A data method for depression state monitoring

Abstract

The application discloses a data method for depression state monitoring, which comprises the following steps: collecting personal basic information of a depression tester, matching corresponding test content data based on the personal basic information; obtaining response content data of the depression tester, extracting first response emotion feature data, corresponding emotion types and emotion scores based on a preset deep learning model, obtaining second response emotion feature data and corresponding emotion types and emotion scores based on a preset emotion driving model; establishing an emotion matrix of the depression tester according to the emotion types and emotion scores, calling a depression emotion matrix in a depression database, calculating first depression state evaluation values and second depression state evaluation values corresponding to the first emotion matrix, the second emotion matrix and the depression emotion matrix, comparing the first depression state evaluation values and the second depression state evaluation values with a preset depression emotion threshold value, obtaining a depression comparison result, and judging a mental health state of the depression tester according to the depression comparison result.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mental health detection, and in particular to a data method for depression state monitoring. BACKGROUND

[0002] Health is not only the absence of disease and physical weakness, but also the perfect state of physical, mental and social adaptation. The status of mental health is no less than that of physical health. A good and perfect mental health intervention system is an indispensable part of the national public health system and preventive medicine.

[0003] At present, a depression state identification method and device based on eye movement sequence space-time feature analysis are disclosed in Chinese patent CN115607159B, which comprises obtaining eye movement data of a to-be-identified tester when watching a combination of positive and negative emotional pictures; inputting the eye movement data of the to-be-identified tester into a depression state identification network based on eye movement sequence space-time features; and obtaining a depression state identification result of the to-be-identified tester according to the eye movement data of the to-be-identified tester and the depression state identification network based on eye movement sequence space-time features. However, in the related art, the preliminary judged emotion and the emotion after subsequent driving are not comprehensively represented in the emotion judgment of the depression state, the judgment of the depression state lacks dynamicity and continuity, the emotion judgment is relatively single and general, and the subsequent induced verification is not performed according to the preliminary emotional performance of the tester, which may cause imperfect detection of accidental emotions and abnormal emotions in the emotion judgment process. SUMMARY

[0004] The technical problem solved by the present application is that the preliminary judged emotion and the emotion after subsequent driving are not comprehensively represented in the emotion judgment of the depression state, the judgment of the depression state lacks dynamicity and continuity, the emotion judgment is relatively single and general, and the subsequent induced verification is not performed according to the preliminary emotional performance of the tester, which may cause imperfect detection of accidental emotions and abnormal emotions in the emotion judgment process.

[0005] To solve the above technical problems, the present application provides the following technical solutions: a data method for depression state monitoring, specifically comprising the following steps:

[0006] Step S100, collecting personal basic information of a depression tester, and matching corresponding test content data based on the personal basic information;

[0007] Step S200, the depression tester responds to the test content data to obtain corresponding response content data, extracts first response emotion feature data of the response content data based on a preset deep learning model, and obtains emotion categories and emotion scores corresponding to the first response emotion feature data, inputs the first response emotion feature data into a preset emotion driving model to obtain second response emotion feature data and corresponding emotion categories and emotion scores;

[0008] Step S300, an emotion matrix of the depression tester is established according to the emotion categories and emotion scores, the emotion matrix includes a first emotion matrix and a second emotion matrix, rows of the emotion matrix are emotion categories, and columns of the emotion matrix are emotion scores corresponding to response emotion feature data;

[0009] Step S400, a depression emotion matrix in a depression database is called, first depression state evaluation values and second depression state evaluation values corresponding to the first emotion matrix, the second emotion matrix and the depression emotion matrix are calculated, the first depression state evaluation values and the second depression state evaluation values are compared with a preset depression emotion threshold value to obtain a depression comparison result, and a mental health state of the depression tester is judged according to the depression comparison result.

[0010] As a preferred scheme of the data method for depression state monitoring, the test content data includes test text data, test image data, test video data and test question data.

[0011] The response content data includes response text data, response image data, response video data and response question data.

[0012] The response emotion feature data includes voice emotion features, eye movement emotion features, facial emotion features and body emotion features.

