Method and system for analyzing trend of psychological state changes based on multidimensional physiological information

Abstract

The present invention proposes a method and system for analyzing the trend of changes in psychological state based on multidimensional physiological information. The method for analyzing the trend of changes in psychological state based on multidimensional physiological information includes: collecting physiological index parameters of users, classifying the physiological index parameters of users, and obtaining multiple data sets corresponding to the physiological index parameters; performing multidimensional intervention data processing on the multiple data sets corresponding to the physiological index parameters, and obtaining multiple data sets after multidimensional intervention data processing; using a deep learning model with a function of analyzing the trend of changes in psychological state to perform data analysis on the multiple data sets, and obtaining a psychological state change trend analysis report. The system includes modules corresponding to the steps of the method.

Description

Technical Field

[0001] The invention proposes a method and system for analyzing the trend of psychological state changes based on multi-dimensional physiological information, and belongs to the technical field of multi-dimensional information processing. Background Art

[0002] With the rapid development of science and technology, people's awareness and monitoring of changes in mental state are increasing. Changes in mental state not only affect the daily quality of an individual, but are also closely related to the occurrence and development of many diseases. Therefore, accurate and effective monitoring and analysis of mental state change trends are of great significance for the prevention, diagnosis and treatment of mental-related diseases.

[0003] In the prior art, monitoring of changes in mental state mainly relies on methods such as self-report, questionnaires, and clinical assessments. However, the above methods have many limitations. First, self-reports and questionnaires are easily affected by subjective factors, such as emotional state, memory bias, etc., resulting in inaccurate results. Second, although clinical assessment is more professional, it can usually only be performed in a specific medical environment and is costly, making it unsuitable for large-scale or long-term mental state monitoring.

[0004] In recent years, with the advancement of sensor technology and computer science, psychological state monitoring methods based on physiological information have gradually attracted attention. This type of method infers changes in an individual's psychological state by collecting and analyzing their physiological signals (such as heart rate, respiration, skin conductance, etc.). Compared with traditional self-reports and questionnaires, monitoring methods based on physiological information have the advantages of being objective, real-time, and continuous. However, existing psychological state monitoring methods based on physiological information still have the problems of poor accuracy and low efficiency in multidimensional data processing in the existing technology. Summary of the invention

[0005] The present invention provides a method and system for analyzing the trend of psychological state changes based on multidimensional physiological information, which is used to solve the problems of poor accuracy and low efficiency of multidimensional data processing in the prior art. The technical solutions adopted are as follows:

[0006] A method for analyzing the trend of changes in psychological states based on multidimensional physiological information, the method comprising:

[0007] Collecting physiological index parameters of the user, and classifying the physiological index parameters of the user to obtain multiple data sets corresponding to the physiological index parameters;

[0008] Performing multi-dimensional intervention data processing on the multiple data sets corresponding to the physiological indicator parameters to obtain multiple data sets after the multi-dimensional intervention data processing;

[0009] The multiple data sets are analyzed using a deep learning model with a psychological state change trend analysis function to obtain a psychological state change trend analysis report.

[0010] Furthermore, the physiological index parameters of the user are collected and classified to obtain multiple data sets corresponding to the physiological index parameters, including:

[0011] Retrieving physiological index parameters with the same collection period as that of the user from a database;

[0012] Classifying the physiological indicator parameters according to their types to obtain a plurality of data sets corresponding to the physiological indicator parameters; wherein the types of the physiological indicator parameters include heart rate information, respiratory rate, and skin electrical response data; and the data sets correspond to the types of the physiological indicator parameters one by one;

[0013] The multiple data sets are subjected to data preprocessing and preprocessing records to obtain multiple data sets after preprocessing; wherein the data preprocessing includes data cleaning processing, data noise reduction processing, and outlier and missing value processing.

[0014] Furthermore, performing multi-dimensional intervention data processing on the multiple data sets corresponding to the physiological indicator parameters to obtain multiple data sets after the multi-dimensional intervention data processing includes:

[0015] Performing multi-dimensional intervention data processing on the abnormal data of the multiple data sets corresponding to the physiological indicator parameters to obtain a data set after the abnormal data corresponding to each data set is replaced;

[0016] Multidimensional intervention data processing is performed on the missing data of multiple data sets corresponding to the physiological indicator parameters to obtain a data set after the missing data corresponding to each data set is filled.

[0017] Furthermore, multi-dimensional intervention data processing is performed on the abnormal data of the multiple data sets corresponding to the physiological indicator parameters to obtain a data set after the abnormal data corresponding to each data set is replaced, including:

[0018] Obtaining data values ​​of abnormal data in each data set and abnormal data collection time nodes corresponding to the abnormal data from the data preprocessing records;

[0019] Performing multi-dimensional intervention data processing on the data values ​​of the abnormal data in each data set in combination with the abnormal data collection time node to obtain abnormal replacement data values ​​corresponding to the abnormal data;

[0020] The abnormal data is replaced by using the abnormal replacement data value to obtain a data set after the abnormal data corresponding to each data set is replaced.

[0021] Furthermore, in combination with the abnormal data collection time node, multi-dimensional intervention data processing is performed on the data value of the abnormal data in each data set to obtain the abnormal replacement data value corresponding to the abnormal data, including:

[0022] Extracting the abnormal data collection time node of the abnormal data contained in each data set;

[0023] Extract normal data values ​​corresponding to two normal data collection time nodes with the closest collection intervals before and after the abnormal data collection time node;

[0024] The first abnormal data adjustment parameter is obtained by using the abnormal data and the normal data value; wherein the first abnormal data adjustment parameter is obtained by the following formula:

[0025]

[0026] Among them, f 01 represents the first abnormal data adjustment parameter; X e Indicates the preset data deviation reference value; X y Data value representing abnormal data; X 01 and X 02 Respectively represent the normal data values ​​corresponding to two normal data collection time nodes;

[0027] Obtaining a second abnormal data adjustment parameter using data at a time corresponding to the abnormal data collection time node in other data sets;

[0028] The abnormal replacement data value corresponding to the abnormal data is obtained by using the first abnormal data adjustment parameter and the second abnormal data adjustment parameter in combination with the data value of the abnormal data; wherein the abnormal replacement data value corresponding to the abnormal data is obtained by the following formula:

[0029]

[0030] Among them, X t represents the abnormal replacement data value corresponding to the abnormal data; f 01 represents the first abnormal data adjustment parameter; f 02 represents the second abnormal data adjustment parameter; m represents the number of normal data in the data set corresponding to the abnormal data; X j Represents the data value corresponding to the j-th normal data.

[0031] Further, obtaining a second abnormal data adjustment parameter by using data at a time corresponding to the abnormal data collection time node in other data sets includes:

[0032] Extracting data at a time corresponding to the abnormal data collection time node from other data sets as first data information;

[0033] Retrieving data from the other data sets at adjacent times before and after the abnormal data collection time node as second data information;

[0034] The second abnormal data adjustment parameter is obtained by using the first data information and the second data information corresponding to the other data sets; wherein the second abnormal data adjustment parameter is obtained by the following formula:

[0035]

[0036]

[0037] Among them, f 02 represents the second abnormal data adjustment parameter; λ represents the abnormal data compensation coefficient; n represents the number of other data sets; X i represents the data value corresponding to the first data information in the i-th data set; X upi and X downi represents the upper and lower limits of the standard parameter range corresponding to the type of physiological index parameter corresponding to the i-th other data set; X ci represents the data median corresponding to the normal data range of the preset i-th data set; X 01i and X 02i They respectively represent the data values ​​of the data at the adjacent moments before and after the abnormal data collection moment node of the i-th other data set.

