are collected.

[0061] For example, in some embodiments of this application, data collection needs to be conducted during the nighttime phase to obtain accurate data on the target subject. The nighttime phase can refer to the period from midnight to 7 a.m., or from midnight to 6 a.m., etc. The nighttime phase can be set according to the sleep patterns of each sample subject, but this embodiment does not involve this. Terminal 100 can collect raw ECG data (as a specific example of raw heart rate data) and raw T-ACC data (as a specific example of raw human acceleration data) of the sample subject during the nighttime phase. Stress labels can be obtained by assessing the sample subject through questionnaires or other methods. For example, stress labels can be determined not only through questionnaires but also through other evidence. Examples include university exam schedules, observations of subjects (i.e., the sample subject) engaging in stressful work (e.g., preparing for a thesis defense), and increased heart rate during planned academic tasks. Furthermore, subjects are only labeled as having chronic stress if the effects of their stressful tasks persist for more than a week. For subjects without chronic stress, their status as being without chronic stress was also assessed using psychometric measurements (e.g., POMS < 100), stressor measurements (e.g., no perceived stressors in the past two weeks), and other objective evidence (e.g., university semester schedule). Because chronic stress lasts a long time, stress-labeled data collection in this application was conducted over several days to improve data collection accuracy. Therefore, the data obtained in this application record richer information about the physiological responses to chronic stress than data collected over a shorter period. Furthermore, the chronic stress data were obtained while subjects were lying down and sleeping, which is far less disruptive to movement and activity compared to data obtained during the day when subjects were awake and engaged in various activities.

[0062] S212, preprocess the original heartbeat data and the original human acceleration data to obtain the heartbeat data and human three-dimensional acceleration high-pass filtered data of the multiple sample objects.

[0063] For example, in some embodiments of this application, after loading such as Figure 3 The raw ECG data shown (i.e. Figure 3 The subgraph (1) and the original T-ACC data (i.e. Figure 3 After subgraph (2) is processed, the following preprocessing operations are performed on it:

[0064] 1) Wavelet decomposition and reconstruction are used to remove the original ECG baseline drift, resulting in the following: Figure 4 The processed neutron map (3) shown is the ECG signal image after baseline drift removal. Low-frequency removal is performed on the original T-ACC data, and AT-ACC is calculated to obtain the following results: Figure 4 The processing results are shown in neutron diagram (4).

[0065] 2) Automatic calculation using an adaptive length running window algorithm Figure 4 The RR interval sequence is obtained as follows: Figure 5 The diagram shows the RR interval sequence.

[0066] 3) Correction Figure 5 The sparse RR interval errors caused by random noise are removed, and RR interval errors caused by device updates and body movements exceeding 10 seconds are removed, resulting in the following: Figure 6 The results are shown in the figure.

[0067] 4) Figure 6 The RR intervals were converted to heart rate per minute, and heart rate data with obvious motion interference were removed according to the rules of AT-ACC < -1.5 and AT-ACC > 1.5, resulting in the following: Figure 7 The ECG data plot shown (as a specific example of heart rate data), namely the data without obvious motion interference and the data with obvious motion interference in subplot (7), and Figure 7 The AT-ACC filtered data shown in subgraph (8) is the triple acceleration mean (as a specific example of human three-dimensional acceleration high-pass filtered data).

[0068] S213, Based on the heartbeat data and human body three-dimensional acceleration high-pass filtered data of the multiple sample objects and the pressure label, construct the total training sample dataset and the total validation sample dataset.

[0069] For example, in some embodiments of this application, labeled datasets for chronic stress and non-chronic stress are constructed using given inclusion and exclusion rules. For instance, using the inclusion and exclusion rules in Table 1, 712 heart rate interval time series samples from 45 subjects over 142 days were obtained, representing chronic stress. Of these, 212 heart rate interval time series samples from 8 subjects over 41 days were used as the validation dataset, and the remaining 500 samples were used as the training dataset. Similarly, 787 heart rate interval time series samples from 41 subjects over 136 days were obtained without chronic stress, of which 240 heart rate interval time series samples from 8 subjects over 40 days were used as the validation dataset, and the remaining 547 samples were used as the training dataset. The total training dataset and the total validation dataset can be obtained in this way. The total training dataset includes 500 samples with chronic stress and 547 samples without chronic stress. The total validation dataset includes 45 samples with chronic stress and 8 samples without chronic stress.

