Intelligent monitoring system convenient for intervening juvenile depression self-injury behavior

Abstract

The invention relates to the technical field of teenager mental health monitoring, and discloses an intelligent monitoring system convenient for intervening teenager depression self-injury behaviors, and the system comprises a data collection module, a data analysis module, a single behavior evaluation module, a continuous behavior evaluation module, an optional data storage module, an early warning feedback module and a user interaction module. The data acquisition module acquires physiological parameters of the smart bracelet through Bluetooth and preprocesses the physiological parameters; the data analysis module identifies fluctuation characteristics and normalizes data; the single behavior evaluation module extracts key points of the physiological parameter curve and generates an abnormal waveform to judge single abnormality; the continuous behavior evaluation module generates an abnormal region based on continuous 24-hour data, and analyzes a change rate to judge continuous abnormality; the data storage module locally stores, encrypts and uploads to the cloud; the early warning feedback module sends graded early warning in real time; and the user interaction module responds to the query instruction. The system realizes multi-dimensional monitoring and accurate early warning of the depression self-injury behavior of the teenagers, and improves the mental health monitoring efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of adolescent mental health monitoring, in particular to an intelligent monitoring system for intervening in the depression and self-injury behavior of adolescents. BACKGROUND

[0002] Adolescent mental health problems have become an important challenge in the field of global public health, and early identification and intervention of depression and self-injury behavior are the difficulties and hotspots of current research. Traditional adolescent depression and self-injury behavior monitoring mainly relies on subjective scale assessment and manual observation, which has significant defects such as strong lag, insufficient objectivity, and inability to track in real time. Scale assessment is limited by the willingness and expression ability of adolescents to self-report, especially for individuals with emotional depression or impaired cognitive function, which can easily lead to missed diagnosis or misdiagnosis; manual observation faces problems such as high labor cost, limited monitoring period, and difficulty in capturing instantaneous physiological changes, which cannot meet the demand for real-time early warning of high-risk behavior.

[0003] With the development of wearable technology, physiological parameter-based monitoring methods provide a new direction to solve the above problems. Although some existing technologies use intelligent devices to collect physiological data such as heart rate and skin electricity for psychological state analysis, they generally have the following technical bottlenecks: first, the physiological data preprocessing method is single, which makes it difficult to effectively distinguish between valid signals and noise, resulting in insufficient accuracy of subsequent analysis; second, there is a lack of specific evaluation model for adolescent depression and self-injury behavior, which simply relies on a single parameter threshold for judgment, and cannot comprehensively analyze the fluctuation characteristics, periodic changes and trend evolution of physiological parameters in continuous periods; third, the early warning mechanism is lagging, and real-time identification and hierarchical feedback of abnormal physiological state cannot be achieved, and the data storage and interaction functions are not perfect, which makes it difficult to meet the needs of clinical intervention and long-term tracking. SUMMARY

[0004] The purpose of the present application is to provide an intelligent monitoring system for intervening in the depression and self-injury behavior of adolescents to solve the problems raised in the background.

[0005] To achieve the above purpose, the present application provides the following technical solution: an intelligent monitoring system for intervening in the depression and self-injury behavior of adolescents, the system comprising:

[0006] A data acquisition module for acquiring physiological parameter data collected by a smart bracelet worn by a user through a Bluetooth transmission method, and preprocessing the physiological parameter data to divide valid data segments and invalid data segments;

[0007] A data analysis module for identifying the fluctuation characteristics of the valid data segments and normalizing the physiological parameter data according to a preset reference value;

[0008] The single behavior evaluation module is configured to select a physiological parameter change curve in an arbitrary time period, extract peak points and valley points of each fluctuation period based on a preset algorithm, mark the peak points and the valley points as key points, obtain a plurality of key points with the largest deviation from a reference value, define the key points as abnormal points, sequentially connect adjacent abnormal points to generate an abnormal waveform, obtain a duration that the abnormal waveform exceeds a preset fluctuation range, and determine whether the duration is greater than a first preset threshold value, if not, determine that the physiological state in the time period is abnormal, and if yes, detect an interval between adjacent peaks and valleys of the waveform, and if the interval is greater than a preset interval, determine that the physiological response is abnormal.

[0009] The continuous behavior evaluation module is configured to select physiological parameter data in a continuous 24 hours, expand a physiological parameter curve in each preset time period range vertically with a reference value as a center line until abnormal waveforms in adjacent time periods overlap, generate a continuous physiological state abnormal area, and calculate a change rate of an abnormal area distribution density, and if a difference between the change rate and an average value of the change rate is greater than a second preset threshold value, determine that the physiological state is continuously abnormal on the day.

[0010] Preferably, the pre-processing process is as follows:

[0011] The physiological parameter data is subjected to smoothing processing to generate a standardized curve, the identification degree of fluctuation of the standardized curve is increased, the standardized curve is converted into segmented mark data by setting an amplitude threshold value, the segmented mark data is subjected to noise filtering based on a sliding window, and effective data segments and invalid data segments are separated based on feature extraction of the filtered segmented mark data.

[0012] Preferably, the preset fluctuation range is composed of 6 reference points, 2 points of which are arranged in a positive fluctuation interval of the reference value, and the other 4 points are arranged in a negative fluctuation interval of the reference value, the two points in the positive interval are positioned at a middle value of an upper limit of a regular fluctuation, two points of the negative interval are positioned at a warning line of a lower limit of the regular fluctuation, i.e., boundary points, and the other two points of the negative interval are positioned at two trisection points of a line connecting the boundary points.

[0013] Preferably, the process of normalizing the physiological parameter data is as follows:

[0014] The fluctuation range of the effective data segment is obtained, the deviation degrees of physiological parameter curves in different time periods from a reference line are calculated, and the physiological parameter data is subjected to amplitude correction according to an average value of the deviation degrees.

[0015] Preferably, the process of detecting the interval between adjacent peaks and valleys of the waveform is as follows:

[0016] According to the chronological order, the physiological parameter curve is traversed in a forward time sequence or a reverse time sequence, the fluctuation amplitude of each minute of data detected is counted, the amplitude value of the abnormal fluctuation is 1, the amplitude value of the normal fluctuation is 0, a data segment with the first amplitude value of 1 and a duration greater than a preset duration t is selected, and the data segment is marked as an abnormal starting segment, the physiological parameter curve is continuously traversed, when a data segment with an amplitude value of 0 and a duration greater than t is detected, the data segment is marked as a transition segment, the time difference T1 between the transition segment and the abnormal starting segment is obtained, if the time difference T1 is greater than or equal to a preset time difference t1, and no data segment with an amplitude value of 1 is detected along the time sequence, the transition segment is marked as an abnormal termination segment, and the abnormality is an isolated event;

[0017] If the time difference is less than t1, the physiological parameter curve is continuously traversed, when a data segment with an amplitude value of 1 and a duration greater than t is detected again, the data segment is marked as a second abnormal starting segment, the time difference T2 between the second abnormal starting segment and the transition segment is obtained, and if the time difference T2 is greater than a preset time difference t2, it is judged that the physiological reaction is an abnormal continuous state.

[0018] Preferably, the specific process of judging the continuous abnormality of the physiological state is:

[0019] For any selected time period, a horizontal line corresponding to the reference value of the time period is selected, the deviation value of all data points in the time period relative to the horizontal line is obtained, the deviation value of each data point relative to the horizontal line is multiplied by the same coefficient, the value of the coefficient is continuously increased, until adjacent time periods are overlapped, a continuous region of abnormal fluctuations is obtained, the corresponding distribution density change curve is obtained by mean filtering the continuous region, and the change rate of the curve is obtained.