[0038] Furthermore, multi-dimensional intervention data processing is performed on the missing data of the multiple data sets corresponding to the physiological index parameters to obtain a data set after the missing data corresponding to each data set is filled, including:

[0039] Obtain the missing data collection time node corresponding to the missing data in each data set from the data preprocessing record;

[0040] Perform multi-dimensional intervention data processing on the missing data in each data set in combination with the missing data collection time node to obtain missing data filling values ​​corresponding to the missing data;

[0041] The missing data positions are filled with the missing data filling values ​​to obtain a missing data-filled data set corresponding to each data set.

[0042] Furthermore, in combination with the missing data collection time node, multi-dimensional intervention data processing is performed on the missing data in each data set to obtain missing data filling values ​​corresponding to the missing data, including:

[0043] Extract the missing data collection time node of the missing data contained in each data set;

[0044] Extract normal data values ​​corresponding to two normal data collection time nodes with the closest collection intervals before and after the missing data collection time node;

[0045] The first missing data parameter is obtained by using the missing data and the normal data value; wherein the first missing data parameter is obtained by the following formula:

[0046]

[0047] Among them, h 01 represents the first missing data parameter; Y c Indicates the median value of the data distribution in the data set corresponding to the missing data; Y 01 and Y 02 Respectively represent the normal data values ​​corresponding to the two normal data collection time nodes with the closest collection intervals before and after the missing data collection time node;

[0048] Extracting data at a time corresponding to the missing data collection time node from other data sets as first reference data information;

[0049] Retrieving data from the other data sets at adjacent times before and after the missing data collection time node as second reference data information;

[0050] The second missing data parameter is obtained by using the first missing data information and the second missing data information corresponding to the other data set; wherein the second missing data parameter is obtained by the following formula:

[0051]

[0052]

[0053] Among them, h 02 represents the second missing data parameter; n represents the number of other data sets; ξ represents the missing data compensation coefficient; Y ci represents the median value of the data distribution corresponding to the i-th other data set; Y zi The data median of the standard parameter range of the type of physiological indicator parameter corresponding to the i-th other data set; Y 01i and Y 02i They respectively represent the normal data values ​​corresponding to the two normal data collection time nodes with the closest collection interval before and after the missing data collection time node corresponding to the i-th other data set; X upi and X downi Indicates the upper limit and lower limit of the standard parameter range corresponding to the type of the physiological indicator parameter corresponding to the i-th other data set;

[0054] The missing data filling value is obtained by using the first missing data parameter and the second missing data parameter; wherein the missing data filling value is obtained by the following formula:

[0055]

[0056] Among them, Y t represents the missing data filling value; h 01 represents the first missing data parameter; h 02 Indicates the second missing data parameter; Y 01 and Y 02 They respectively represent the normal data values ​​corresponding to the two normal data collection time nodes with the closest collection intervals before and after the missing data collection time node.

[0057] Further, a deep learning model with a psychological state change trend analysis function is used to perform data analysis on the multiple data sets to obtain a psychological state change trend analysis report, including:

[0058] Performing standardization processing on the multiple data sets to obtain each standardized data set;

[0059] Inputting each of the standardized data sets into a deep learning model with a psychological state change trend analysis function, and outputting the analysis result through the deep learning model, wherein the analysis result is a psychological state change trend analysis report;

[0060] The psychological state change trend analysis report is visualized.

[0061] A psychological state change trend analysis system based on multi-dimensional physiological information, the psychological state change trend analysis system based on multi-dimensional physiological information comprises:

[0062] A data collection and processing module is used to collect the user's physiological index parameters, and classify the user's physiological index parameters to obtain multiple data sets corresponding to the physiological index parameters;

[0063] A multidimensional intervention data processing module, used for performing multidimensional intervention data processing on the multiple data sets corresponding to the physiological index parameters, and obtaining multiple data sets after the multidimensional intervention data processing;

[0064] The data analysis and processing module is used to input the multiple data sets processed by the multi-dimensional intervention data into a deep learning model with a psychological state change trend analysis function, and to visualize and generate a report on the psychological state change trend analysis results output by the deep learning model.

[0065] Beneficial effects of the present invention:

[0066] The method and system for analyzing the trend of changes in psychological state based on multidimensional physiological information proposed in the present invention analyzes multidimensional physiological information through a deep learning model, and can more accurately identify and predict the trend of changes in the psychological state of the user, with higher accuracy than the traditional method based on questionnaires or interviews. Since the above technical scheme is based on physiological indicator data under long-term monitoring, it can reflect the changes in the psychological state of the user in real time, providing a basis for timely discovery and intervention of psychological problems. By classifying and processing the physiological indicator data of different users, personalized psychological state analysis can be achieved, providing more accurate mental health services for each user. By continuously monitoring and analyzing the trend of changes in the psychological state of the user, potential psychological problems can be discovered in a timely manner, providing a scientific basis for psychological intervention and prevention, and helping to reduce the incidence of mental illness. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 is a flow chart of the method of the present invention;

[0068] Figure 2 The system block diagram of the system of the present invention. DETAILED DESCRIPTION

[0069] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0070] A method for analyzing the trend of changes in psychological states based on multidimensional physiological information, the method comprising:

[0071] S1. Collect the physiological index parameters of the user, and classify the physiological index parameters of the user to obtain multiple data sets corresponding to the physiological index parameters; wherein the physiological index parameters of the user are the physiological index parameters monitored and obtained by a body physiological index monitoring instrument under long-term monitoring, and the physiological index parameters are stored in a database.

[0072] S2, performing multi-dimensional intervention data processing on the multiple data sets corresponding to the physiological indicator parameters to obtain multiple data sets after the multi-dimensional intervention data processing;

[0073] S3. Utilize a deep learning model with a psychological state change trend analysis function to perform data analysis on the multiple data sets to obtain a psychological state change trend analysis report.

[0074] The working principle of the above technical solution is: through the long-term monitoring of the body physiological index monitoring instrument, the user's physiological index parameters, such as heart rate, blood pressure, galvanic skin response (GSR), breathing rate, etc., are continuously collected. The collected physiological index parameters are classified to form multiple data sets according to different physiological indicators or time periods. The above data sets are stored in the database for subsequent analysis and processing.

[0075] Multidimensional intervention data processing is performed on each data set, including data cleaning (removing noise and outliers), data standardization (enabling data of different dimensions to be compared), and data enhancement (increasing the amount of data through simulation or other means). The goal of multidimensional intervention data processing is to ensure the data quality of subsequent analysis and improve the accuracy and reliability of the analysis.

[0076] Use a pre-trained deep learning model with the function of analyzing the trend of psychological state changes to analyze multiple processed data sets. The deep learning model can automatically learn complex patterns and features in the data to predict or identify the trend of the user's psychological state changes. Through the analysis of the model, a psychological state change trend analysis report is generated, which includes the fluctuation of the user's psychological state, psychological problems or suggestions, etc.

[0077] The effect of the above technical solution is: by analyzing multi-dimensional physiological information through a deep learning model, it is possible to more accurately identify and predict the changing trend of the user's psychological state, which has higher accuracy than the traditional questionnaire-based or interview-based method. Since the above technical solution is based on physiological indicator data under long-term monitoring, it can reflect the changes in the user's psychological state in real time, providing a basis for timely discovery and intervention of psychological problems. By classifying and processing the physiological indicator data of different users, personalized psychological state analysis can be achieved, providing more accurate mental health services for each user. By continuously monitoring and analyzing the changing trends of the user's psychological state, potential psychological problems can be discovered in a timely manner, providing a scientific basis for psychological intervention and prevention, and helping to reduce the incidence of mental illness.