[0013] As a preferred scheme of the data method for depression state monitoring, the step S100 specifically includes:

[0014] Step S101, personal basic information of the depression tester is collected, the personal basic information includes age, gender and region of the depression tester, the age and the gender are input into a test content database, and the test content database outputs first test content data corresponding to the depression tester;

[0015] Step S102, the region is input into the test content database, the test content database obtains cultural information data of the region, and second test content data corresponding to the depression tester is output according to the cultural information data;

[0016] Step S103, the first test content data and the second test content data are fused and spliced, and the corresponding test content data is matched.

[0017] As a preferred scheme of the data method for depression state monitoring, the step S200 specifically comprises:

[0018] Step S201, obtaining the test text data, test image data, test video data and test question data corresponding to the depression tester;

[0019] Step S202, the depression tester responds to the test text data, test image data, test video data and test question data to obtain the corresponding response text data, response image data, response video data and response question data;

[0020] Step S203, inputting the response text data, response image data, response video data and response question data into the preset deep learning model to extract the first response emotion feature data, and obtaining the emotion type and emotion score corresponding to the first response emotion feature data;

[0021] Step S204, inputting the first response emotion feature data into the preset emotion driving model to obtain the second response emotion feature data and the corresponding emotion type and emotion score.

[0022] As a preferred scheme of the data method for depression state monitoring, the step S300 specifically comprises:

[0023] Step S301, taking the emotion type corresponding to the first response emotion feature data output by the deep learning model as the row of the first emotion matrix, and taking the emotion score corresponding to the first response emotion feature data as the column of the first emotion matrix;

[0024] Step S302, taking the emotion type corresponding to the second response emotion feature data output by the deep learning model as the row of the second emotion matrix, and taking the emotion score corresponding to the second response emotion feature data as the column of the second emotion matrix;

[0025] Step S303, filling the first emotion matrix and the second emotion matrix, and performing a 0 operation if the corresponding emotion score and emotion type do not exist.

[0026] As a preferred scheme of the data method for depression state monitoring, the step S400 specifically comprises:

[0027] Step S401, inputting the personal basic information of the depression tester into the depression database to obtain a corresponding depression emotion matrix, and the expression of the depression emotion matrix is i is 1-4;

[0028] Step S402, calculating the first emotion matrix the second emotion matrix and the first depression state evaluation value and the second depression state evaluation value corresponding to the depression emotion matrix, and the mathematical expression of the first depression state evaluation value is as follows:

[0029]

[0030] Wherein, P1 represents the first depression state evaluation value, β j represents the weight value of the emotion category of the first emotion matrix, d 1i represents the emotion score of the emotion category of the first emotion matrix, v 1i represents the emotion score of the emotion category of the depression emotion matrix;

[0031] The mathematical expression of the second depression state evaluation value is as follows:

[0032]

[0033] Wherein, P2 represents the second depression state evaluation value, γ j represents the weight value of the emotion category of the second emotion matrix, and r1i represents the emotion score of the emotion category of the second emotion matrix;

[0034] Step S403, the depression emotion threshold includes the first depression emotion threshold and the second depression emotion threshold, the first depression state evaluation value and the first depression emotion threshold T1 obtain the first depression comparison result, the second depression state evaluation value and the second depression emotion threshold T2 obtain the second depression comparison result; according to the first depression comparison result and the second depression comparison result, the mental health state of the depression tester is judged.

[0035] As a preferred scheme of the data method for monitoring the depression state, wherein: the emotion categories include fear, sadness, disgust, anger, happiness, calmness and surprise; the range of the emotion score is 0-1.

[0036] As a preferred scheme of the data method for monitoring the depression state, wherein: the specific method for obtaining the voice emotion features, eye movement emotion features, facial emotion features and body emotion features includes:

[0037] Real-time collection of response question data and response video data generated by the depression tester answering multiple test question data, and putting the collected response question data and response video data into the pre-trained deep learning model to extract voice emotion features and facial emotion features;

[0038] Real-time collection of response video data generated by the depression tester watching multiple test video data, and putting the collected response video data into the pre-trained deep learning model to extract facial emotion features and body emotion features;

[0039] Real-time collection of response image data generated by the depression tester watching multiple test image data, and putting the collected response image data into the pre-trained deep learning model to extract eye movement emotion features;

[0040] Real-time collection of response text data generated by the depression tester reading multiple test text data, and putting the collected response text data into the pre-trained deep learning model to extract voice emotion features.