[0020] Preferably, the process of generating an abnormal waveform is:

[0021] First, three abnormal points with the largest deviation reference value amplitude are selected, adjacent abnormal points are connected to generate a to-be-determined abnormal waveform, then the largest deviation amplitude point in the remaining abnormal points is selected and added to the connection line of the abnormal waveform one by one, until the amplitude of the waveform corresponding to the original abnormal point in the connection line formed by the newly added abnormal point becomes smaller, the added point is removed, and the connection line formed by the original abnormal point is the abnormal waveform of the time period.

[0022] Preferably, the system further comprises:

[0023] The data storage module is configured to store the preprocessed physiological parameter data in the local database in chronological order, and upload the physiological parameter data to the cloud server through an encrypted channel for backup.

[0024] Preferably, the system further comprises:

[0025] An early warning feedback module is configured to send a hierarchical early warning signal to the bound monitoring terminal through Bluetooth transmission when an abnormal physiological state or an abnormal physiological response is detected.

[0026] Preferably, the system further comprises:

[0027] A user interaction module is configured to receive a query instruction sent by the monitoring terminal and retrieve corresponding physiological parameter data analysis results from the local database or the cloud server according to the instruction content.

[0028] Compared with the prior art, the present application has the following advantages:

[0029] The data acquisition module acquires physiological parameter data collected by the smart bracelet through Bluetooth transmission and pre-processes the data to divide valid data segments and invalid data segments. This process generates a standardized curve through smoothing processing, improves the recognition degree of fluctuations, and effectively eliminates noise interference in the data by setting an amplitude threshold and a sliding window noise filter, thereby ensuring that subsequent analysis is based on high-quality physiological data and improving the reliability and usability of the data.

[0030] The data analysis module identifies the fluctuation characteristics of the valid data segments and performs normalization processing. By obtaining the fluctuation range of the valid data segments, the deviation degree of physiological parameter curves at different time periods from the baseline is calculated and amplitude correction is performed, so that the physiological data of different time periods and different individuals are comparable, thereby laying a scientific foundation for subsequent behavior assessment.

[0031] The single behavior assessment module extracts key points and identifies abnormal points by selecting a physiological parameter change curve in an arbitrary time period, generates an abnormal waveform, and determines whether the physiological state is abnormal. This module not only considers the duration of abnormal waveforms exceeding the preset fluctuation range, but also distinguishes between periodic fluctuations and sudden fluctuations, further detects the interval between adjacent wave peaks and troughs, and evaluates single physiological responses from multiple dimensions, which can accurately capture instantaneous abnormal physiological states and improve the ability to identify sudden or occasional abnormal behaviors.

[0032] The continuous behavior assessment module selects twenty-four hours of continuous physiological parameter data, generates a continuous physiological state abnormal area, and calculates the change rate of the abnormal area distribution density, to determine the persistent abnormality of the physiological state. This analysis method based on continuous time period data can effectively identify long-term trends and change rules of the physiological state, avoid misjudgments that may be caused by single assessment, and provide strong support for monitoring of persistent depression or self-injury tendency.

[0033] The data storage module stores the preprocessed physiological parameter data in chronological order in the local database and uploads it to the cloud server backup through an encrypted channel. This not only guarantees the security and integrity of the data, but also facilitates subsequent retrospective analysis and long-term tracking, providing rich data resources for mental health research and clinical intervention.

[0034] The early warning feedback module sends a hierarchical early warning signal to the bound monitoring terminal through Bluetooth when an anomaly is detected, achieving real-time response to abnormal states. The hierarchical early warning mechanism enables monitoring personnel to take appropriate intervention measures according to the warning level, improving the relevance and timeliness of intervention and helping to prevent self-injury behavior from occurring in a timely manner.

[0035] The user interaction module receives the query instruction of the monitoring terminal and retrieves the data analysis results, making it convenient for monitoring personnel to understand the physiological state and historical data of adolescents at any time, promoting information exchange between monitoring personnel and the monitoring system, and providing convenience for personalized psychological intervention and health management. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 A working principle diagram of the intelligent monitoring system for intervening in the depression and self-injury behavior of adolescents according to the present application;

[0037] Figure 2 A flowchart of continuous behavior assessment;

[0038] Figure 3 A working principle diagram of the early warning feedback module;

[0039] Figure 4 A working principle diagram of the user interaction module. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the present application.

[0041] Please refer to Figures 1-4 The specific implementation of the intelligent monitoring system for intervening in the depression and self-injury behavior of adolescents according to the present application is as follows:

[0042] The intelligent monitoring system for intervening in the depression and self-injury behavior of adolescents includes a data acquisition module, a data analysis module, a single behavior assessment module, and a continuous behavior assessment module.

[0043] The data acquisition module acquires physiological parameter data collected by a smart bracelet worn by a user through a Bluetooth transmission mode. The physiological parameter data includes heart rate, skin electrical response, body temperature, and other indicators reflecting the physiological state of an individual. The data acquisition module performs a preprocessing operation on the physiological parameter data. The specific process is to divide the valid data segment and the invalid data segment.

[0044] The data analysis module identifies the fluctuation characteristics of the valid data segment and performs normalization processing on the physiological parameter data based on a preset reference value to eliminate analysis errors caused by dimensional differences of different physiological parameters.

[0045] The single behavior evaluation module selects a physiological parameter change curve in an arbitrary time period, extracts the peak point and the valley point of each fluctuation period based on a preset algorithm, and marks both as key points. The module further acquires a number of key points with the largest deviation amplitude from the reference value, which are defined as abnormal points. Adjacent abnormal points are connected in sequence to generate an abnormal waveform, and the duration of the abnormal waveform exceeding a preset fluctuation range is calculated. When the duration is greater than a first preset threshold, the fluctuation type is determined: if it is non-periodic fluctuation or sudden fluctuation, it is directly determined that the physiological state in the selected time period is abnormal; if it is periodic fluctuation, the interval between adjacent peaks and valleys of the waveform is detected, and if the interval is greater than a preset interval, it is determined that the physiological response is abnormal. The implementation of the preset algorithm for extracting the peak point and the valley point of each fluctuation period of the physiological parameter change curve includes:

[0046] The physiological parameter change curve data after preprocessing and normalization is acquired. The preprocessing has smoothed and separated the effective data segment from the original data, and the normalization processing has corrected the amplitude difference of different time periods, so that the data can be uniformly analyzed.

[0047] In the valid data segment, the system uses a sliding window method to traverse the physiological parameter change curve. For each data point on the curve, check the N data points before and after it (N is a preset window size, which can be adjusted according to the sampling frequency and physiological parameter characteristics). If the value of the point is greater than the value of all other points in the window, it is marked as a peak point; if the value of the point is less than the value of all other points in the window, it is marked as a valley point.

[0048] The marked peak points and valley points are verified and screened. For each candidate point, calculate its deviation amplitude from the reference value. If the deviation amplitude is less than a preset minimum threshold, the point is considered to be noise interference and is removed. At the same time, check the time interval between adjacent peak points or valley points. If the interval is less than a preset minimum time interval, keep the point with a larger deviation amplitude and remove the other point.