[0078] In one embodiment of the present invention, physiological index parameters of a user are collected and classified to obtain multiple data sets corresponding to the physiological index parameters, including:

[0079] S101, retrieving physiological index parameters with the same collection period as that of the user from a database;

[0080] S102, classifying the physiological indicator parameters according to the types of the physiological indicator parameters to obtain a plurality of data sets corresponding to the physiological indicator parameters; wherein the types of the physiological indicator parameters include heart rate information, respiratory rate and skin electrical response data; and the data sets correspond to the types of the physiological indicator parameters one by one;

[0081] S103, performing data preprocessing and preprocessing records on the multiple data sets to obtain multiple data sets after preprocessing; wherein the data preprocessing includes data cleaning processing, data noise reduction processing, and outlier and missing value processing.

[0082] The working principle of the above technical solution is as follows: first, the physiological index parameters corresponding to the specific user and the specified collection period are retrieved from the pre-established database. The above parameters are continuously collected by the body physiological index monitoring instrument within the specified time period.

[0083] The retrieved physiological index parameters are classified according to their types. For example, heart rate information, respiratory rate, and skin galvanic response data are classified into different categories. Each physiological index parameter will form an independent data set to ensure that the data set corresponds to the type of physiological index parameter.

[0084] Data preprocessing is performed on multiple classified data sets to improve data quality and ensure the accuracy of subsequent analysis. Data preprocessing includes multiple steps, such as data cleaning (removing duplicate, invalid or erroneous data), data noise reduction (reducing or eliminating noise interference in the data), and processing of outliers and missing values ​​(such as through interpolation or elimination). At the same time, during the preprocessing process, the corresponding preprocessing operations will be recorded to facilitate tracking and review of the data processing process.

[0085] The effect of the above technical solution is: through data cleaning, noise reduction and outlier processing, the overall quality of the data can be effectively improved and the analysis errors caused by data problems can be reduced. Classifying the data according to the types of physiological indicator parameters and preprocessing them separately can ensure that each parameter is processed in a targeted manner and improve the precision and accuracy of the processing. Recording each step of the preprocessing process will help to review and verify the data processing process later to ensure the reliability and repeatability of the analysis results. The preprocessed data set will provide high-quality data support for the subsequent analysis of psychological state change trends, which will help improve the accuracy and reliability of the analysis. By classifying and preprocessing the physiological indicator parameters of different users, a more personalized psychological state analysis can be provided for each user to meet the needs of different users.

[0086] In one embodiment of the present invention, multi-dimensional intervention data processing is performed on multiple data sets corresponding to the physiological indicator parameters to obtain multiple data sets after multi-dimensional intervention data processing, including:

[0087] S201, performing multi-dimensional intervention data processing on abnormal data of multiple data sets corresponding to the physiological index parameters, and obtaining a data set after the abnormal data corresponding to each data set is replaced;

[0088] S202, performing multi-dimensional intervention data processing on the missing data of the multiple data sets corresponding to the physiological indicator parameters, and obtaining a data set after the missing data corresponding to each data set is filled.

[0089] The working principle of the above technical solution is: in the data set, abnormal data is caused by equipment failure, abnormal user behavior or other reasons. If the above abnormal data is not processed, it will have a negative impact on the subsequent analysis results. The system will identify the abnormal data in each data set and apply multi-dimensional intervention data processing technology. After processing, each data set will get a new version after the abnormal data is replaced. Among them, the abnormal data refers to the data value in the physiological information data set that has a significant and unreasonable deviation from most of the data of objective and reasonable vital signs. The abnormal data is often caused by equipment failure and / or data collection errors.

[0090] At the same time, missing data is another common problem in data sets. The above missing values ​​are caused by reasons such as the failure of the device to successfully record data and the failure of the user to cooperate. The system will identify the missing data in each data set and apply multi-dimensional intervention data processing technology to fill the above missing values. Filling methods include using statistical values ​​such as mean, median, mode, or predicted values ​​based on machine learning algorithms. After processing, each data set will get a new version with missing data filled.

[0091] The effect of the above technical solution is: by multi-dimensional intervention processing of abnormal data and missing data, the overall quality of the data set can be effectively improved. The above technical solution helps to reduce analysis errors caused by data problems and improve the accuracy of subsequent analysis of psychological state change trends.

[0092] Enhanced reliability of analysis: The processed data set is more complete and accurate, and can provide reliable data support for subsequent analysis of psychological state change trends. The above technical solution helps to ensure the reliability and effectiveness of the analysis results. Multidimensional intervention data processing technology can flexibly adjust the processing method according to different data sets and different data problems. This makes the technical solution highly adaptable and flexible, and can cope with various complex data situations. The data set processed by multidimensional intervention data will provide high-quality data support for subsequent analysis of psychological state change trends. The above technical solution helps to more accurately identify the user's psychological state change trend and provide a scientific basis for mental health monitoring and intervention.

[0093] In one embodiment of the present invention, multi-dimensional intervention data processing is performed on the abnormal data of multiple data sets corresponding to the physiological indicator parameters to obtain a data set after the abnormal data corresponding to each data set is replaced, including:

[0094] S2011, obtaining the data value of the abnormal data in each data set and the abnormal data collection time node corresponding to the abnormal data from the data preprocessing record;

[0095] S2012, performing multi-dimensional intervention data processing on the data values ​​of the abnormal data in each data set in combination with the abnormal data collection time node, to obtain abnormal replacement data values ​​corresponding to the abnormal data;

[0096] S2013: Replace the abnormal data with the abnormal replacement data value to obtain a data set corresponding to each data set after the abnormal data has been replaced.

[0097] The working principle of the above technical solution is: in the data preprocessing stage, the system will clean and reduce noise on the data, and record the abnormal data. In the above technical solution, the system will obtain the data value of the abnormal data in each data set and the corresponding abnormal data collection time node from the data preprocessing record. The above information is crucial for subsequent data processing because it can help the system identify and process abnormal data more accurately.

[0098] After obtaining the data value and collection time node of the abnormal data, the system will perform multi-dimensional intervention data processing on the above abnormal data. The processing method of the above technical solution includes a variety of strategies, such as statistical-based replacement (such as using the median, mode, etc. to replace abnormal values), prediction replacement based on time series analysis (such as using historical data or adjacent data points to predict and replace abnormal values), prediction replacement based on machine learning (such as using trained models to predict and replace abnormal values), etc. The "multi-dimensional" here means that the system not only considers the data value itself, but also combines other relevant information such as the abnormal data collection time node (ie, time dimension), adjacent data points, etc. for processing. This can make the processing results more accurate and reasonable.

[0099] After obtaining the abnormal replacement data value, the system will use the above value to replace the abnormal data in the original data set. In this way, each data set will get a new version with more accurate and reasonable data values ​​after the abnormal data is replaced.

[0100] The effect of the above technical solution is: by multi-dimensional intervention data processing and replacement of abnormal data, the overall quality of the data set can be significantly improved. The existence of abnormal data will have a negative impact on data analysis and mining, so timely identification and processing of the above abnormal data is the key to ensuring data quality. Using the processed data set for subsequent psychological state change trend analysis can improve the accuracy of the analysis. Because the existence of abnormal data will cause deviations or misleading analysis results, the replaced data set is more accurate and reasonable, and can better reflect the real situation of the user.

[0101] This technical solution can automatically identify and process abnormal data in the data set, improving the robustness of the system. Even if data anomalies occur in actual applications, the system can respond quickly and give reasonable processing results, ensuring the stability and reliability of the system. The processed data set provides high-quality data support for subsequent analysis of psychological state change trends. The above technical solution helps to more accurately identify the user's psychological state change trend and provide a scientific basis for mental health monitoring and intervention.