[0070] Table 1

[0071]

[0072] S214, the nighttime phase is divided into multiple time phases in chronological order;

[0073] For example, in some embodiments of this application, the data can be divided into three time stages based on the sampling time, such as the first hour, second hour, etc. (as a specific example of multiple time stages). Alternatively, it can be divided into three groups based on sleep state (as another specific example of multiple time stages), namely, the first group belongs to the early stage of sleep, the second group belongs to the middle stage of sleep, and the third group belongs to the late stage of sleep.

[0074] S215, the total training sample dataset and the total validation sample dataset are divided according to the multiple time stages to obtain the training dataset and validation dataset for the multiple stages.

[0075] For example, in some embodiments of this application, the data in the total training sample dataset and the total validation sample dataset are divided into the three time stages mentioned above to obtain training datasets and validation datasets for three stages (as a specific example of training datasets and validation datasets for multiple stages).

[0076] S220, the initial classifier is trained using the training datasets of each stage in the training dataset of the multiple stages to obtain the classifiers to be verified for each stage.

[0077] For example, in some embodiments of this application, initial classifiers for each of the three stages are trained using training datasets from three stages, resulting in classifiers to be validated for each stage. The initial classifier can be a CNN+SENet deep learning neural network classifier. It should be noted that the type of initial classifier can be selected based on actual circumstances, and the embodiments of this application are not limited to this.

[0078] S230, the classifiers to be verified in each stage are verified using the verification datasets of each stage in the multiple stages of the verification dataset, thereby obtaining the multiple target classifiers.

[0079] For example, in some embodiments of this application, three-stage validation datasets are used to validate the respective classifiers to be validated, and three target classifiers are obtained after the validation is passed (as a specific example of multiple target classifiers).

[0080] As a specific example of this application, for instance, a CNN+SENet deep learning neural network classifier is trained and validated using the first set of training and validation datasets from the early stages of sleep, as shown below. Figure 8The diagram illustrates the changes in accuracy and loss rate. The CNN+SENet deep learning neural network classifier was trained and validated using the training and validation datasets from the second group of early sleep periods, as shown below. Figure 9 The diagram illustrates the changes in accuracy and loss rate. The CNN+SENet deep learning neural network classifier was trained and validated using the training and validation datasets from the third group of early sleep periods, as shown below. Figure 10 The diagram shows the changes in accuracy and loss rate.

[0081] As can be seen from the above embodiments of this application, the target classifier model of this application has high accuracy.

[0082] The following is in conjunction with the appendix Figure 11 The present application provides an exemplary embodiment of the process for identifying chronic stress, executed by server 200.

[0083] Please see the appendix Figure 11 , Figure 11 A flowchart of a method for identifying chronic stress is provided for some embodiments of this application. The method for identifying chronic stress includes:

[0084] S1110, acquire multiple sleep data of the target object at night, wherein each stage of sleep data includes: dynamic electrocardiogram data and human three-dimensional acceleration data.

[0085] For example, in some embodiments of this application, it is assumed that a person (as a specific example of the target object) has N1, N2, and N3 samples during three sleep periods on a certain night (i.e., early sleep, middle sleep, and late sleep). Each of the N1, N2, and N3 samples (as a specific example of multiple sleep stages, with each sample representing one sleep stage) includes multiple sets of dynamic ECG data and human T-ACC data (as a specific example of dynamic electrocardiogram data and human three-dimensional acceleration data). For example, sample N1 includes 10 sets of dynamic ECG data and human T-ACC data, sample N2 includes 5 sets of dynamic ECG data and human T-ACC data, and sample N3 includes 3 sets of dynamic ECG data and human T-ACC data.

[0086] In some embodiments of this application, S1110 may include: acquiring the nighttime sleep data of the target object; dividing the nighttime sleep data according to a preset sleep time threshold to obtain the multiple stages of sleep data.

[0087] For example, in some embodiments of this application, according to Figure 2The method provided in the Chinese embodiment divides a person's nighttime sleep data into three sleep periods: early sleep, middle sleep, and late sleep. Then, N1, N2, and N3 samples are obtained based on these three sleep periods.