[0049] Further according to the characteristics of the physiological parameters, the screened peak points and valley points are classified. For example, the first M points with the largest deviation from the reference value are defined as abnormal points, which may be related to the depressive self-harm behavior of adolescents. All the peak points and valley points are marked as key points for generating abnormal waveforms subsequently.

[0050] The continuous behavior assessment module selects the physiological parameter data of the past twenty-four hours, and expands the physiological parameter curve of each time period vertically with the reference value as the center line until the abnormal waveforms in adjacent time periods overlap to generate a continuous physiological state abnormal area. The module calculates the change rate of the abnormal area distribution density, and when the difference between the change rate and the average change rate is greater than a second preset threshold, it is determined that the physiological state is continuously abnormal on that day. The implementation of expanding the physiological parameter curve of each time period vertically with the reference value as the center line includes:

[0051] A reference value is determined for the physiological parameter curve of each time period. The reference value can be the average of all data points in the time period, or a value calculated by a personalized reference model established through historical data.

[0052] For the physiological parameter curve of each time period, the system calculates the deviation of each data point from the reference value. A positive deviation value indicates that the data point is higher than the reference value, and a negative deviation value indicates that the data point is lower than the reference value.

[0053] The system expands the physiological parameter curve vertically by multiplying each data point's deviation from the reference value by the same coefficient. Initially, the coefficient is set to 1, i.e., the original curve is maintained. Subsequently, the system gradually increases the value of the coefficient by a preset step each time.

[0054] During the process of increasing the coefficient, the system checks in real time whether the physiological parameter curves of adjacent time periods overlap. Specifically, for two adjacent time periods, the expanded curve values at the same time points are calculated. If there is a time point such that the expanded curve value of one time period is greater than that of the other time period, it is considered that the curves of the two time periods have overlapped.

[0055] When the system detects that the physiological parameter curves of adjacent time periods all overlap with each other, it stops increasing the coefficient. At this time, the system obtains a continuous area of abnormal fluctuations, and the curve changes in this area reflect the continuous abnormality of the physiological state of adolescents.

[0056] The system performs mean filtering on the obtained continuous area to smooth the curve and reduce noise interference, obtaining the corresponding distribution density change curve. Finally, the change rate of the curve is calculated and compared with the average change rate. When the difference is greater than a second preset threshold, it is determined that the physiological state is continuously abnormal in the past twenty-four hours.

[0057] The application will be further described in connection with Examples 1 to 5:

[0058] Example 1:

[0059] In an intelligent monitoring system for intervening in the depressive self-harming behavior of adolescents, the preprocessing process is a key link to ensure the accuracy of subsequent data analysis. This process optimizes the original physiological parameter data through a series of ordered operations, and the specific implementation is as follows:

[0060] The system first performs smoothing processing on the physiological parameter data obtained from the smart bracelet through Bluetooth transmission. Physiological parameter data may have high-frequency noise due to user activity, device interference, etc. For example, heart rate data may have instantaneous sharp fluctuations due to exercise, and skin electrical response data may have spiky interference due to environmental humidity. The purpose of smoothing processing is to eliminate such noise and improve the continuity and fluctuation recognition of the data curve. In specific operations, Gaussian filtering algorithm or median filtering algorithm can be used to process the original data. Gaussian filtering smooths the data points based on normal distribution weights, which is suitable for noise following normal distribution; median filtering eliminates impulse noise by taking the median value of the data in the sliding window, which has good inhibition effect on salt and pepper noise. Regardless of which algorithm is used, the appropriate filter window size needs to be set according to the characteristics of the physiological parameter, for example, for skin electrical response data with a sampling frequency of 5 Hz, the window size can be set to 30 data points (i.e. 6 seconds), so that the filtered curve can not only retain the true physiological fluctuation characteristics, but also effectively reduce noise interference, generating a standardized curve.

[0061] After completing the smoothing processing, the system enters the segmented labeled data link. In this phase, an amplitude threshold needs to be set to distinguish different states of the data. The setting of the amplitude threshold needs to be combined with the conventional fluctuation range of the specific physiological parameter, for example, the conventional fluctuation range of heart rate is usually 50-100 times / min, and the conventional fluctuation range of body temperature is 36-37.5℃. The system marks the region in the standardized curve that is higher than the positive amplitude threshold as a high activity segment, marks the region that is lower than the negative amplitude threshold as a low activity segment, and marks the region between the positive and negative thresholds as a normal segment. For example, if the positive amplitude threshold of heart rate is set to 90 times / min and the negative amplitude threshold is set to 60 times / min, the part of the heart rate data greater than 90 times / min is marked as a high activity segment, the part less than 60 times / min is marked as a low activity segment, and the part between 60-90 times / min is marked as a normal segment. Through this operation, the continuous standardized curve is converted into segmented labeled data with clear state identification, which facilitates subsequent targeted processing of different state data segments.

[0062] Next, the system uses a sliding window technique to filter noise from the segmented labeled data. The size of the sliding window is determined based on the sampling frequency of the physiological parameter and the actual application scenario. For example, for temperature data with a sampling frequency of 1Hz, the window size can be set to 60 data points (i.e., 1 minute), with the window sliding by one data point at a time. Within each window, the system calculates a data statistic (such as the mean, variance, or median) and compares each data point within the window with the statistic, removing data points that deviate from the statistic by more than three standard deviations. For example, if the mean of the temperature data within a window is 36.5°C and the standard deviation is 0.2°C, data points that deviate from the mean by more than 3×0.2=0.6°C (i.e., data points below 35.9°C or above 37.1°C) are considered noise points and removed. This method further removes any residual random noise from the segmented labeled data, ensuring that the retained data points truly reflect the actual fluctuations of the physiological parameter.

[0063] After noise filtering, the system needs to perform feature extraction on the processed segmented labeled data to separate valid data segments from invalid data segments. The parameters for feature extraction include but are not limited to fluctuation amplitude, fluctuation frequency, duration, and number of state transitions. The fluctuation amplitude reflects the range of change of physiological parameters within a certain time period. For example, in a certain data segment, the heart rate rises from 70 beats / minute to 85 beats / minute, with a fluctuation amplitude of 15 beats / minute. The fluctuation frequency indicates the number of fluctuations per unit time, such as the number of cycles of heart rate fluctuation per minute. The duration refers to the duration of a certain state (such as a high-activity segment, a low-activity segment, or a normal segment). The number of state transitions counts the frequency of switching between different states within a data segment, such as the number of times a normal segment switches to a high-activity segment and then to a normal segment.

[0064] The system evaluates data segments based on these characteristic parameters: valid data segments must meet conditions such as containing a complete physiological fluctuation cycle, the noise level is below a certain threshold, and the data integrity is higher than a certain ratio (such as 90%). For example, a heart rate data segment lasting 10 minutes is judged to be a valid data segment if it contains at least 5 complete fluctuation cycles (i.e., the frequency is not lower than 0.08Hz), the noise point ratio is lower than 5%, and there is no data missing for more than 30 seconds continuously. On the contrary, if there is severe noise interference (noise points account for more than 20%), the data missing rate is higher than 10%, or the fluctuation cycle is incomplete, it is classified as an invalid data segment.