[0102] In one embodiment of the present invention, multi-dimensional intervention data processing is performed on the data values ​​of the abnormal data in each data set in combination with the abnormal data collection time node to obtain the abnormal replacement data value corresponding to the abnormal data, including:

[0103] Step 1: extracting the abnormal data collection time node of the abnormal data contained in each data set;

[0104] Step 2: extract the normal data values ​​corresponding to the two normal data collection time nodes with the closest collection interval before and after the abnormal data collection time node; wherein the normal data values ​​are physiological indicator parameters that meet the rationality of the appearance of vital signs, specifically including physiological indicator parameters that meet the prescribed range and physiological indicator parameters that do not meet the prescribed range;

[0105] Step 3: Obtain a first abnormal data adjustment parameter using the abnormal data and the normal data value; wherein the first abnormal data adjustment parameter is obtained by the following formula:

[0106]

[0107] Among them, f 01 represents the first abnormal data adjustment parameter; X e Indicates the preset data deviation reference value; X y Data value representing abnormal data; X 01 and X 02 Respectively represent the normal data values ​​corresponding to two normal data collection time nodes;

[0108] Step 4: using data at a time corresponding to the abnormal data collection time node in other data sets to obtain a second abnormal data adjustment parameter;

[0109] Step 5: Utilize the first abnormal data adjustment parameter and the second abnormal data adjustment parameter in combination with the data value of the abnormal data to obtain the abnormal replacement data value corresponding to the abnormal data; wherein the abnormal replacement data value corresponding to the abnormal data is obtained by the following formula:

[0110]

[0111] Among them, X t represents the abnormal replacement data value corresponding to the abnormal data; f 01 represents the first abnormal data adjustment parameter; f 02 represents the second abnormal data adjustment parameter; m represents the number of normal data in the data set corresponding to the abnormal data; X j Represents the data value corresponding to the j-th normal data.

[0112] The working principle of the above technical solution is as follows: First, the system extracts the abnormal data collection time node containing abnormal data from each data set. The above time node is the time point when the abnormal data occurs and is the key basis for subsequent processing.

[0113] Next, the system will find the two normal data collection time nodes with the closest collection interval before and after the abnormal data collection time node. The normal data values ​​corresponding to these two time nodes will be extracted as the basis for subsequent calculation and adjustment parameters. The normal data values ​​here are physiological indicator parameters that meet the rationality of the appearance of vital signs, including data within the specified range and data that exceeds the range but can be considered normal.

[0114] Using the abnormal data and the two extracted normal data values, the system will calculate the first abnormal data adjustment parameter through a preset formula. The above parameter reflects the degree of deviation between the abnormal data and the normal data, and is an important basis for subsequent adjustment of the abnormal data.

[0115] In addition to the information in the current data set, the system also uses the data in other data sets corresponding to the abnormal data collection time node to obtain the second abnormal data adjustment parameter. The above technical solution is to consider the correlation between cross-data sets to make the adjustment more comprehensive and accurate.

[0116] Finally, the system will combine the first abnormal data adjustment parameter, the second abnormal data adjustment parameter and the data value of the abnormal data itself, and calculate the abnormal replacement data value corresponding to the abnormal data through a preset formula. The above replacement data value takes into account both the information in the current data set and the correlation between cross-data sets, and is an adjustment result that combines multiple factors.

[0117] The effect of the above technical solution is that the overall quality of the data set can be effectively improved by multi-dimensional intervention processing of abnormal data. Accurate replacement of abnormal data can reduce analysis deviations caused by data errors and improve the accuracy of subsequent analysis.

[0118] The data set processed by this technical solution is used to analyze the trend of psychological state changes, which can more accurately reflect the real situation of the user. Because the replacement of abnormal data is based on multidimensional information and the correlation across data sets, the replaced data is more reasonable and reliable. This technical solution can automatically identify and process abnormal data in the data set, enhancing the robustness of the system. Even if data anomalies occur in actual applications, the system can respond quickly and give reasonable processing results, ensuring the stability and reliability of the system. This technical solution not only considers the information in the current data set, but also considers the correlation between cross-data sets, and has strong adaptability and flexibility. It can cope with various complex data situations and give reasonable processing results.

[0119] On the other hand, by performing multi-dimensional intervention processing on abnormal data, including combining the characteristics of the abnormal data itself, the normal data values ​​before and after, and the information of other data sets, the abnormal data can be more accurately identified and corrected, thereby improving the quality of the overall data. When extracting normal data values, it is clearly pointed out that these data values ​​are physiological indicator parameters that meet the rationality of the appearance of vital signs, ensuring that the replaced data is physiologically reasonable, and increasing the credibility and practicality of the data. By introducing the first abnormal data adjustment parameter and the second abnormal data adjustment parameter, the relationship between the abnormal data itself and the normal data and the information of other data sets are comprehensively considered, and the accuracy of abnormal processing is improved. The abnormal data after multi-dimensional intervention processing is replaced with reasonable values, which reduces the impact of abnormal values ​​on the stability and reliability of the overall data set, so that the data set can better reflect the actual situation. The above technical solution automatically calculates the abnormal replacement data value in a formulaic way, reduces the need for manual intervention, and improves the intelligent level of data processing. At the same time, the above technical solution is not only applicable to a single data set, but also can assist in processing abnormal data by utilizing the information of other data sets. It has strong adaptability and can process data sets of different types and sources. By comprehensively considering the information of multiple dimensions to correct abnormal data, the data deviation and error that may be caused by a single factor are reduced, and the accuracy and reliability of the data are improved.

[0120] To sum up, the technical effects of the above-mentioned technical solutions in terms of performance indicators are mainly reflected in improving data quality, ensuring the physiological rationality of data, improving the accuracy of exception handling, enhancing the stability and reliability of data sets, improving the intelligence level of data processing, strong adaptability, and reducing data deviations and errors.

[0121] In one embodiment of the present invention, obtaining a second abnormal data adjustment parameter by using data at a time corresponding to the abnormal data collection time node in other data sets includes:

[0122] Step 401: extracting data at a time corresponding to the abnormal data collection time node from other data sets as first data information;

[0123] Step 402: retrieve data from the other data sets at adjacent times before and after the abnormal data collection time node as second data information;

[0124] Step 403: Obtain a second abnormal data adjustment parameter using the first data information and the second data information corresponding to the other data sets; wherein the second abnormal data adjustment parameter is obtained by the following formula:

[0125]

[0126]

[0127] Among them, f 02 represents the second abnormal data adjustment parameter; λ represents the abnormal data compensation coefficient; n represents the number of other data sets; X i represents the data value corresponding to the first data information in the i-th data set; X upi and X downi represents the upper and lower limits of the standard parameter range corresponding to the type of physiological index parameter corresponding to the i-th other data set; X ci represents the data median corresponding to the normal data range of the preset i-th data set; X 01i and X 02i They respectively represent the data values ​​of the data at the adjacent moments before and after the abnormal data collection moment node of the i-th other data set.

[0128] The working principle of the above technical solution is as follows: First, the system extracts data corresponding to the abnormal data collection time node from other data sets as the first data information. The above data reflects the physiological index parameter values ​​recorded in other data sets at the same time point.

[0129] Next, the system will retrieve data from other data sets at adjacent times before and after the abnormal data collection time node as the second data information. The above data provides the change trend of the physiological index parameter value before and after the abnormal data collection time node.

[0130] Using the first data information and the second data information, the system calculates the second abnormal data adjustment parameter through a preset formula. The above parameters comprehensively consider the data values ​​at the same time point and adjacent time points in other data sets, as well as the normal data range and standard parameter range of each data set, to evaluate the rationality of the abnormal data and the adjustment direction. λ in the above technical solution represents the abnormal data compensation coefficient, which is used to adjust the weight or influence between the abnormal data and the data of other data sets.