[0041] As a preferred scheme of the data method for monitoring the depression state, the weight value of the emotion category is obtained, and the specific method includes: calculating the sub-attention allocation result of the response text data, response image data, response video data and response question data to depression emotion based on an attention mechanism.

[0042] The emotion weights of fear, sadness, disgust, anger, happiness, calmness and surprise to depression emotion are calculated based on the sub-attention allocation result.

[0043] As a preferred scheme of the data method for monitoring the depression state, the specific method for judging the mental health state of the depression tester according to the first depression comparison result and the second depression comparison result includes:

[0044] When P1 is greater than T1 and P2 is greater than T2, the depression is severe, and immediate psychological intervention is needed;

[0045] When P1 is greater than T1 and P2 is less than T2, the depression is moderate, and communication with a psychology teacher is needed;

[0046] When P1 is less than T1 and P2 is greater than T2, the depression is mild, and the depression tester needs to adjust his mood;

[0047] When P1 is less than T1 and P2 is less than T2, the mental health is good, and no intervention is needed.

[0048] The beneficial effects of the present application are: by matching the corresponding test content data to the personal basic information (including age, gender and region) of the depression tester, the depression tester can be provided with targeted and personalized test content data, avoiding the single test content data and the wrong judgment of depression caused by individual differences, and the accuracy and precision of the depression state judgment can be improved. Based on the test text data, test image data, test video data and test question data, the emotion type and corresponding emotion score of the depression tester are preliminarily judged, based on the first response emotion feature data of the depression tester to the test content data, the emotion driving model matches the corresponding driving emotion question according to the response emotion feature data, obtains the second response emotion feature data of the depression tester to the driving emotion question, and compares the first response emotion feature data with the second response emotion feature data with the preset first depression emotion threshold and the first depression emotion threshold. The development trend of the depression tester's emotion can be mobilized, so as to avoid one-sided judgment of emotion when diagnosing depression, improve the anti-interference of emotion judgment, so as to obtain the data that can best reflect the real depression state of the depression tester, and realize the multi-modal double judgment of depression emotion, thereby improving the judgment of the psychological health state of the depression tester. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 A basic flow diagram of a data method for depression state monitoring is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0050] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings. Obviously, the described embodiments are part of the embodiments of the present application, not all embodiments.

[0051] Embodiment, refer to Figure 1 For an embodiment of the present application, a data method for depression state monitoring is provided, which specifically includes the following steps:

[0052] Step S100, collecting the personal basic information of the depression tester, and matching the corresponding test content data based on the personal basic information;

[0053] Step S200, the depression tester responds to the test content data to obtain corresponding response content data, extracts the first response emotion feature data of the response content data based on the preset deep learning model, and obtains the emotion type and emotion score corresponding to the first response emotion feature data. The first response emotion feature data is input into the preset emotion driving model to obtain the second response emotion feature data and the corresponding emotion type and emotion score;

[0054] Step S300, according to the emotion category and the emotion score, an emotion matrix of the depression tester is established, the emotion matrix includes a first emotion matrix and a second emotion matrix, the rows of the emotion matrix are emotion categories, and the columns of the emotion matrix are emotion scores corresponding to the response emotion feature data;

[0055] Step S400, the depression emotion matrix in the depression database is called, the first depression state evaluation value and the second depression state evaluation value corresponding to the first emotion matrix, the second emotion matrix and the depression emotion matrix are calculated, the first depression state evaluation value and the second depression state evaluation value are compared with the preset depression emotion threshold value, a depression comparison result is obtained, and the mental health state of the depression tester is judged according to the depression comparison result.

[0056] The test content data includes test text data, test image data, test video data and test question data;

[0057] The response content data includes response text data, response image data, response video data and response question data;

[0058] The response emotion feature data includes voice emotion features, eye movement emotion features, facial emotion features and body emotion features.