[0088] In some embodiments of this application, S1110 may include: acquiring the original nighttime electrocardiogram data and the original human three-dimensional acceleration data of the target object; and filtering the original nighttime electrocardiogram data and the original human three-dimensional acceleration data to obtain the nighttime sleep data.

[0089] For example, in some embodiments of this application, since raw nighttime ECG data (as a specific example of raw nighttime electrocardiogram data) and raw T-ACC data (as a specific example of raw human three-dimensional acceleration data) can be collected non-invasively using existing wearable devices connected to terminal 100, high signal-to-noise ratio data can be easily obtained through changes in sleep state, reducing the preprocessing requirements for raw nighttime ECG data and raw T-ACC data, thereby improving recognition efficiency. It should be noted that the filtering processing of raw nighttime ECG data and raw T-ACC data can refer to... Figure 2 The specific processes of processing the original data of the sample object in steps 1) to 4) of the provided method embodiments will not be repeated here to avoid repetition.

[0090] S1120, the multiple stages of sleep data are input into multiple target classifiers to obtain multiple output results, wherein one stage of sleep data corresponds to one target classifier, and each output result indicates whether there is chronic stress or no chronic stress.

[0091] For example, in some embodiments of this application, N1, N2, and N3 samples are respectively input into corresponding target classifiers. Ten sets of dynamic ECG data and human T-ACC data from sample N1 are input into the target classifier corresponding to early sleep, yielding 10 output results. Five sets of dynamic ECG data and human T-ACC data from sample N2 are input into the target classifier corresponding to mid-sleep, yielding 5 output results. Three sets of dynamic ECG data and human T-ACC data from sample N3 are input into the target classifier corresponding to late sleep, yielding 3 output results. Finally, 18 output results are obtained from samples N1, N2, and N3 (as a specific example of multiple output results). Each of the 18 output results indicates either chronic stress or no chronic stress.

[0092] S1130, Statistically analyze the multiple output results to obtain the chronic stress identification result of the target object.

[0093] In some embodiments of this application, S1130 may include: counting the number of results with chronic stress and the number of results without chronic stress among the plurality of output results; and obtaining the chronic stress identification result based on the number of results with chronic stress and the number of results without chronic stress.

[0094] For example, in some embodiments of this application, out of 18 output results, 10 votes indicate chronic stress and 8 votes indicate no chronic stress. Alternatively, out of 18 output results, 9 votes indicate chronic stress and 9 votes indicate no chronic stress. The chronic stress identification result is determined based on the number of votes.

[0095] In some embodiments of this application, S1130 may include: if the number of results with chronic stress is not equal to the number of results without chronic stress, then the output result corresponding to the larger value between the number of results with chronic stress and the number of results without chronic stress shall be used as the chronic stress identification result.

[0096] For example, in some embodiments of this application, 10 out of 18 output results indicate chronic stress and 8 out of 8 results indicate no chronic stress, thus the chronic stress identification result for a person can be determined as having chronic stress.

[0097] In some other embodiments of this application, S1130 may include: if the number of results with chronic stress is equal to the number of results without chronic stress, then the last output result among the plurality of output results is removed to obtain the removed data of results with chronic stress and the data of results without chronic stress; the output result corresponding to the larger value among the removed data of results with chronic stress and the data of results without chronic stress is taken as the chronic stress identification result.

[0098] For example, in some other embodiments of this application, out of 18 output results, 9 votes indicate chronic stress and 9 votes indicate no chronic stress, which is an equal number. In this case, the output result of the latest set of data in the N3 samples (i.e., the last output result) is removed. For example, if the output result of the latest set of data indicates chronic stress, after removal, there are 8 votes for chronic stress and 9 votes for no chronic stress, and the final chronic stress identification result for a person is no chronic stress.

[0099] In addition, some embodiments of this application have statistically analyzed the parameters of the three target classifiers, as shown in Table 2. The experimental data in Table 2 demonstrates that the target classifier can achieve a recognition accuracy of 81%.

[0100] Table 2

[0101]

[0102] Furthermore, in some embodiments of this application, an early warning can be issued regarding whether a target individual is likely to develop a chronic stress state. By continuously collecting sleep data from multiple stages of the target individual over a preset number of days, and then inputting this data into multiple target classifiers, daily chronic stress identification results are obtained, thereby providing an early warning regarding whether the target individual is likely to develop a chronic stress state.