[0131] The effect of the above technical solution is: by comprehensively considering the data at the same time point and adjacent time points in other data sets, the system can more accurately evaluate the rationality of abnormal data and the adjustment direction, thereby improving the accuracy of the replaced data. Using the information of multiple data sets to process abnormal data makes the system more robust and reliable when facing data anomalies. Even if a data set is abnormal, the information of other data sets can provide strong support for processing. The quality of the data set processed by this technical solution has been significantly improved because the abnormal data has been reasonably replaced. The above technical solution helps to make the subsequent trend analysis of psychological state changes or other related analyses more accurate and reliable. The technical solution can be flexibly applied to the processing of physiological indicator parameter data sets in different scenarios, and has strong adaptability and flexibility. At the same time, by adjusting parameters such as the abnormal data compensation coefficient, customized data processing can be performed according to actual needs.

[0132] On the other hand, by extracting and utilizing the data information of other data sets corresponding to or adjacent to the abnormal data collection time node, the above technical solution can more accurately evaluate and adjust the abnormal data, reducing the possibility of misjudgment or over-adjustment. When obtaining the second abnormal data adjustment parameter, the normal data range of the data set, the standard parameter range of the physiological index parameter and other factors are considered, so that the data correction is more reasonable and more in line with the actual situation. Through the reasonable adjustment of the abnormal data, the above technical solution can improve the stability and reliability of the entire data set, so that the data set can better reflect the actual situation and provide a more accurate basis for subsequent data analysis and mining. The above technical solution not only considers the abnormal data point itself, but also combines the data information of other data sets at the same or similar time, thereby enhancing the robustness of anomaly detection and making the system more reliable in identifying and processing abnormal data. The above technical solution is applicable to different data sets and different types of data, and has strong adaptability. Regardless of the size, structure or data type of the data set, effective abnormal data processing can be performed through the above technical solution. The abnormal data adjustment parameter is automatically calculated by a formulaic method, which reduces the need for manual intervention and improves the efficiency of data processing.

[0133] To sum up, the technical effects of the above technical solutions in terms of performance indicators are mainly reflected in improving the accuracy of abnormal data processing, enhancing the rationality of data correction, improving the stability and reliability of data sets, enhancing the robustness of anomaly detection, strong adaptability and improving data processing efficiency.

[0134] In one embodiment of the present invention, multi-dimensional intervention data processing is performed on the missing data of multiple data sets corresponding to the physiological indicator parameters to obtain a data set after the missing data corresponding to each data set is filled, including:

[0135] S2021. Obtaining the missing data collection time node corresponding to the missing data in each data set from the data preprocessing record;

[0136] S2022, performing multi-dimensional intervention data processing on the missing data in each data set in combination with the missing data collection time node, to obtain missing data filling values ​​corresponding to the missing data;

[0137] S2023. Fill the missing data positions with the missing data filling values ​​to obtain a missing data-filled data set corresponding to each data set.

[0138] The working principle of the above technical solution is: first, the missing data collection time node corresponding to the missing data in each data set is retrieved from the data preprocessing record. The above time node is the time point when the missing data occurs and is the benchmark for subsequent data filling.

[0139] Next, combined with the missing data collection time node, multi-dimensional intervention data processing is performed on the missing data in each data set. The purpose of this step is to estimate the reasonable value of the missing data at the time of missingness, that is, the missing data filling value, based on the existing data and other relevant information. The specific method of multi-dimensional intervention data processing will be customized according to the characteristics of the data set and business needs. For example, if the data set shows obvious periodicity or trend, time series analysis or machine learning models will be used to predict missing values; if the data set has a large number of repeated patterns or periodicity, interpolation methods will be used to estimate missing values.

[0140] Finally, the missing data positions in the original data set are filled with the calculated missing data filling values. In this way, each data set will get a new version with missing data filled, which contains more complete and accurate data.

[0141] The effect of the above technical solution is: by multi-dimensional intervention data processing and filling of missing data, the integrity of the data set can be significantly improved. The existence of missing data may have a negative impact on data analysis and mining, so timely identification and filling of the above missing data is the key to ensuring data quality. The use of the filled data set for subsequent psychological state change trend analysis or other related analysis can improve the accuracy of the analysis. Because the filling of missing data is based on existing data and other relevant information, the filled data is closer to the actual situation and can better reflect the changes in the user's physiological index parameters. The technical solution can automatically identify and process missing data in the data set, thereby improving the robustness of the system. Even if data is missing in actual applications, the system can respond quickly and give reasonable filling results, ensuring the stability and reliability of the system. The filled data set provides high-quality data support for subsequent psychological state change trend analysis or other related analysis. The above technical solution helps to more accurately identify the user's physiological index parameter change trend and provide a scientific basis for related research and applications.

[0142] In one embodiment of the present invention, multi-dimensional intervention data processing is performed on the missing data in each data set in combination with the missing data collection time node to obtain the missing data filling value corresponding to the missing data, including:

[0143] Step 1: For each data set, extract the missing data collection time node of the missing data contained therein;

[0144] Step 2: extract the normal data values ​​corresponding to the two normal data collection time nodes with the closest collection interval before and after the missing data collection time node;

[0145] The first missing data parameter is obtained by using the missing data and the normal data value; wherein the first missing data parameter is obtained by the following formula:

[0146]

[0147] Among them, h 01 represents the first missing data parameter; Y c Indicates the median value of the data distribution in the data set corresponding to the missing data; Y 01 and Y 02 Respectively represent the normal data values ​​corresponding to the two normal data collection time nodes with the closest collection intervals before and after the missing data collection time node;

[0148] Step 3: extracting data at a time corresponding to the missing data collection time node from other data sets as first reference data information;

[0149] Step 4: retrieve the data of the other data sets at the time points before and after the missing data collection time point as the second reference data information;

[0150] Step 5: Obtain a second missing data parameter using the first missing data information and the second missing data information corresponding to the other data sets; wherein the second missing data parameter is obtained by the following formula:

[0151]

[0152]

[0153] Among them, h 02 represents the second missing data parameter; n represents the number of other data sets; ξ represents the missing data compensation coefficient; Y ci represents the median value of the data distribution corresponding to the i-th other data set; Y zi The data median of the standard parameter range of the type of physiological indicator parameter corresponding to the i-th other data set; Y 01i and Y 02i They respectively represent the normal data values ​​corresponding to the two normal data collection time nodes with the closest collection interval before and after the missing data collection time node corresponding to the i-th other data set; X upi and X downi Indicates the upper limit and lower limit of the standard parameter range corresponding to the type of the physiological indicator parameter corresponding to the i-th other data set;

[0154] Step 6: Obtain a missing data filling value using the first missing data parameter and the second missing data parameter; wherein the missing data filling value is obtained by the following formula:

[0155]

[0156] Among them, Y t represents the missing data filling value; h 01 represents the first missing data parameter; h 02 Indicates the second missing data parameter; Y 01 and Y 02 They respectively represent the normal data values ​​corresponding to the two normal data collection time nodes with the closest collection intervals before and after the missing data collection time node.

[0157] The working principle of the above technical solution is as follows: First, the system extracts the missing data collection time nodes of the missing data contained in each data set. The above time nodes are the basis for subsequent data processing, because it is necessary to find data values ​​for filling based on the above time nodes.

[0158] Next, the system will find the two normal data collection time nodes with the closest collection interval before and after the missing data collection time node, and extract the normal data values ​​corresponding to these two time nodes. Then, using the missing data and these two normal data values, the first missing data parameter is calculated through a preset formula. The above parameter describes the relationship between the missing data and the surrounding normal data values, which is used for subsequent missing data filling.