[0065] In practical applications, the focus of feature extraction for different physiological parameters may differ. For heart rate data, more attention is paid to the stability of fluctuation frequency and amplitude. For skin conductance data, more attention is paid to the duration and transition rules of low activity segments and high activity segments. Through targeted feature extraction and evaluation criteria, the system can accurately separate effective data segments from ineffective data segments, providing high-quality input data for subsequent data analysis modules, and ensuring that the evaluation results of the intelligent monitoring system for adolescent depression and self-injury behavior have high reliability and accuracy. The entire preprocessing process is executed in an orderly manner through four links of smoothing processing, segment marking, noise filtering and feature extraction, forming a complete data optimization mechanism, effectively improving the usability of the original physiological parameter data, and laying a solid data foundation for subsequent single behavior evaluation and continuous behavior evaluation of the system.

[0066] Embodiment 2:

[0067] In an intelligent monitoring system for facilitating intervention in adolescent depression and self-injury behavior, the construction of a preset fluctuation range is an important benchmark for judging physiological parameter abnormalities. It forms a structured judgment framework by setting a certain number and position of reference points in the positive and negative intervals of the reference value. The specific implementation is as follows:

[0068] The preset fluctuation range is composed of 6 reference points, divided into positive fluctuation interval and negative fluctuation interval. The positive interval sets 2 reference points, and the negative interval sets 4 reference points. The reference value is the central reference line for judging whether the physiological parameter is normal or not, which is usually determined based on the statistical average of the physiological parameters of the adolescent group. For example, the reference value of heart rate can be set to 75 times per minute, the reference value of skin conductance can be set to 5 microsiemens, and the reference value of body temperature can be set to 36.8℃.

[0069] In the positive fluctuation interval, the positioning of the two reference points needs to be combined with the conventional fluctuation characteristics of the physiological parameters. The positive interval refers to the fluctuation range of the physiological parameters higher than the reference value, and the upper limit of the conventional fluctuation refers to the highest value that the physiological parameters of the teenager can reach in the normal activity state. This value can be obtained through large sample statistical data, for example, the upper limit of the conventional positive fluctuation of heart rate can be set to 100 times / min (corresponding to +33.3% of the reference value). The two reference points of the positive interval are located at the middle value position of the upper limit of the conventional fluctuation, that is, in the interval from the reference value to the upper limit of the conventional fluctuation, the middle position and the upper limit position are taken as the reference points. Taking heart rate as an example, if the reference value is 75 times / min and the upper limit of the conventional positive fluctuation is 100 times / min, the difference between them is 25 times / min, then the middle value position is the reference value plus half of the difference (i.e. 75+12.5=87.5 times / min), and the upper limit position is 100 times / min. Therefore, the two reference points of the positive interval are 87.5 times / min and 100 times / min, respectively, corresponding to the middle value and the upper limit value of the upper limit of the conventional fluctuation. These two reference points are used to identify the progressive upper limit area of normal physiological fluctuation, where the middle value point is the warning boundary of the conventional fluctuation, and the upper limit value point is the physical boundary of the normal fluctuation. When the physiological parameter exceeds the middle value point but does not reach the upper limit value point, it is prompted that there may be mild abnormal fluctuation; when it exceeds the upper limit value point, it is considered to be close to the limit of the conventional fluctuation.

[0070] The construction of the four reference points of the negative fluctuation interval is more complex and needs to be hierarchically set in combination with the risk level of abnormal fluctuation. The negative interval refers to the fluctuation range of the physiological parameters lower than the reference value, and the lower limit of the conventional fluctuation is also determined based on group statistical data, for example, the lower limit of the conventional negative fluctuation of heart rate can be set to 50 times / min (corresponding to -33.3% of the reference value). Among the four reference points of the negative interval, two are boundary points located at the warning line position of the lower limit of the conventional fluctuation, and the other two are trisection points located at the trisection position of the connecting line of the boundary points. Specifically, the boundary points need to be set at a position close to the lower limit of the conventional fluctuation but not yet reaching the limit value. Taking heart rate as an example, if the lower limit of the conventional negative fluctuation is 50 times / min, the boundary points can be set to 55 times / min and 50 times / min, respectively, where 55 times / min is the warning line (corresponding to -26.7% of the reference value) and 50 times / min is the physical lower limit. These two boundary points are used to distinguish the critical region of normal fluctuation and abnormal fluctuation. When the physiological parameter is lower than the warning line but higher than the physical lower limit, it is prompted that there may be moderate abnormal fluctuation; when it is lower than the physical lower limit, it is considered to enter the severe abnormal region.

[0071] In the interval between the two boundary points (e.g. 55 and 50 times / minute), the system further sets two trisection points to refine the levels of negative fluctuation. The difference between the boundary points (55-50 = 5 times / minute) is divided into three segments, each about 1.67 times / minute, so the first trisection point is 55-1.67 = 53.33 times / minute, and the second trisection point is 55-3.33 = 51.67 times / minute. Thus, the four reference points of the negative interval are 55 times / minute (warning line), 53.33 times / minute (first trisection point), 51.67 times / minute (second trisection point), and 50 times / minute (physical lower limit) in turn. Through the two trisection points, the system can divide the negative fluctuation into three sub-intervals: 55-53.33 times / minute for mild abnormal interval, 53.33-51.67 times / minute for moderate abnormal interval, and 51.67-50 times / minute for severe abnormal interval. This hierarchical setting allows the system to more accurately judge the severity of abnormal fluctuations according to the physiological parameter falling into different sub-intervals, providing more detailed quantitative basis for subsequent single behavior assessment and continuous behavior assessment.

[0072] In practical applications, the setting of reference points for different physiological parameters needs to be adjusted in combination with their physiological significance and clinical experience. For example, the positive fluctuation of skin conductance response may be related to emotional arousal, and the upper limit of regular fluctuation can be set to +50% of the reference value, with the two reference points of the positive interval being +25% and +50%; the negative fluctuation may be related to emotional depression, and the lower limit of regular fluctuation can be set to -40% of the reference value, with the boundary points of the negative interval being -30% (warning line) and -40% (physical lower limit), and the trisection points being -33.3% and -36.7%. Regardless of the specific parameters, the construction of the preset fluctuation range follows the following principles: the reference points of the positive interval are used to identify the upper limit of normal fluctuation, the reference points of the negative interval are used to achieve hierarchical warning of abnormal fluctuation through boundary points and trisection points, and all reference points together constitute a three-dimensional judgment framework containing different risk levels.

[0073] When the system generates an abnormal waveform in a single behavior assessment, it needs to determine whether the waveform is beyond the preset fluctuation range. Specifically, if the peak or valley value of the abnormal waveform touches the middle value point, the upper limit value point of the positive interval, or the warning line point, the three equal division points, and the physical lower limit point of the negative interval, the level and type of abnormal fluctuation are determined according to the position of the touched reference point. For example, if the valley value of the heart rate curve falls into the 53.33-55 beats / minute interval (mild abnormal interval) of the negative interval, it is determined as a mild abnormal fluctuation; falls into the 51.67-53.33 beats / minute interval (moderate abnormal interval), it is determined as a moderate abnormal fluctuation; falls into the 50-51.67 beats / minute interval (severe abnormal interval), it is determined as a severe abnormal fluctuation. Through this quantitative judgment mechanism based on reference points, the system can avoid the limitations of single threshold judgment, more comprehensively capture the abnormal changes of physiological parameters, and provide a more scientific basis for early identification of adolescent depressive self-injurious behavior.