[0059] In the embodiment, the test content data covers a wide range of aspects and a diversity of forms, and can realize multi-modal emotion judgment of the depression tester, so that the judgment of the mental health state of the depression tester is more comprehensive and persuasive.

[0060] Step S100 specifically includes:

[0061] Step S101, personal basic information of the depression tester is collected, the personal basic information includes age, gender and region of the depression tester, the age and gender are input into a test content database, and the test content database outputs first test content data corresponding to the depression tester;

[0062] Step S102, the region is input into the test content database, the test content database acquires cultural information data of the region, and according to the cultural information data, second test content data corresponding to the depression tester is output;

[0063] Step S103, the first test content data and the second test content data are fused and spliced to match the corresponding test content data.

[0064] In this embodiment, by matching the corresponding test content data with the basic personal information (including age, gender and region) of the depression test subject, targeted and personalized test content data can be provided to the depression test subject, avoiding the erroneous judgment of depressive emotions caused by the single test content data and individual differences, and improving the accuracy and precision of the judgment of the depressive state.

[0065] Step S200 specifically includes:

[0066] Step S201, obtaining test text data, test image data, test video data and test question data corresponding to the depression test subject;

[0067] Step S202: The depression test subject responds to the test text data, test image data, test video data, and test question data to obtain corresponding response text data, response image data, response video data, and response question data;

[0068] Step S203: Input the response text data, response image data, response video data, and response question data into a preset deep learning model to extract the first response emotion feature data, and obtain the emotion type and emotion score corresponding to the first response emotion feature data;

[0069] Step S204: input the first response emotion feature data into a preset emotion-driven model to obtain second response emotion feature data and corresponding emotion types and emotion scores.

[0070] In this embodiment, the deep learning model can analyze the response text data, response image data, response video data and response question data of the depression test subject to obtain the corresponding emotion type and emotion score. The emotion score represents the intensity of the emotion type. For example, fear is divided into mild fear, moderate fear and strong fear. The emotion-driven model matches the corresponding driving emotion question according to the response emotion feature data. The depression test subject responds to the corresponding driving emotion question to obtain the second response emotion feature data of the depression test subject to the driving emotion question. The driving question is committed to reversing the emotion type and emotion score corresponding to the first response emotion feature data, thereby achieving double verification of the test subject's emotion type and avoiding the erroneous indicators brought to the emotion judgment by the accidental appearance of a certain emotion type.

[0071] Step S300 specifically includes:

[0072] Step S301: Using the emotion types corresponding to the first response emotion feature data output by the deep learning model as rows of the first emotion matrix, and using the emotion scores corresponding to the first response emotion feature data as columns of the first emotion matrix;

[0073] Step S302: Using the emotion types corresponding to the second response emotion feature data output by the deep learning model as rows of the second emotion matrix, and using the emotion scores corresponding to the second response emotion feature data as columns of the second emotion matrix;

[0074] Step S303: Fill the first emotion matrix and the second emotion matrix. If the corresponding emotion scores and emotion types do not exist, fill them with 0.

[0075] In this embodiment, the number of rows of the first emotion matrix is ​​7, the number of columns of the first emotion matrix is ​​5, the number of rows of the second emotion matrix is ​​7, and the number of columns of the second emotion matrix is ​​5. An emotion matrix is ​​made for the emotion types and the corresponding emotion scores, so that mathematical operations and weighting can be performed on the emotion matrix, making the emotion judgment operational and digital, and being able to intuitively express the emotion form.

[0076] Step S400 specifically includes:

[0077] Step S401: Input the basic personal information of the depression test subject into the retrieved depression database to obtain the corresponding depression emotion matrix. The expression of the depression emotion matrix is: i ranges from 1 to 4;

[0078] Step S402: Calculate the first emotion matrix Second Emotion Matrix The first depression state evaluation value and the second depression state evaluation value corresponding to the depression mood matrix, the mathematical expression of the first depression state evaluation value is as follows:

[0079]

[0080] Wherein, P1 represents the first depression state assessment value, β j represents the weight of the emotion type of the first emotion matrix, d 1i represents the emotion score of the emotion category of the first emotion matrix, v 1i The emotion score representing the emotion category of the depression emotion matrix;