[0159] The system not only considers the information of the current data set, but also extracts the data of the time corresponding to the missing data collection time node in other data sets (first reference data information) and the data of the previous and next adjacent time nodes (second reference data information). The above reference data provides an additional source of information for filling in missing data.

[0160] Using reference data extracted from other datasets, the system calculates a second missing data parameter that takes into account information across datasets and more fully reflects the possible values ​​of missing data.

[0161] Finally, the system will combine the first missing data parameter and the second missing data parameter, as well as the normal data values ​​corresponding to the two normal data collection time nodes with the closest collection interval before and after the missing data collection time node, and calculate the missing data filling value through a preset formula. The above filling value is calculated based on multi-dimensional information, so it has high accuracy and reliability.

[0162] The effect of the above technical solution is: by multi-dimensional intervention data processing and filling of missing data, the technical solution can significantly improve the integrity of the data set. The complete data set provides a richer information basis for subsequent data analysis and mining. Since the filling value is calculated based on multi-dimensional information and correlation across data sets, the filled data has high accuracy and reliability. The above technical solution helps to improve the accuracy of subsequent analysis. Using the filled data set for psychological state change trend analysis or other related analysis can obtain more accurate analysis results. This is of great significance for applications in the fields of medical care, health monitoring, etc. The technical solution can automatically identify and process missing data in the data set, thereby improving the robustness of the system. Even if data is missing in actual applications, the system can respond quickly and give reasonable filling results, ensuring the stability and reliability of the system.

[0163] On the other hand, through multi-dimensional data reference (including normal data values ​​at time points before and after the missing data, data values ​​of other data sets at the same time point or adjacent time points), and the missing data parameters calculated using these data values, the missing data values ​​can be estimated and filled more accurately, thereby improving the completeness and accuracy of the data. Filling in missing data makes the data set more complete, thereby improving the availability of the data. A complete data set is more likely to be used for various data analysis, mining and machine learning tasks, providing strong support for decision making, trend prediction, etc. The above technical solution automatically calculates the filling value of missing data through a formulaic method, reduces the need for manual intervention, and improves the intelligence and automation level of data processing. The technical solution takes into account factors such as the median of data distribution, the standard parameter range of physiological indicator parameters, and the correlation between data sets, making it highly adaptable when processing data sets of different types and distributions. By comprehensively considering information from multiple data sources and multiple time points, the above technical solution can reduce the deviations and errors that may be caused by a single data source or a single time point data, and improve the accuracy and reliability of data filling.

[0164] To sum up, the technical effects of this technical solution in terms of performance indicators are mainly reflected in improving data integrity and accuracy, enhancing data availability, improving the intelligence and automation level of data processing, strong adaptability, and reducing data deviations and errors.

[0165] In one embodiment of the present invention, a deep learning model with a psychological state change trend analysis function is used to analyze the multiple data sets to obtain a psychological state change trend analysis report, including:

[0166] S301, performing standardization processing on the multiple data sets to obtain each data set after the standardization processing;

[0167] S302, inputting each of the standardized data sets into a deep learning model with a psychological state change trend analysis function, and outputting an analysis result through the deep learning model, wherein the analysis result is a psychological state change trend analysis report;

[0168] S303: Visually display the psychological state change trend analysis report.

[0169] The working principle of the above technical solution is as follows: First, multiple data sets are standardized. Data standardization is an important step before data analysis. Its purpose is to eliminate the dimensional differences between different features and make the data more suitable for subsequent model analysis. Standardization processing usually includes removing the mean and scaling of the data so that the mean of the data set is 0 and the standard deviation is 1.

[0170] The standardized data set is input into the pre-trained deep learning model. The above deep learning model has the function of analyzing the trend of changes in psychological state. It can identify the patterns and trends in the data through learning and analyzing the data set, and output the analysis results. In the above scenario, the analysis result is the psychological state change trend analysis report. The analysis process of the deep learning model usually includes the steps of forward propagation, back propagation and weight update. Forward propagation means that the model calculates the output result based on the input data and the current weight; back propagation is to calculate the gradient information based on the difference between the output result and the true label; weight update is to update the weight of the model based on the gradient information, so that the model can predict the result more accurately in the next iteration.

[0171] Finally, the psychological state change trend analysis report output by the deep learning model is visualized. Visualization helps users understand the analysis results more intuitively and discover the patterns and trends in the data. Visualization can include charts, images, animations and other forms.

[0172] Among them, the process of building a deep learning model with the function of analyzing the trend of psychological state changes is as follows:

[0173] Step 1: Construct feature extraction layer:

[0174] For heart rate information and respiratory rate data, one-dimensional convolutional neural network (1D CNN) or recurrent neural network (RNN / LSTM / GRU) is used to extract time series features. The network structure can capture the time dependency and periodic patterns in the data.

[0175] For galvanic skin response data (such as skin conductance response, GSR), since it usually has a short response time, convolutional neural network (CNN) or simple fully connected layers are used to extract features.

[0176] Step 2: Construct the fusion layer:

[0177] The features extracted from different physiological indicators are fused to obtain fused feature information. In this embodiment, this is achieved through concatenation, addition, or a more complex attention mechanism. The role of the fusion layer is to combine information from different sources so that subsequent layers can learn the associations and interactions between them.

[0178] Step 3: Build the middle layer:

[0179] Use multiple fully connected layers (or dense layers, hidden layers) to process the above fused feature information. The above multiple fully connected layers can learn complex nonlinear relationships and extract higher-level feature representations.

[0180] In the above-mentioned multiple fully connected layers, activation functions (such as ReLU, Sigmoid or Tanh) are introduced to increase the nonlinearity of the model. At the same time, in order to prevent overfitting, regularization techniques such as dropout are used.

[0181] Step 4. Build the output layer:

[0182] Depending on the requirements of the specific task, the output layer can be designed as a regression problem or a classification problem.

[0183] For the regression problem (predicting the degree of change in the psychological state), the output layer has only one neuron and uses a linear activation function. The output of the model will be a continuous value that represents the trend of change in the psychological state.

[0184] For classification problems (classifying mental states into different categories), the output layer has multiple neurons and uses a softmax activation function to generate a probability distribution. The output of the model will be the probabilities of different mental state categories.

[0185] Step 5: Training and Optimization:

[0186] The model is trained using a labeled dataset. The labels can be scores or categories of psychological states obtained through professional assessments or questionnaires.

[0187] Choose an appropriate loss function (such as mean squared error MSE for regression problems and cross entropy loss for classification problems) and optimizer (such as gradient descent, Adam, etc.) to train the model.

[0188] Optimize the performance of the model by adjusting the model's hyperparameters (such as learning rate, batch size, number of iterations, etc.).

[0189] Step 6: Real-time monitoring and early warning:

[0190] After the model training is completed, it can be deployed in a real-time monitoring system to analyze the individual's physiological indicator data in real time and predict the changing trend of the psychological state.

[0191] When the model detects that an individual's mental state is abnormal or high-risk, an early warning mechanism can be triggered so that timely measures can be taken to intervene.

[0192] The effect of the above technical solution is: using a deep learning model to analyze the trend of changes in mental state can automatically learn and identify complex patterns and trends in the data, thereby improving the accuracy of the analysis. Compared with traditional manual analysis methods, deep learning models can process larger data sets and discover more subtle and hidden laws. Deep learning models can quickly process and analyze large data sets and output analysis results in a short time. This greatly improves the efficiency of data analysis and enables users to obtain useful information faster. By presenting the trend analysis report of mental state changes in a visual way, users can more intuitively understand the analysis results and discover the laws and trends in the data. The above technical solution helps users better understand the changes in mental state and make more informed decisions. The deep learning model can be customized according to the specific needs of the user to meet the needs of mental state change trend analysis in different scenarios. This makes the technical solution more flexible and scalable.