[0103] As a specific example of this application, a data collection subject (i.e., the target subject) had an argument with a close relative at approximately 11:30 PM. This stressor caused the data collection subject to experience frequent premature ventricular contractions (PVCs) after falling asleep, resulting in a series of abnormal heart rhythm variability events. Figure 11 The provided method embodiment identifies daily chronic stress in the target object, and obtains the following results: Figure 12 The diagram illustrates an abnormal state, suggesting that the data collection subject may develop a chronic stress state. Subsequent monitoring showed that the abnormal heart rhythm variability largely returned to normal after 26 days. Figure 13 As shown.

[0104] In some embodiments of this application, it is also possible to provide early warning of extreme stress responses during sleep that have developed into chronic stress. For example... Figure 14 As shown, the data collection subject in this example experienced frequent cardiac arrhythmia variability during sleep due to months of continuous exam preparation. This manifested as an irregular heartbeat, with periods of rapid and slow heart rate, particularly slow periods with approximately four seconds of no heartbeat. By using [method / approach] on this data collection subject... Figure 11 If the abnormal heart rhythm variability does not return to normal after 3 months, an early warning can be given that the data collection subject has developed chronic stress.

[0105] Please refer to Figure 15 , Figure 15 The diagram illustrates a block diagram of a device for identifying chronic stress according to some embodiments of this application. It should be understood that this device for identifying chronic stress corresponds to the method embodiments described above and is capable of performing the various steps involved in the method embodiments. The specific functions of this device for identifying chronic stress can be found in the description above; detailed descriptions are omitted here to avoid repetition.

[0106] Figure 15The device for identifying chronic stress includes at least one software functional module that can be stored in a memory or embedded in the device in the form of software or firmware. The device includes: a data acquisition module 1510 for acquiring multiple sleep stages of a target object at night, wherein each stage of sleep data includes: dynamic electrocardiogram data and human three-dimensional acceleration data; a data processing module 1520 for inputting the multiple sleep stages of sleep data into multiple target classifiers to obtain multiple output results, wherein one stage of sleep data corresponds to one target classifier, and each output result indicates whether chronic stress is present or not; and a result output module 1530 for statistically analyzing the multiple output results to obtain the chronic stress identification result of the target object.

[0107] Those skilled in the art will understand that the device for identifying chronic stress described above can perform the various steps in the above method embodiments, and will not be repeated here to avoid repetition.

[0108] Some embodiments of this application also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can perform the operation of any of the methods corresponding to the methods provided in the above embodiments.

[0109] Some embodiments of this application also provide a computer program product, which includes a computer program, wherein when the computer program is executed by a processor, it can implement the operation of any of the methods corresponding to the above embodiments provided in the above embodiments.

[0110] like Figure 16 As shown, some embodiments of this application provide an electronic device 1600, which includes a memory 1610, a processor 1620, and a computer program stored in the memory 1610 and executable on the processor 1620. When the processor 1620 reads the program from the memory 1610 via a bus 1630 and executes the program, it can implement the methods of any of the above embodiments.

[0111] Processor 1620 can process digital signals and can include various computing architectures. Examples include complex instruction set computer architectures, reduced instruction set computer architectures, or architectures that implement multiple instruction set combinations. In some examples, processor 1620 can be a microprocessor.

[0112] The memory 1610 can be used to store instructions executed by the processor 1620 or data related to the execution of instructions. These instructions and / or data may include code for implementing some or all of the functions of one or more modules described in the embodiments of this application. The processor 1620 of this disclosure embodiment can be used to execute the instructions in the memory 1610 to implement the methods shown above. The memory 1610 includes dynamic random access memory, static random access memory, flash memory, optical memory, or other memories well known to those skilled in the art.