[0081] The mathematical expression of the second depressive state assessment value is as follows:

[0082]

[0083] Wherein, P2 represents the second depressive state assessment value, γ j represents the weight of the emotion category of the second emotion matrix, and r1i represents the emotion score of the emotion category of the second emotion matrix;

[0084] Step S403, the threshold of depression mood includes a first threshold of depression mood and a second threshold of depression mood, the first depression state evaluation value and the first threshold of depression mood T1 obtain a first depression comparison result, the second depression state evaluation value and the second threshold of depression mood T2 obtain a second depression comparison result; the mental health state of the depression tester is judged according to the first depression comparison result and the second depression comparison result.

[0085] In the embodiment, based on the first response emotion feature data of the depression tester to the test content data, the emotion driving model matches the corresponding driving emotion problem according to the response emotion feature data, obtains the second response emotion feature data of the depression tester to the driving emotion problem, can eliminate the abnormal emotion in the preliminary judgment, makes the process of emotion judgment have a feedback correction effect, improves the comprehensiveness of emotion judgment and avoids the one-sidedness of emotion judgment.

[0086] The emotion categories include fear, sadness, disgust, anger, happiness, calmness and surprise; and the range of the emotion score is 0-1.

[0087] In the embodiment, the range of the emotion score is 0-1, and the size of the emotion score represents the emotion intensity of the corresponding emotion category. When the emotion score is less than 0.5, the emotion intensity of the corresponding emotion category is mild. When the emotion score is greater than 0.5 and less than 0.8, the emotion intensity of the corresponding emotion category is moderate. When the emotion score is greater than 0.8, the emotion intensity of the corresponding emotion category is strong. Subdividing the emotion of the depression tester can increase the accuracy of emotion judgment.

[0088] The specific method for obtaining the voice emotion feature, the eye movement emotion feature, the facial emotion feature and the body emotion feature includes:

[0089] The response question data and the response video data generated by the depression tester when answering the multiple test question data are collected in real time, and the collected response question data and the response video data are put into the pre-trained deep learning model to extract the voice emotion feature and the facial emotion feature;

[0090] The response video data generated by the depression tester when watching the multiple test video data is collected in real time, and the collected response video data is put into the pre-trained deep learning model to extract the facial emotion feature and the body emotion feature;

[0091] The response image data generated by the depression tester when watching the multiple test image data is collected in real time, and the collected response image data is put into the pre-trained deep learning model to extract the eye movement emotion feature;

[0092] Real-time collection of response text data generated by the depression tester when reading multiple test text data, and the collected response text data is put into a pre-trained deep learning model to extract voice emotion features.

[0093] The specific method for obtaining the weight of the emotion category includes: calculating a sub-attention allocation result of the response text data, response image data, response video data, and response question data to depression emotion based on an attention mechanism;

[0094] Based on the sub-attention allocation result, the emotion weights of fear, sadness, disgust, anger, happiness, calmness, and surprise to depression emotion are calculated.

[0095] In the embodiment, even if the emotion categories determined by different methods are the same, the corresponding emotion category has different degrees of persuasion. The persuasion degrees of the response text data, response image data, response video data, and response question data based on the attention mechanism can improve the comprehensiveness and authenticity of emotion determination.

[0096] The specific method for determining the mental health state of the depression tester according to the first depression comparison result and the second depression comparison result includes:

[0097] When P1 is greater than T1 and P2 is greater than T2, the depression is severe, and immediate psychological intervention is needed;

[0098] When P1 is greater than T1 and P2 is less than T2, the depression is moderate, and communication with a psychology teacher is needed;

[0099] When P1 is less than T1 and P2 is greater than T2, the depression is mild, and the depression tester needs to adjust his mood;

[0100] When P1 is less than T1 and P2 is less than T2, the mental health is good, and no intervention is needed.