[0193] The embodiment of the present invention proposes a psychological state change trend analysis system based on multi-dimensional physiological information, such as Figure 2 As shown, the psychological state change trend analysis system based on multi-dimensional physiological information includes:

[0194] The data collection and processing module is used to collect the user's physiological indicator parameters, classify the user's physiological indicator parameters, and obtain multiple data sets corresponding to the physiological indicator parameters; wherein the user's physiological indicator parameters are physiological indicator parameters monitored and obtained by a body physiological indicator monitoring instrument under long-term monitoring, and the physiological indicator parameters are stored in a database.

[0195] A multidimensional intervention data processing module, used for performing multidimensional intervention data processing on the multiple data sets corresponding to the physiological index parameters, and obtaining multiple data sets after the multidimensional intervention data processing;

[0196] The data analysis and processing module is used to input the multiple data sets processed by the multi-dimensional intervention data into a deep learning model with a psychological state change trend analysis function, and to visualize and generate a report on the psychological state change trend analysis results output by the deep learning model.

[0197] The working principle of the above technical solution is: through the long-term monitoring of the body physiological index monitoring instrument, the user's physiological index parameters, such as heart rate, blood pressure, galvanic skin response (GSR), breathing rate, etc., are continuously collected. The collected physiological index parameters are classified to form multiple data sets according to different physiological indicators or time periods. The above data sets are stored in the database for subsequent analysis and processing.

[0198] Multidimensional intervention data processing is performed on each data set, including data cleaning (removing noise and outliers), data standardization (enabling data of different dimensions to be compared), and data enhancement (increasing the amount of data through simulation or other means). The goal of multidimensional intervention data processing is to ensure the data quality of subsequent analysis and improve the accuracy and reliability of the analysis.

[0199] Use a pre-trained deep learning model with the function of analyzing the trend of psychological state changes to analyze multiple processed data sets. The deep learning model can automatically learn complex patterns and features in the data to predict or identify the trend of the user's psychological state changes. Through the analysis of the model, a psychological state change trend analysis report is generated, which includes the fluctuation of the user's psychological state, psychological problems or suggestions, etc.

[0200] The effect of the above technical solution is: by analyzing multi-dimensional physiological information through a deep learning model, it is possible to more accurately identify and predict the changing trend of the user's psychological state, which has higher accuracy than the traditional questionnaire-based or interview-based method. Since the above technical solution is based on physiological indicator data under long-term monitoring, it can reflect the changes in the user's psychological state in real time, providing a basis for timely discovery and intervention of psychological problems. By classifying and processing the physiological indicator data of different users, personalized psychological state analysis can be achieved, providing more accurate mental health services for each user. By continuously monitoring and analyzing the changing trends of the user's psychological state, potential psychological problems can be discovered in a timely manner, providing a scientific basis for psychological intervention and prevention, and helping to reduce the incidence of mental illness.

[0201] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if the above-mentioned modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include the above-mentioned modifications and variations.

Claims

1. A method for analyzing the trend of psychological state changes based on multidimensional physiological information, characterized in that: The method for analyzing the trend of psychological state changes based on multidimensional physiological information includes: Collecting physiological index parameters of the user, and classifying the physiological index parameters of the user to obtain multiple data sets corresponding to the physiological index parameters; Performing multi-dimensional intervention data processing on the multiple data sets corresponding to the physiological indicator parameters to obtain multiple data sets after the multi-dimensional intervention data processing; Performing data analysis on the multiple data sets using a deep learning model with a psychological state change trend analysis function to obtain a psychological state change trend analysis report; Performing multi-dimensional intervention data processing on the multiple data sets corresponding to the physiological indicator parameters to obtain multiple data sets after the multi-dimensional intervention data processing, including: Performing multi-dimensional intervention data processing on the abnormal data of the multiple data sets corresponding to the physiological index parameters to obtain a data set after the abnormal data corresponding to each data set is replaced; Performing multi-dimensional intervention data processing on the missing data of multiple data sets corresponding to the physiological index parameters to obtain a data set after the missing data corresponding to each data set is filled; Performing multi-dimensional intervention data processing on the abnormal data of the multiple data sets corresponding to the physiological index parameters to obtain a data set after the abnormal data corresponding to each data set is replaced, including: Obtaining data values ​​of abnormal data in each data set and abnormal data collection time nodes corresponding to the abnormal data from the data preprocessing records; Performing multi-dimensional intervention data processing on the data values ​​of the abnormal data in each data set in combination with the abnormal data collection time node to obtain abnormal replacement data values ​​corresponding to the abnormal data; The abnormal data is replaced by using the abnormal replacement data value to obtain a data set after the abnormal data is replaced corresponding to each data set; In combination with the abnormal data collection time node, multi-dimensional intervention data processing is performed on the data value of the abnormal data in each data set to obtain the abnormal replacement data value corresponding to the abnormal data, including: Extracting the abnormal data collection time node of the abnormal data contained in each data set; Extract normal data values ​​corresponding to two normal data collection time nodes with the closest collection intervals before and after the abnormal data collection time node; The first abnormal data adjustment parameter is obtained by using the abnormal data and the normal data value; wherein the first abnormal data adjustment parameter is obtained by the following formula: Wherein, f01 represents the first abnormal data adjustment parameter; Xe represents the preset data deviation reference value; Xy represents the data value of the abnormal data; X01 and X02 represent the normal data values ​​corresponding to two normal data collection time nodes respectively; Obtain a second abnormal data adjustment parameter using data at a time corresponding to the abnormal data collection time node in other data sets; The abnormal replacement data value corresponding to the abnormal data is obtained by using the first abnormal data adjustment parameter and the second abnormal data adjustment parameter in combination with the data value of the abnormal data; wherein the abnormal replacement data value corresponding to the abnormal data is obtained by the following formula: Wherein, Xt represents the abnormal replacement data value corresponding to the abnormal data; f01 represents the first abnormal data adjustment parameter; f02 represents the second abnormal data adjustment parameter; m represents the number of normal data in the data set corresponding to the abnormal data; Xj represents the data value corresponding to the jth normal data; Using data at a time corresponding to the abnormal data collection time node in other data sets to obtain a second abnormal data adjustment parameter includes: Extracting data at a time corresponding to the abnormal data collection time node from other data sets as first data information; Retrieving data from the other data sets at adjacent times before and after the abnormal data collection time node as second data information; The second abnormal data adjustment parameter is obtained by using the first data information and the second data information corresponding to the other data sets; wherein the second abnormal data adjustment parameter is obtained by the following formula: Among them, f 02 represents the second abnormal data adjustment parameter; λ represents the abnormal data compensation coefficient; n represents the number of other data sets; X i represents the data value corresponding to the first data information in the i-th data set; X upi and X downi represents the upper and lower limits of the standard parameter range corresponding to the type of physiological index parameter corresponding to the i-th other data set; X ci represents the data median corresponding to the normal data range of the preset i-th data set; X 01i and X 02i They respectively represent the data values ​​of the data at the adjacent moments before and after the abnormal data collection moment node of the i-th other data set.

2. The method for analyzing the trend of psychological state changes based on multidimensional physiological information according to claim 1, characterized in that: Collect the user's physiological index parameters, and classify the user's physiological index parameters to obtain multiple data sets corresponding to the physiological index parameters, including: Retrieving physiological index parameters with the same collection period as that of the user from a database; Classifying the physiological indicator parameters according to their types to obtain a plurality of data sets corresponding to the physiological indicator parameters; wherein the types of the physiological indicator parameters include heart rate information, respiratory rate, and skin electrical response data; and the data sets correspond to the types of the physiological indicator parameters one by one; The multiple data sets are subjected to data preprocessing and preprocessing records to obtain multiple data sets after preprocessing; wherein the data preprocessing includes data cleaning processing, data noise reduction processing, and outlier and missing value processing.