[0113] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application. It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0114] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0115] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A method for identifying chronic stress, characterized in that, include: The method acquires multiple sleep stages of a target object during the night, wherein each stage of sleep data includes: dynamic electrocardiogram data and human three-dimensional acceleration data; the multiple sleep stages are obtained by dividing the nighttime sleep data; the nighttime sleep data is obtained by preprocessing the original nighttime electrocardiogram data and the original human three-dimensional acceleration data; the preprocessing operations include wavelet decomposition and reconstruction method, adaptive length running window algorithm, error correction, and removal of motion interference from heart rate data; The multiple sleep stages are input into multiple target classifiers to obtain multiple output results. Each stage of sleep data corresponds to one target classifier, and each output result indicates whether there is chronic stress or no chronic stress. The number of output results is statistically analyzed to obtain the chronic stress identification result of the target object.

2. The method as described in claim 1, characterized in that, The acquisition of multiple sleep stages of the target object at night includes: Obtain the nighttime sleep data of the target object; The nighttime sleep data is divided according to a preset sleep time threshold to obtain multiple stages of sleep data.

3. The method as described in claim 2, characterized in that, The acquisition of the target object's nighttime sleep data includes: Acquire the raw nighttime electrocardiogram data and raw human three-dimensional acceleration data of the target object; The original nighttime electrocardiogram data and the original human body three-dimensional acceleration data are filtered to obtain the nighttime sleep data.

4. The method as described in claim 1 or 2, characterized in that, The step of statistically analyzing the multiple output results to obtain the chronic stress identification result of the target object includes: Count the number of results that indicate chronic stress and the number of results that indicate no chronic stress among the multiple output results; The chronic stress identification result is obtained based on the number of results with chronic stress and the number of results without chronic stress.

5. The method as described in claim 4, characterized in that, The process of obtaining the chronic stress identification result based on the number of results with chronic stress and the number of results without chronic stress includes: If the number of results indicating chronic stress is not equal to the number of results indicating no chronic stress, then the output corresponding to the larger value between the number of results indicating chronic stress and the number of results indicating no chronic stress is taken as the chronic stress identification result; or, If the number of results with chronic stress is equal to the number of results without chronic stress, then the last output result among the multiple output results is removed to obtain the data of results with chronic stress and results without chronic stress after removal. The output corresponding to the larger value between the removed chronic stress result data and the non-chronic stress result data is taken as the chronic stress identification result.

6. The method as described in claim 1 or 2, characterized in that, Before inputting the multiple sleep stages into multiple target classifiers, the method further includes: Multiple training and validation datasets are obtained, wherein the training and validation datasets each include: heart rate data of multiple sample objects at night, human three-dimensional acceleration high-pass filtered data, and stress labels, wherein the stress labels include: chronic stress labels and no chronic stress labels; The initial classifier is trained using the training datasets of each stage in the training datasets of the multiple stages to obtain the classifiers to be verified for each stage. The classifiers to be validated at each stage are validated using the validation datasets of each stage in the multiple validation datasets to obtain the multiple target classifiers.

7. The method as described in claim 6, characterized in that, Before obtaining the training and validation datasets for multiple stages, the method further includes: During the nighttime phase, raw heart rate data, raw human acceleration data, and pressure labels of the multiple sample objects are collected. The raw heartbeat data and the raw human acceleration data are preprocessed to obtain heartbeat data and human three-dimensional acceleration high-pass filtered data of the multiple sample objects. Based on the heartbeat data and human body 3D acceleration high-pass filtered data of the multiple sample objects, as well as the pressure labels, a total training sample dataset and a total validation sample dataset are constructed.

8. The method as described in claim 7, characterized in that, The acquisition of training and validation datasets for multiple stages includes: The nighttime phase is divided into multiple time phases in chronological order; The total training sample dataset and the total validation sample dataset are divided according to the multiple time stages to obtain the training dataset and validation dataset for the multiple stages.

9. A device for identifying chronic stress, characterized in that, The apparatus is used to perform the method as described in claim 1, comprising: The data acquisition module is used to acquire multiple sleep data of the target object at night, wherein each stage of sleep data includes: dynamic electrocardiogram data and human three-dimensional acceleration data. The data processing module is used to input the multiple stages of sleep data into multiple target classifiers to obtain multiple output results. Each stage of sleep data corresponds to one target classifier, and each output result indicates whether there is chronic stress or no chronic stress. The result output module is used to perform statistical analysis on the multiple output results to obtain the chronic stress identification result of the target object.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, performs the method as described in any one of claims 1-8.

11. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the computer program is executed by the processor to perform the method as claimed in any one of claims 1-8.

Images