An ai-based mental health assessment system and method

Abstract

The application discloses a mental health assessment system and method based on AI, wherein the method comprises the following steps: authenticating a first wearable device worn by a user, and acquiring first physiological data of the user collected by the first wearable device worn by the user in the recent N consecutive days; acquiring first answering data filled by the user on an electronic terminal for a mental health assessment questionnaire; during the answering of the questionnaire, acquiring second physiological data of the user collected by the wearable device in real time and generating real-time data features; comparing the first physiological data with the real-time data features generated by the second physiological data, and judging whether the first physiological data and the second physiological data are of the same user; if not, terminating the assessment and alarming; if yes, inputting the following data into an AI assessment model: the first answering data and the first physiological data, and outputting a mental health state assessment result and a personalized intervention suggestion of the user. The application can effectively improve the accuracy of mental health assessment.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of mental health assessment, in particular to an AI-based mental health assessment system and method. BACKGROUND

[0002] Mental health assessment is a systematic tool and method that can be used to assess the user's mental state. Questionnaire survey is a common means of mental health assessment, but the questionnaire survey of mental health generally only stays in the scoring of the questionnaire answer content, and it is easy to appear subjective bias and cheating phenomenon (the actual answerer does not match the evaluation target) of the user, resulting in inaccurate evaluation. The following are several common subjective biases: 1. Individuals may tend to conceal their true feelings and give answers that meet social expectations (such as denying depressive emotions or exaggerating positive performance). 2. Questionnaires usually require a review of the mental state within a period of time (such as "the past two weeks"), but memory may not be accurate or influenced by recent emotions. 3. Some people lack the ability to accurately perceive their own emotions, resulting in distorted answers. Therefore, the questionnaire survey of mental health has low accuracy. SUMMARY

[0003] The applicant found that in the questionnaire mental health assessment, the behavior parameters (such as mouse trajectory, answer time, option modification times) can capture the user's potential mental state. If the mental state expressed by the behavior parameters and the mental state expressed by the questionnaire answers can effectively improve the accuracy of mental health assessment.

[0004] In view of the above part of the defects of the prior art, the technical problem to be solved by the present application is to provide an AI-based mental health assessment system and method, which aims to improve the accuracy of mental health assessment.

[0005] To achieve the above-mentioned purpose, the first aspect of the present application discloses an AI-based mental health assessment method, which comprises:

[0006] Step S1, authenticating a first wearable device worn by a user, and acquiring first physiological data of the user collected by the first wearable device worn by the user in the recent N consecutive days; the first physiological data includes heart rate data, blood pressure data, sleep quality data, and skin electricity reaction data;

[0007] Step S2, acquiring first answer data filled by the user on an electronic terminal for a mental health assessment questionnaire; and during the questionnaire answering period, acquiring second physiological data of the user collected by the wearable device in real time and generating real-time data features;

[0008] Step S3, comparing the first physiological data with the real-time data features generated by the second physiological data to determine whether the first physiological data and the second physiological data are of the same user; if not, terminating the evaluation and alarming;

[0009] Step S4, in response to the first physiological data and the second physiological data being of the same user, inputting the first answer data and the first physiological data into an AI evaluation model and outputting the mental health state evaluation result and the personalized intervention suggestion of the user.

[0010] Optionally, the step S3 comprises:

[0011] Step S31, pre-processing and feature extraction are performed on the first physiological data to generate a first multi-dimensional feature vector;

[0012] The first multi-dimensional feature vector comprises a combination of at least two types of features, including:

[0013] resting heart rate, time / frequency domain indicators of heart rate variability (HRV);

[0014] blood pressure trend features and body position change response features; the body position change response features are generated from the three-axis acceleration data of the first wearable device in combination with the blood pressure data;

[0015] skin electrical response amplitude, rise / fall time, and basal skin electrical level;

[0016] Step S32, inputting the first multi-dimensional feature vector into a machine learning model to train and generate a personalized physiological baseline model corresponding to the user;

[0017] Step S33, extracting a second feature vector of the same dimension as the first multi-dimensional feature vector from the second physiological data during the answering of the questionnaire;

[0018] Step S34, calculating a matching score of the second feature vector and the personalized physiological baseline model;

[0019] Step S35, when the matching score exceeds a preset threshold, determining that the current wearer and the registered user are the same person.

[0020] Optionally, the step S31 comprises:

[0021] extracting the mean resting heart rate and the post-exercise heart rate recovery slope from the heart rate data;

[0022] extracting the low frequency power (LF) and high frequency power (HF) ratio (LF / HF) and the adjacent R-R interval difference root mean square (RMSSD) from the heart rate variability analysis;

[0023] The specific behavior event is identified in combination with acceleration data, and an EDA response peak and a blood pressure fluctuation amplitude in a preset time window before and after the event are extracted;

[0024] The personalized physiological baseline model of the step S32 is constructed by using any of the following algorithms:

[0025] A Hidden Markov Model (HMM) is used to model the time sequence transition probability of a physiological state;

[0026] A Support Vector Data Description (SVDD) is used to establish a high-dimensional hyper-sphere boundary of a user's physiological features;

[0027] A Long Short-Term Memory Network (LSTM) is used to learn a multi-parameter joint time sequence dependency pattern;

[0028] The matching degree calculation in the step S34 is performed by using one of the following methods:

[0029] A Dynamic Time Warping (DTW) distance between a new feature vector and a baseline model feature vector is calculated;

[0030] A probability value of belonging to a registered user is output by a pre-trained binary classifier;

[0031] A Mahalanobis Distance is used to measure the deviation degree of new data relative to a baseline model distribution.