3. The method for analyzing the trend of psychological state changes based on multidimensional physiological information according to claim 1, characterized in that: Performing multi-dimensional intervention data processing on the missing data of multiple data sets corresponding to the physiological index parameters to obtain a data set after the missing data corresponding to each data set is filled, including: Obtain the missing data collection time node corresponding to the missing data in each data set from the data preprocessing record; Perform multi-dimensional intervention data processing on the missing data in each data set in combination with the missing data collection time node to obtain missing data filling values ​​corresponding to the missing data; The missing data positions are filled with the missing data filling values ​​to obtain a missing data-filled data set corresponding to each data set.

4. The method for analyzing the trend of psychological state changes based on multidimensional physiological information according to claim 3, characterized in that: Performing multi-dimensional intervention data processing on the missing data in each data set in combination with the missing data collection time node to obtain missing data filling values ​​corresponding to the missing data, including: Extract the missing data collection time node of the missing data contained in each data set; Extract normal data values ​​corresponding to two normal data collection time nodes with the closest collection intervals before and after the missing data collection time node; The first missing data parameter is obtained by using the missing data and the normal data value; wherein the first missing data parameter is obtained by the following formula: Among them, h 01 represents the first missing data parameter; Y c Indicates the median value of the data distribution in the data set corresponding to the missing data; Y 01 and Y 02 Respectively represent the normal data values ​​corresponding to the two normal data collection time nodes with the closest collection intervals before and after the missing data collection time node; Extracting data at a time corresponding to the missing data collection time node from other data sets as first reference data information; Retrieving data from the other data sets at adjacent times before and after the missing data collection time node as second reference data information; The second missing data parameter is obtained by using the first missing data information and the second missing data information corresponding to the other data set; wherein the second missing data parameter is obtained by the following formula: Among them, h 02 represents the second missing data parameter; n represents the number of other data sets; ξ represents the missing data compensation coefficient; Y ci represents the median value of the data distribution corresponding to the i-th other data set; Y zi The data median of the standard parameter range of the type of physiological indicator parameter corresponding to the i-th other data set; Y 01i and Y 02i They respectively represent the normal data values ​​corresponding to the two normal data collection time nodes with the closest collection interval before and after the missing data collection time node corresponding to the i-th other data set; X upi and X downi Indicates the upper limit and lower limit of the standard parameter range corresponding to the type of the physiological indicator parameter corresponding to the i-th other data set; The missing data filling value is obtained by using the first missing data parameter and the second missing data parameter; wherein the missing data filling value is obtained by the following formula: Among them, Y t represents the missing data filling value; h 01 represents the first missing data parameter; h 02 Indicates the second missing data parameter; Y 01 and Y 02 They respectively represent the normal data values ​​corresponding to the two normal data collection time nodes with the closest collection intervals before and after the missing data collection time node.

5. The method for analyzing the trend of psychological state changes based on multi-dimensional physiological information according to claim 1, characterized in that: The multiple data sets are analyzed using a deep learning model with a psychological state change trend analysis function to obtain a psychological state change trend analysis report, including: Performing standardization processing on the multiple data sets to obtain each standardized data set; Inputting each of the standardized data sets into a deep learning model with a psychological state change trend analysis function, and outputting the analysis result through the deep learning model, wherein the analysis result is a psychological state change trend analysis report; The psychological state change trend analysis report is visualized.

6. A psychological state change trend analysis system based on multi-dimensional physiological information, characterized in that: The psychological state change trend analysis system based on multi-dimensional physiological information includes: A data collection and processing module is used to collect the user's physiological index parameters, and classify the user's physiological index parameters to obtain multiple data sets corresponding to the physiological index parameters; A multidimensional intervention data processing module, used to perform multidimensional intervention data processing on the multiple data sets corresponding to the physiological index parameters, and obtain multiple data sets after the multidimensional intervention data processing; A data analysis and processing module, used to input the multiple data sets processed by the multi-dimensional intervention data into a deep learning model with a psychological state change trend analysis function, and to visualize and report the psychological state change trend analysis results output by the deep learning model; Wherein, performing multi-dimensional intervention data processing on the multiple data sets corresponding to the physiological indicator parameters to obtain multiple data sets after the multi-dimensional intervention data processing includes: Performing multi-dimensional intervention data processing on the abnormal data of the multiple data sets corresponding to the physiological index parameters to obtain a data set after the abnormal data corresponding to each data set is replaced; Performing multi-dimensional intervention data processing on the missing data of multiple data sets corresponding to the physiological index parameters to obtain a data set after the missing data corresponding to each data set is filled; Performing multi-dimensional intervention data processing on the abnormal data of the multiple data sets corresponding to the physiological index parameters to obtain a data set after the abnormal data corresponding to each data set is replaced, including: Obtaining data values ​​of abnormal data in each data set and abnormal data collection time nodes corresponding to the abnormal data from the data preprocessing records; Performing multi-dimensional intervention data processing on the data values ​​of the abnormal data in each data set in combination with the abnormal data collection time node to obtain abnormal replacement data values ​​corresponding to the abnormal data; The abnormal data is replaced by using the abnormal replacement data value to obtain a data set after the abnormal data is replaced corresponding to each data set; In combination with the abnormal data collection time node, multi-dimensional intervention data processing is performed on the data value of the abnormal data in each data set to obtain the abnormal replacement data value corresponding to the abnormal data, including: Extracting the abnormal data collection time node of the abnormal data contained in each data set; Extract normal data values ​​corresponding to two normal data collection time nodes with the closest collection intervals before and after the abnormal data collection time node; The first abnormal data adjustment parameter is obtained by using the abnormal data and the normal data value; wherein the first abnormal data adjustment parameter is obtained by the following formula: Wherein, f01 represents the first abnormal data adjustment parameter; Xe represents the preset data deviation reference value; Xy represents the data value of the abnormal data; X01 and X02 represent the normal data values ​​corresponding to two normal data collection time nodes respectively; Obtain a second abnormal data adjustment parameter using data at a time corresponding to the abnormal data collection time node in other data sets; The abnormal replacement data value corresponding to the abnormal data is obtained by using the first abnormal data adjustment parameter and the second abnormal data adjustment parameter in combination with the data value of the abnormal data; wherein the abnormal replacement data value corresponding to the abnormal data is obtained by the following formula: Wherein, Xt represents the abnormal replacement data value corresponding to the abnormal data; f01 represents the first abnormal data adjustment parameter; f02 represents the second abnormal data adjustment parameter; m represents the number of normal data in the data set corresponding to the abnormal data; Xj represents the data value corresponding to the jth normal data; Using data at a time corresponding to the abnormal data collection time node in other data sets to obtain a second abnormal data adjustment parameter includes: Extracting data at a time corresponding to the abnormal data collection time node from other data sets as first data information; Retrieving data from the other data sets at adjacent times before and after the abnormal data collection time node as second data information; The second abnormal data adjustment parameter is obtained by using the first data information and the second data information corresponding to the other data sets; wherein the second abnormal data adjustment parameter is obtained by the following formula: Among them, f 02 represents the second abnormal data adjustment parameter; λ represents the abnormal data compensation coefficient; n represents the number of other data sets; X i represents the data value corresponding to the first data information in the i-th data set; X upi and X downi represents the upper and lower limits of the standard parameter range corresponding to the type of physiological index parameter corresponding to the i-th other data set; X ci represents the data median corresponding to the normal data range of the preset i-th data set; X 01i and X 02i They respectively represent the data values ​​of the data at the adjacent moments before and after the abnormal data collection moment node of the i-th other data set.

Images