[0101] In the embodiment, according to the preliminary first depression comparison result of the depression tester and the emotional driving model, the corresponding driving emotional problems are matched according to the first response emotion feature data, the second response emotion feature data of the depression tester to the driving emotional problems is obtained, and the corresponding second depression state evaluation value is calculated. The second depression state evaluation value is compared with the second depression emotion threshold value to obtain the second depression comparison result, which can reflect the degree to which the depression tester can be motivated to have positive emotions. According to the first depression comparison result and the second depression comparison result, the mental health state can be diagnosed to improve the anti-interference of emotion determination.

[0102] By matching the test content data with the depression test subject's basic personal information (including age, gender, and region), targeted and personalized test content data can be provided to the depression test subject, avoiding misjudgments of depression caused by single test content data and individual differences, and improving the accuracy and precision of depression judgment. Based on multiple methods including test text data, test image data, test video data, and test question data, a preliminary judgment of the depression test subject's emotion type and corresponding emotion score is made. Based on the emotion feature data of the depression test subject's first response to the test content data, the emotion-driven model matches the response emotion feature data with the corresponding driving emotion question to obtain the emotion feature data of the depression test subject's second response to the driving emotion question. The first and second response emotion feature data are compared with the pre-set first and second depression emotion thresholds. This can mobilize the depression test subject's emotional development direction, thereby avoiding one-sided emotion judgments when diagnosing depression, improving the anti-interference ability of emotion judgments, and obtaining data that best reflects the depression test subject's true depression state. This facilitates multimodal dual judgment of depression emotion, thereby improving the judgment of the depression test subject's mental health status.

[0103] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a fully hardware embodiment, a fully software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium may be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1one or more processes and / or blocks Figure 1 the function specified in the one or more blocks.

[0104] It should be noted that the above-mentioned embodiments are only used to illustrate the technical solutions of the present application, not to limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalent replaced without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A data method for monitoring depression status, characterized in that: include: Step S100: collecting basic personal information of a depression test subject, and matching corresponding test content data based on the basic personal information; In step S200, the depression test subject responds to the test content data to obtain corresponding response content data, extracts first response emotional feature data of the response content data based on a preset deep learning model, obtains the emotion type and emotion score corresponding to the first response emotional feature data, inputs the first response emotional feature data into a preset emotion-driven model, and obtains second response emotional feature data and the corresponding emotion type and emotion score; specifically, based on the first response emotional feature data of the depression test subject to the test content data, the emotion-driven model matches the corresponding driving emotional question according to the response emotional feature data, and obtains second response emotional feature data of the depression test subject to the driving emotional question; Step S300, establishing an emotion matrix of the depression test subject based on the emotion types and emotion scores, wherein the emotion matrix includes a first emotion matrix and a second emotion matrix, wherein the rows of the emotion matrix are emotion types, and the columns of the emotion matrix are emotion scores corresponding to the response emotion feature data; The step S300 specifically includes: Step S301: Using the emotion types corresponding to the first response emotion feature data output by the deep learning model as rows of the first emotion matrix, and using the emotion scores corresponding to the first response emotion feature data as columns of the first emotion matrix; Step S302: Using the emotion types corresponding to the second response emotion feature data output by the deep learning model as rows of the second emotion matrix, and using the emotion scores corresponding to the second response emotion feature data as columns of the second emotion matrix; Step S303: Fill the first emotion matrix and the second emotion matrix. If the corresponding emotion score and emotion type do not exist, fill them with 0. Step S400, retrieve the depression mood matrix in the depression database, calculate the first depression state assessment value and the second depression state assessment value corresponding to the first emotion matrix, the second emotion matrix and the depression mood matrix, compare the first depression state assessment value and the second depression state assessment value with a preset depression mood threshold to obtain a depression comparison result, and judge the mental health status of the depression tester based on the depression comparison result.

2. The method for monitoring depression according to claim 1, wherein: The test content data includes test text data, test image data, test video data and test question data; The response content data includes response text data, response image data, response video data and response question data; The response emotion feature data includes voice emotion features, eye movement emotion features, facial emotion features and body emotion features.

3. The method for monitoring depression according to claim 2, wherein: Step S100 specifically includes: Step S101, collecting basic personal information of a depression test subject, the basic personal information including the age, gender and region of the depression test subject, inputting the age and gender into a test content database, and the test content database outputting first test content data corresponding to the depression test subject; Step S102: inputting the region into the test content database, the test content database acquiring cultural information data of the region, and outputting second test content data corresponding to the depression test subject according to the cultural information data; Step S103: Merge and splice the first test content data and the second test content data to match corresponding test content data.