[0074] The construction process of the entire preset fluctuation range realizes the hierarchical division of physiological parameter fluctuations through the differential design of the reference value positive and negative intervals, combined with statistical data and clinical experience of conventional fluctuation range. This structured reference point setting not only enhances the accuracy and operability of the system's judgment, but also provides a clear reference standard for subsequent abnormal waveform duration calculation, periodic fluctuation detection, etc., ensuring that the intelligent monitoring system can more accurately identify the abnormal physiological state of adolescents, providing reliable support for the warning and intervention of depressive self-injurious behavior.

[0075] Embodiment 3:

[0076] In an intelligent monitoring system for intervening in adolescent depressive self-injurious behavior, normalizing physiological parameter data is a key step to eliminate differences in different time periods and different parameter dimensions. Its implementation is achieved through fluctuation range analysis, deviation degree calculation, and amplitude correction, and the specific process is as follows:

[0077] The system obtains the fluctuation range of the effective data segment. The effective data segment is high-quality data separated after preprocessing by the data acquisition module, containing complete physiological fluctuation periods and noise levels meeting the requirements. The fluctuation range is determined by the maximum and minimum values of the physiological parameters in this data segment. For example, the maximum value of heart rate data in a certain period is 95 beats / minute, and the minimum value is 60 beats / minute, so the fluctuation range is 95-60=35 beats / minute. If it is skin galvanic response data, the maximum value is 8 microsiemens, and the minimum value is 2 microsiemens, the fluctuation range is 6 microsiemens. The fluctuation range reflects the actual amplitude of change of the physiological parameter in this period, and is the basis parameter for subsequent normalization processing.

[0078] The system needs to calculate the degree of deviation between the physiological parameter curve and the baseline at different time periods. The baseline is a horizontal line formed by the baseline value of the corresponding physiological parameter. The baseline value is usually determined based on the statistical average of the physiological parameters of the adolescent group, such as the heart rate baseline value B is 75 beats / minute and the skin galvanic response baseline value B is 5 microsiemens. For the physiological parameter data of each time period, the degree of deviation is characterized by the average of the absolute value of the difference between all data points and the baseline value in that period. The calculation formula is:

[0079]

[0080] Where D represents the mean of the deviation, n is the number of data points in the period, and x i is the measured value of the i-th data point, and B is the baseline value of the corresponding physiological parameter. For example, if the heart rate data for a certain period contains 300 data points (sampling frequency is 5 Hz, duration is 1 minute), and the sum of the absolute values ​​of the differences between each data point and the baseline value of 75 is 1500, then the mean deviation is D = 1500 ÷ 300 = 5 beats / minute. If the galvanic skin response data contains 200 data points with a sum of the absolute values ​​of the differences is 400, then the mean deviation is D = 400 ÷ 200 = 2 microsiemens. This mean reflects the average deviation of the physiological parameter from the baseline value during that period and is a key indicator for determining the center of the data distribution.

[0081] Based on the mean of the degree of deviation, the system performs amplitude correction on the physiological parameter data to achieve normalization. The core idea of ​​amplitude correction is to adjust the distribution center of the data in different time periods to the reference value to eliminate the analysis error caused by the overall offset. The specific operation is as follows: If the mean of the degree of deviation D is greater than 0, it means that the overall distribution of the data in this period is biased towards one side of the reference value (positive or negative), and the data needs to be shifted to align the distribution center with the reference value. For example, if the mean of the degree of deviation of the heart rate data in a certain period is +5 beats / minute (that is, the data as a whole is 5 beats / minute higher than the reference value), then all data points in this period will be subtracted by 5 beats / minute, so that the adjusted data fluctuates around 75 beats / minute; if the mean of the degree of deviation is -3 beats / minute (the data as a whole is 3 beats / minute lower than the reference value), then all data points will be added by 3 beats / minute to achieve central adjustment.

[0082] In practical applications, the normalization of different physiological parameters needs to be combined with their physical meaning. For parameters with a clear physiological lower limit or upper limit, such as heart rate and body temperature, amplitude correction should avoid adjusting the data beyond the reasonable physiological range. For example, if the baseline value of the body temperature data in a certain period is 36.8°C, the minimum measured value is 35.5°C, and the calculated mean deviation is -0.8°C, if all data points are directly added by 0.8°C, the minimum value after adjustment is 36.3°C, which is still within the normal physiological range (36-37.5°C), and the adjustment is effective; if the minimum measured value is 35.0°C, the mean deviation is -1.8°C, and the minimum value after adjustment is 36.8°C, although it meets the normalization requirements, it needs to be combined with clinical experience to judge whether the original data has device error or user abnormal state, to avoid mechanical adjustment to cover up the real physiological signal.

[0083] The key role of normalization processing is to unify the dimensions of different time periods and different physiological parameters, making it possible to compare and evaluate across time periods. For example, a teenager's heart rate fluctuation range in the morning is 65-85 beats per minute (baseline value 75 beats per minute), with a mean deviation of 5 beats per minute; the skin electric response fluctuation range in the afternoon is 3-7 microsiemens (baseline value 5 microsiemens), with a mean deviation of 2 microsiemens. Through amplitude correction, the morning heart rate data is adjusted to 60-80 beats per minute (centered on 75, fluctuation range ±5), and the afternoon skin electric response data is adjusted to 3-7 microsiemens (centered on 5, fluctuation range ±2). Although the physical dimensions of the two are different, after normalization, both are centered on the baseline value, with fluctuation ranges of ±5 and ±2 respectively, and the system can directly compare the fluctuation stability of the two to determine whether there is a physiological state abnormality.

[0084] In addition, normalization processing also needs to consider the time sequence continuity of the data. For physiological parameter data of continuous twenty-four hours, the normalization processing of each period needs to be carried out independently to avoid the influence of the adjustment of the previous period on the original data characteristics of the next period. For example, in the continuous behavior evaluation module, the system sequentially normalizes the data of each hour to ensure that the abnormal waveform analysis of each period is based on independently adjusted data, avoiding cumulative errors that may lead to abnormal region recognition bias.

[0085] In the implementation process, the system needs to optimize the calculation logic of the mean deviation to cope with the influence of data missing or abnormal values. For example, if there is missing data in a certain period, the data points need to be completed by interpolation method (such as linear interpolation, polynomial interpolation) first, and then the mean deviation is calculated; if there is an individual abnormal value (such as a sudden increase in heart rate caused by exercise), the abnormal value can be removed or corrected by a sliding window before calculation to ensure that the mean value truly reflects the overall deviation trend of the data.

[0086] In summary, the normalization of physiological parameter data realizes the centralization adjustment of data, eliminates the interference of dimensional difference and overall deviation on the analysis results through the three links of fluctuation range acquisition, deviation degree calculation and amplitude correction. The process takes the mean of deviation degree as the core index, and through simple and effective mathematical transformation, the physiological parameter data with different characteristics have comparability, which provides a unified quantitative basis for the identification of abnormal points in single behavior evaluation, the generation of abnormal waveforms and the construction of abnormal regions in continuous behavior evaluation, and ensures that the intelligent monitoring system can accurately capture the abnormal changes of the physiological state of adolescents, and provide reliable support for early warning of depressive self-injurious behavior.

[0087] Example 4:

[0088] In an intelligent monitoring system for facilitating intervention in depressive self-injurious behavior of adolescents, the detection process of the waveform adjacent peak-trough interval is a key link for judging physiological reaction abnormalities, which is realized through steps such as time sequence traversal, fluctuation amplitude marking, abnormal segment identification and interval calculation, and the specific implementation is described in detail as follows with examples:

[0089] The system first traverses the physiological parameter curve in time sequence. The time sequence can be selected from the forward time sequence from the start time to the end time of data collection, or the reverse time sequence from the end time to the start time, to adapt to different data processing logic. Taking the forward time sequence as an example, assuming that the heart rate data of a certain adolescent from 9:00 to 10:00 is divided into 60 data segments per minute, the system starts analyzing the data from 9:00 minute by minute.