4. The method for monitoring depression according to claim 3, wherein: Step S200 specifically includes: Step S201, obtaining test text data, test image data, test video data and test question data corresponding to the depression test subject; Step S202: The depression test subject responds to the test text data, test image data, test video data, and test question data to obtain corresponding response text data, response image data, response video data, and response question data; Step S203: Input the response text data, response image data, response video data, and response question data into a preset deep learning model to extract the first response emotion feature data, and obtain the emotion type and emotion score corresponding to the first response emotion feature data; Step S204: input the first response emotion feature data into a preset emotion-driven model to obtain second response emotion feature data and corresponding emotion types and emotion scores.

5. The method for monitoring depression according to claim 1, wherein: Step S400 specifically includes: Step S401: Input the basic personal information of the depression test subject into the retrieved depression database to obtain the corresponding depression emotion matrix. The expression of the depression emotion matrix is: , i ranges from 1 to 4; Step S402: Calculate the first emotion matrix , Second Emotion Matrix The first depression state evaluation value and the second depression state evaluation value corresponding to the depression mood matrix, the mathematical expression of the first depression state evaluation value is as follows: ； Wherein, P1 represents the first depression state assessment value, represents the weight of the emotion type of the first emotion matrix, represents the emotion score of the emotion category of the first emotion matrix, The emotion score representing the emotion category of the depression emotion matrix; The mathematical expression of the second depressive state assessment value is as follows: ； Wherein, P2 represents the second depression state assessment value, represents the weight of the emotion type of the second emotion matrix, The emotion score representing the emotion category of the second emotion matrix; Step S403, the depression mood threshold includes a first depression mood threshold and a second depression mood threshold, the first depression state assessment value and the first depression mood threshold T1 are compared to obtain a first depression comparison result, and the second depression state assessment value and the second depression mood threshold T2 are compared to obtain a second depression comparison result; the mental health state of the depression tester is judged based on the first depression comparison result and the second depression comparison result.

6. The method for monitoring depression according to claim 5, wherein: The emotion types include fear, sadness, disgust, anger, happiness, calmness and surprise; the emotion score ranges from 0 to 1.

7. The method for monitoring depression according to claim 6, wherein: The specific method of obtaining the voice emotion features, eye movement emotion features, facial emotion features and body emotion features includes: The system collects the response data and video data generated by depression test subjects answering multiple test questions in real time, feeds the collected response data and video data into a pre-trained deep learning model, and extracts voice emotion features and facial emotion features. Real-time depression detection collects the response video data generated by the test subject when watching multiple test video data, and puts the collected response video data into a pre-trained deep learning model to extract facial emotion features and body emotion features; The response image data generated by depression testers when viewing multiple test image data is collected in real time, and the collected response image data is put into a pre-trained deep learning model to extract eye movement emotion features; The response text data generated by depression testers when reading multiple test text data is collected in real time, and the collected response text data is put into a pre-trained deep learning model to extract voice emotion features.

8. The method for monitoring depression according to claim 7, wherein: The specific method for obtaining the weight of the emotion type includes: calculating the sub-attention allocation result of the response text data, response image data, response video data and response question data to the depressive emotion based on the attention mechanism; The emotion weights of fear, sadness, disgust, anger, happiness, calmness and surprise to depression are calculated based on the sub-attention allocation results.

9. The method for monitoring depression according to claim 8, wherein: The specific method for judging the mental health status of the depression test subject according to the first depression comparison result and the second depression comparison result includes: When P1 is greater than T1 and P2 is greater than T2, it indicates severe depression and requires immediate psychological intervention; When P1 is greater than T1 and P2 is less than T2, it is moderate depression and needs to be communicated with a psychology teacher; When P1 is less than T1 and P2 is greater than T2, it is mild depression and the depressed test-taker needs to adjust his / her mood. When P1 is less than T1 and P2 is less than T2, the person is mentally healthy and no intervention is required.

Images