[0090] In the traversal process, the system needs to count the fluctuation amplitude of each minute data, and convert the fluctuation amplitude into binary markers: abnormal fluctuation marker 1, normal fluctuation marker 0. The judgment standard of fluctuation amplitude is based on the preset fluctuation threshold, for example, set the heart rate fluctuation more than ±15% of the reference value (such as 75 times / minute) (i.e. more than 86 times / minute or lower than 64 times / minute) as abnormal fluctuation, amplitude value 1; within ±15% range as normal fluctuation, amplitude value 0. Assuming that the maximum heart rate data in 9:05-9:06 minutes is 88 times / minute (more than +15% threshold 86 times / minute), and the minimum value is 62 times / minute (lower than -15% threshold 64 times / minute), then the fluctuation amplitude value of this minute is marked as 1; the heart rate data in 9:07-9:08 minutes is between 65-85 times / minute, marked as 0.

[0091] After the amplitude range marking, the system starts to identify the abnormal starting segment. The abnormal starting segment is defined as "the first data segment with amplitude value 1 and duration greater than preset duration t". The preset duration t should be set according to the physiological parameter characteristics, for example, for heart rate, t can be set to 5 minutes to avoid accidental fluctuations in a short time triggering false positives. Assuming that the fluctuation amplitude value is 1 for 6 consecutive minutes during 9:10-9:15 (duration 6 minutes > t = 5 minutes), the system marks 9:10-9:15 as an abnormal starting segment, records the starting time as 9:10, and the ending time as 9:15.

[0092] Next, the system continues to traverse the subsequent data to find the transition segment. The transition segment is defined as "the data segment with amplitude value 0 and duration greater than t". Assuming that the fluctuation amplitude value is 0 for 6 consecutive minutes during 9:16-9:21 (duration 6 minutes > t = 5 minutes), it is marked as a transition segment, with the starting time recorded as 9:16 and the ending time recorded as 9:21. At this time, the system calculates the time difference T1 between the transition segment and the abnormal starting segment, i.e. 9:16-9:15 = 1 minute (the actual calculation is the time difference between the two time segments, and here the example is simplified in minutes). If the preset time difference t1 is 30 minutes, since T1 = 1 minute < t1 = 30 minutes, the system judges that the time interval between the abnormal starting segment and the transition segment is too short, and needs to further judge whether there is a subsequent abnormal fluctuation.

[0093] Assuming that the fluctuation amplitude value is 1 for 6 consecutive minutes again during 9:22-9:27 (duration > t), it is marked as a second abnormal starting segment, with the starting time recorded as 9:22 and the ending time recorded as 9:27. The system calculates the time difference T2 between the second abnormal starting segment and the transition segment, i.e. 9:22-9:21 = 1 minute. If the preset time difference t2 is 15 minutes, since T2 = 1 minute < t2 = 15 minutes, the system judges that the interval between the two abnormal fluctuations is too short, which belongs to the continuous abnormal state, so the abnormal starting segment of 9:10-9:15, the transition segment of 9:16-9:21, and the second abnormal starting segment of 9:22-9:27 are merged, and it is determined as a continuous abnormal state of physiological response.

[0094] Another case is that the abnormal starting segment is 9:10-9:15 (lasting 6 minutes), and the transition segment is 9:30-9:35 (lasting 6 minutes), and the time difference T1 = 9:30-9:15 = 15 minutes. If t1 = 30 minutes, T1 = 15 minutes < t1, and no data segment with amplitude value 1 is detected during the subsequent traversal to 10:00, then the system judges that the transition segment is an abnormal ending segment, and this abnormality is an isolated event, i.e. only a short abnormal fluctuation occurs during 9:10-9:15, and then it returns to normal and has no subsequent abnormalities.

[0095] In the scenario of reverse time series traversal, the system analyzes from the end of the data (e.g., 10:00) to the beginning (e.g., 9:00). Assuming that an abnormal starting segment with an amplitude value of 1 is detected at 9:50-9:55 (lasting 6 minutes), forward traversal finds a transition segment with an amplitude value of 0 at 9:40-9:45 (lasting 6 minutes), and the time difference T1 = 9:50-9:45 = 5 minutes (reverse calculation as the absolute value of the time interval) is calculated. If t1 = 30 minutes, T1 = 5 minutes < t1, and continue to forward traverse to 9:00 without other abnormal segments, it is determined as an isolated event; if an abnormal starting segment is detected again at 9:30-9:35, T2 = 9:40-9:35 = 5 minutes is calculated, and if t2 = 15 minutes, T2 = 5 minutes < t2, it is determined as an abnormal persistent state.

[0096] In practical applications, the preset time length t and time differences t1 and t2 of different physiological parameters need to be differentiated. For example, the fluctuation of skin electric response is usually more sensitive, t can be set to 3 minutes, t1 to 20 minutes, and t2 to 10 minutes; the body temperature fluctuation is slower, t can be set to 10 minutes, t1 to 60 minutes, and t2 to 30 minutes. This differentiated setting is based on the natural fluctuation period of different physiological parameters, avoiding false positives or false negatives caused by uniform thresholds.

[0097] During the detection process, the system needs to handle the continuity of the data segment boundaries. For example, if the abnormal starting segment ends at 9:15:59 and the transition segment starts at 9:16:00, they seamlessly connect in time, and the system needs to consider them as continuous periods to avoid interval calculation deviation caused by sampling timestamp errors. In addition, for continuous twenty-four-hour data across days, the system needs to automatically handle date boundaries to ensure the consistency of traversal logic.

[0098] Through the above detection logic, the system can effectively distinguish between isolated abnormal fluctuations and persistent abnormal fluctuations. Isolated abnormal fluctuations may be caused by transient emotional fluctuations, physiological stress reactions, etc., while persistent abnormal fluctuations may indicate that the adolescent is in a long-term stress or depression state and needs to be paid special attention. For example, if a certain adolescent has persistent heart rate abnormal fluctuations multiple times for three consecutive days, the system can combine the results of the continuous behavior evaluation module to comprehensively judge whether the physiological state of the adolescent has persistent abnormalities, providing key evidence in the time dimension for the early warning of depression and self-harm behavior.

[0099] The entire detection process uses time sequence as a clue, simplifying the determination of fluctuation states through binary labeling. By combining preset durations and time difference thresholds, the system logically associates abnormal segments, avoiding the one-sidedness of judgments based solely on a single abnormal point. Furthermore, the forward and reverse time sequence traversal method enhances the robustness of the algorithm, ensuring that the temporal characteristics of abnormal fluctuations can be accurately identified regardless of the order in which data is collected. This time interval-based detection mechanism provides a quantitative basis for systematic analysis of the regularity and persistence of physiological responses, enabling the intelligent monitoring system to more accurately capture abnormal patterns of change in adolescents' physiological states and enhance the scientific nature and effectiveness of monitoring depressive and self-harm behaviors.

[0100] Example 5:

[0101] In an intelligent monitoring system for facilitating intervention in adolescent depression and self-harm behaviors, Example 5 involves the specific implementation of a data storage module, a warning feedback module, and a user interaction module. These modules work together to improve the practicality and reliability of the system. The following describes the system in detail with reference to specific examples.

[0102] Implementation method of the data storage module: The data storage module is responsible for storing the pre-processed physiological parameter data in a local database in chronological order and uploading it to the cloud server for backup through an encrypted channel. Taking heart rate data as an example, suppose a smart bracelet worn by a teenager collects 10 minutes of heart rate data between 9:00 and 9:10 on May 30, 2025. After pre-processing, valid data segments are separated (such as continuous heart rate data from 9:02 to 9:08). The data format is a structured record containing timestamps and values ​​(such as "2025-05-30 09:02:00, 78 beats / minute", "2025-05-30 09:02:01, 76 beats / minute", etc.).

[0103] When using SQLite as the local database, the system creates a table named "physiological_data" with fields such as "timestamp" (text type), "parameter_type" (text type, such as "heart_rate"), "value" (floating point value), and "segment_status" (text type, such as "valid"). Each valid data record after preprocessing is inserted into the table using the SQL statement "INSERT INTOphysiological_data(timestamp, parameter_type, value, segment_status) VALUES(?,?,?,?)" to ensure chronological data order. If using MySQL, establish a connection pool to manage database connections to improve data writing efficiency in high-concurrency scenarios.

[0104] In terms of cloud backup, the system establishes a connection with the cloud server through an SSL / TLS encrypted channel. Taking the AWS S3 cloud storage as an example, after AES-256 encryption in the local, the data is uploaded to the designated storage bucket through the PUT request. The encryption key is randomly generated by the system and stored in the local secure key library, and is dynamically called each time of uploading. For example, the 10 minutes of data processed at 9:10 is encrypted to generate a file name with timestamp "202505300900_heart_rate_encrypted.dat", and uploaded to the cloud path " / youth_depression_monitor / data / 2025 / 05 / 30 / ", and at the same time, the cloud storage path is recorded in the local database for subsequent query and traceability.

[0105] Implementation of the early warning feedback module: When the physiological state is abnormal, the early warning feedback module sends a hierarchical early warning signal to the bound monitoring terminal through Bluetooth. Assuming that the system analyzes the heart rate data from 9:02 to 9:08, it finds that the abnormal waveform exceeds the preset fluctuation range for 6 minutes (more than the first preset threshold of 5 minutes), and detects non-periodic fluctuation, and determines that the physiological state in this period is abnormal.

[0106] The system first determines the warning level: single abnormal fluctuation triggers level one warning (yellow), persistent abnormal fluctuation (such as abnormal for two hours in a row) triggers level two warning (orange), and continuous three-day abnormal area distribution density change rate exceeding threshold triggers level three warning (red). In this example, because it is a single abnormality, level one warning is triggered.

[0107] The early warning signal is transmitted to the monitoring terminal (such as a parent's mobile phone) through Bluetooth Low Energy (BLE) protocol. The matching APP installed on the mobile phone needs to be paired with the system in advance, and after successful pairing, the system sends the early warning data through the characteristic value in the GATT (Generic Attribute Profile) service. The early warning data format is a JSON string, which includes:

[0108] "alarm_level": warning level ("level1" "level2" "level3")

[0109] "parameter_type": abnormal parameter type ("heart_rate")

[0110] "abnormal_time": abnormal time ("2025-05-3009:02:00 to 2025-05-3009:08:00")

[0111] "description": abnormal description ("heart rate fluctuation exceeds the preset range for 6 minutes")

[0112] After receiving the signal, the APP immediately pops up a warning notification, showing the warning level, abnormal parameters and time, accompanied by vibration and sound reminders. The notification interface of the first-level warning is highlighted in yellow, showing "single physiological abnormality reminder", the second-level warning is orange and prompts "need to pay attention to continuous abnormal fluctuations", and the third-level warning is red and marked "emergency: continuous abnormal exacerbation".

[0113] Embodiment of user interaction module: the user interaction module receives the query instruction of the monitoring terminal, calls local or cloud data and returns the analysis result. Taking the query of the heart rate data of a teenager on May 30, 2025 by the parent as an example, the parent selects the query time range "2025-05-30 00:00:00 to 2025-05-30 23:59:59" and specifies the parameter type "heart rate" in the APP of the mobile phone, and clicks "query". After that, the APP sends an instruction to the system through Bluetooth, and the instruction format is as follows:

[0114] {

[0115] "command":"query",

[0116] "time_range":["2025-05-30 00:00:00","2025-05-30 23:59:59"],

[0117] "parameter":"heart_rate"

[0118] }

[0119] After receiving the instruction, the system first queries whether there is heart rate data in the local database within the time range. If the local data is complete (such as containing all valid data segments), it is directly called; if the local data is missing (such as not storing data in a certain time period due to device failure), the corresponding encrypted data is downloaded from the cloud server through an encrypted channel, and then loaded after decryption using the local key.

[0120] After the data is called, the system generates the analysis result:

[0121] Basic statistics: the total effective data duration is 22 hours, and the invalid data duration is 2 hours; the average heart rate is 78 times per minute, the highest is 95 times per minute (which occurs at 14:30-14:35), and the lowest is 62 times per minute (which occurs at 22:00-22:05).

[0122] Abnormal event summary: 3 first-level warnings are detected (at 9:02-9:08, 11:15-11:20 and 16:45-16:50, respectively), and there is no second-level or third-level warning.

[0123] Trend chart: Generate a 24-hour heart rate fluctuation line chart, with the baseline value of 75 beats per minute as the center line, mark the abnormal waveform range with a shaded area (such as 9:02-9:08 during the curve exceeds the positive baseline point 87.5 times per minute), and mark the time period corresponding to each warning level with different colors.

[0124] The analysis results are returned to the monitoring terminal in JSON format, and the APP presents them to the parents: a warning summary card is displayed at the top of the interface (yellow mark 3 times of first-level warning), the middle is an interactive line chart, and the bottom is a table listing the specific time, duration, and fluctuation amplitude of each abnormality. Parents can click on a time period on the chart to view detailed data points, or export a PDF report for medical reference.

[0125] Module coordination application example: Suppose a teenager has continuous heart rate abnormal fluctuations from 14:00 to 16:00 on May 30th, the system detects three abnormal periods of 14:30-14:35, 15:10-15:20, and 15:45-15:50 through the single behavior assessment module, and the intervals are all less than the preset time difference t2 (such as 15 minutes), which is determined as an abnormal continuous state, triggering a second-level warning. The data storage module stores the original data and analysis results of these three periods in local and cloud storage, and the warning feedback module sends an orange warning notification to the parents, prompting "heart rate has been continuously abnormal from 14:30 to 15:50 today, please pay attention to the teenager's emotional state". After receiving the notification, the parents query the detailed data of this time period through the user interaction module, the system retrieves the encrypted data stored in the cloud, generates a continuous physiological state chart containing abnormal waveform superposition, and displays the abnormal area distribution density change rate as 0.8 (higher than the average value of 0.3, with a difference of 0.5 greater than the preset 0.4), which helps parents judge the severity of the abnormality and take timely communication or professional intervention measures.

[0126] Key implementation details:

[0127] Data encryption: The original data stored locally and the data packets transmitted in the cloud are all encrypted using AES-256, and the key is automatically updated every 7 days to ensure data security.

[0128] Bluetooth stability: Warning signals and query instructions are transmitted through the notification feature (Notification) of BLE, supporting the disconnection and reconnection mechanism to ensure that the communication success rate is higher than 95% in complex environments.

[0129] Cross-platform compatibility: The user interaction module supports multiple monitoring terminals (iOS, Android phones and tablets), and realizes data interaction through a unified API interface, with consistent interface style and operation logic.

[0130] Storage strategy: the local database retains data for the last 30 days, and data beyond the deadline is automatically archived to cloud cold storage, reducing local storage pressure while meeting long-term data tracing needs.

[0131] Through reliable data management of the data storage module, real-time hierarchical response of the early warning feedback module, and convenient data query of the user interaction module, embodiment 5 constructs a complete monitoring-early warning-intervention closed loop, so that the system can not only accurately identify physiological abnormalities, but also provide timely and comprehensive information support for guardians through multi-module cooperation, thereby improving the early intervention efficiency of adolescent depression and self-injury behavior.

[0132] It should be noted that, in this article, relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the term "includes" "contains" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment.

[0133] Although the embodiments of the present application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. An intelligent monitoring system for intervening in adolescent depression and self-harm behavior, characterized by: include: The data acquisition module is used to obtain physiological parameter data collected by the smart bracelet worn by the user through Bluetooth transmission, and pre-process the physiological parameter data to divide it into valid data segments and invalid data segments; A data analysis module, configured to identify fluctuation characteristics of the valid data segments and normalize the physiological parameter data according to a preset reference value; A single behavior assessment module is used to select a physiological parameter change curve within any time period, extract the peak points and valley points of each fluctuation cycle of the physiological parameter change curve based on a preset algorithm, mark them as key points, obtain several key points with the largest deviation from the baseline value, define them as abnormal points, connect adjacent abnormal points in sequence to generate an abnormal waveform, obtain the duration of the abnormal waveform exceeding the preset fluctuation range, and when the duration is greater than a first preset threshold, identify whether it is a periodic fluctuation or a sudden fluctuation. If not, it is determined that the physiological state of the selected time period is abnormal; If so, the intervals between adjacent peaks and troughs of the abnormal waveform are detected, and if the intervals are greater than a preset interval, it is determined that the physiological response in the selected time period is abnormal; The continuous behavior assessment module is used to select physiological parameter data for 24 consecutive hours, vertically expand the physiological parameter curve of each preset time period with the baseline as the center line until the abnormal waveforms in adjacent time periods overlap, generate continuous abnormal physiological state areas, and calculate the rate of change of the distribution density of the abnormal areas. When the difference between the rate of change and the average value of the rate of change is greater than a second preset threshold, it is judged that the physiological state of the continuous 24-hour day is continuously abnormal.

2. The intelligent monitoring system for facilitating intervention in adolescent depression and self-harm behavior according to claim 1, characterized in that: The process of the pretreatment is: The physiological parameter data is smoothed to generate a standardized curve, the fluctuation recognition of the standardized curve is increased, and the standardized curve is converted into segmented labeled data by setting an amplitude threshold. The segmented labeled data is noise filtered based on a sliding window, and feature extraction is performed on the filtered segmented labeled data to separate valid data segments from invalid data segments.

3. The intelligent monitoring system for intervening in adolescent depression and self-harm behavior according to claim 1, characterized in that: The preset fluctuation range is composed of 6 reference points, among which 2 reference points are set in the positive fluctuation range of the reference value, and the other 4 reference points are set in the negative fluctuation range of the reference value. The two reference points in the positive range are positioned at the middle value of the upper limit of the regular fluctuation; two of the reference points in the negative range are positioned at the warning line of the lower limit of the regular fluctuation, that is, the boundary points, and the remaining two reference points in the negative range are positioned at the two trisection points of the line connecting the two boundary points.

4. The intelligent monitoring system for intervening in adolescent depression and self-harm behavior according to claim 1, characterized in that: The process of normalizing physiological parameter data is as follows: The fluctuation range of the effective data segment is obtained, the deviation degree between the physiological parameter curve and the baseline of different time periods is calculated respectively, and the physiological parameter data is amplitude corrected according to the average value of the deviation degree.

5. The intelligent monitoring system for intervening in adolescent depression and self-harm behavior according to claim 1, characterized in that: The process of detecting the intervals between adjacent peaks and troughs of a waveform is: Traverse the physiological parameter curve according to the time sequence, the time sequence is a forward time sequence or a reverse time sequence, and count the fluctuation amplitude of the detected data for each minute, the amplitude value of the abnormal fluctuation is 1, and the amplitude value of the normal fluctuation is 0, select the first data segment with an amplitude value of 1 and a duration greater than a preset duration t, mark it as the abnormal starting segment, continue to traverse the physiological parameter curve, when a data segment with an amplitude value of 0 and a duration greater than t is detected, mark it as a transition segment, obtain the time difference T1 between the transition segment and the abnormal starting segment, if the time difference T1 is greater than or equal to the preset time difference t1, and continue to detect along the time sequence and no data segment with an amplitude value of 1, then mark the transition segment as the abnormal ending segment, and the abnormality is an isolated event; If the time difference is less than t1, continue to traverse the physiological parameter curve. When a data segment with an amplitude value of 1 and a duration greater than t is detected again, it is marked as the second abnormal starting segment. The time difference T2 between the second abnormal starting segment and the transition segment is obtained. If the time difference T2 is greater than the preset time difference t2, it is judged that the physiological reaction is in an abnormal continuous state.

6. The intelligent monitoring system for intervening in adolescent depression and self-harm behavior according to claim 5, characterized in that: The specific process of judging persistent abnormal physiological status is as follows: For any selected time period, select the horizontal line corresponding to the baseline value of the time period, obtain the deviation values ​​of all data points in the time period relative to the horizontal line, multiply the deviation value of each data point relative to the horizontal line by the same coefficient, and continuously increase the value of the coefficient until adjacent time periods overlap in pairs to obtain a continuous area of ​​abnormal fluctuations. Obtain the corresponding distribution density change curve for the continuous area through mean filtering, and obtain the change rate of the curve.

7. The intelligent monitoring system for facilitating intervention in adolescent depression and self-harm behavior according to claim 1, characterized in that: The process of generating abnormal waveforms is: First, select the three abnormal points with the largest deviation from the baseline value, connect the adjacent abnormal points to generate the pending abnormal waveform, and then select the points with the largest deviation from the remaining abnormal points one by one and add them to the connection line of the abnormal waveform until the amplitude of the waveform corresponding to the original abnormal point in the connection line formed by the newly added abnormal point becomes smaller. In this case, the added abnormal point is removed, and the connection line formed by the original abnormal point is the abnormal waveform of this period.

8. The intelligent monitoring system for facilitating intervention in adolescent depression and self-harm behavior according to claim 1, characterized in that: Also includes: The data storage module is used to store the pre-processed physiological parameter data in the local database in chronological order and upload it to the cloud server for backup through an encrypted channel.

9. The intelligent monitoring system for facilitating intervention in adolescent depression and self-harm behavior according to claim 1, characterized in that: Also includes: The early warning feedback module is used to send a graded early warning signal to the bound monitoring terminal via Bluetooth transmission when an abnormal physiological state or abnormal physiological reaction is detected.

10. The intelligent monitoring system for intervening in adolescent depression and self-harm behavior according to claim 1, characterized in that: Also includes: The user interaction module is used to receive query instructions sent by the monitoring terminal and retrieve the corresponding physiological parameter data analysis results from the local database or cloud server according to the instruction content.

Images