[0032] Optionally, the step S2 further includes:

[0033] The step S21, collecting answer content data of a first user to a mental health assessment questionnaire and answer behavior of the first user in the answering process; wherein the answer behavior at least includes a mouse trajectory, an answer duration, and an option modification times;

[0034] The step S22, obtaining a content score of the first user according to the answer content data; wherein the content score is positively or negatively correlated with the mental health assessment of the first user;

[0035] The step S23, obtaining a mouse trajectory complexity and an invalid movement proportion of the first user according to the mouse trajectory, obtaining a cognitive load number of the first user according to the answer duration, obtaining a decision conflict degree according to the option modification times, and obtaining an attention dispersion degree of the first user according to the mouse trajectory, the answer duration, and the option modification times;

[0036] Step S24, obtaining a behavior score of the first user according to the mouse trajectory complexity, the invalid movement proportion, the cognitive load number, the decision conflict degree and the attention dispersion degree; wherein the behavior score is positively or negatively correlated with the mental health assessment of the first user;

[0037] Step S25, obtaining a consistency score of the first user according to the content score and the behavior score; wherein the scores of the content score and the behavior score are positively correlated with the mental health assessment of the first user;

[0038] Step S26, obtaining a mental health assessment result of the first user according to the content score, the behavior score and the consistency score.

[0039] Optionally, the step S24 comprises:

[0040] According to

[0041]

[0042] obtaining the behavior score of the first user; wherein, the behavior score is, 4, the mouse trajectory complexity, a weight of the mouse trajectory complexity, the invalid movement proportion, a weight of the invalid movement proportion, the cognitive load number, a weight of the cognitive load number, the decision conflict degree, a weight of the decision conflict degree, the attention dispersion degree, a weight of the attention dispersion degree.

[0043] Optionally, the step S26 comprises

[0044] According to

[0045]

[0046] obtaining the mental health assessment result; wherein, a score corresponding to the mental health assessment result, the content score, a weight of the content score, the behavior score, a weight of the behavior score, the consistency score, weighting the consistency score.

[0047] Optionally, the method further comprises

[0048] determining the weight corresponding to the content score, the behavior score and the consistency score according to the specific score of the content score, the behavior score and the consistency score; wherein the weight corresponding to the consistency score is positively correlated with the content score and the behavior score.

[0049] Optionally, after the step S25, the method further comprises:

[0050] in response to the consistency score being less than a first consistency threshold, determining that the first user has a cheating behavior; wherein the cheating behavior includes concealing the mental health condition of the first user.

[0051] Optionally, after the step S21, the method further comprises:

[0052] in response to the answering time being less than a minimum answering time threshold, determining that the first user has a random answering behavior.

[0053] Optionally, in the step S23, the cognitive load number of the first user is obtained according to the answering time, comprising:

[0054] obtaining the reading speed of the first user;

[0055] obtaining the cognitive load number of the first user according to the reading speed of the first user and the answering time.

[0056] The second aspect of the present application discloses an AI-based mental health assessment system, which comprises a first physiological data acquisition unit, a second physiological data acquisition unit, a user comparison unit and an assessment unit.

[0057] The first physiological data acquisition unit is configured to authenticate a first wearable device worn by an input user, and acquire first physiological data of the user collected by the first wearable device worn by the user in the recent N consecutive days; the first physiological data comprises heart rate data, blood pressure data, sleep quality data and skin electricity reaction data.

[0058] The second physiological data acquisition unit is configured to acquire first answering data filled by the user on an electronic terminal for a mental health assessment questionnaire; and during the answering of the questionnaire, the second physiological data acquisition unit is configured to acquire second physiological data of the user collected by the wearable device in real time and generate real-time data features.

[0059] The user comparison unit is configured to compare the first physiological data with the real-time data feature generated by the second physiological data, and determine whether the first physiological data and the second physiological data are of the same user; if not, terminate the evaluation and alarm.

[0060] The evaluation unit is configured to, in response to the first physiological data and the second physiological data being of the same user, input the first answer data and the first physiological data into an AI evaluation model, and output a mental health state evaluation result and a personalized intervention suggestion of the user.

[0061] Optionally, the user comparison unit comprises a first multi-dimensional feature vector generation subunit, a model training generation subunit, a second multi-dimensional feature vector extraction subunit, a matching score calculation subunit, and a judgment subunit.

[0062] The first multi-dimensional feature vector generation subunit is configured to pre-process and extract features from the first physiological data to generate a first multi-dimensional feature vector.

[0063] The first multi-dimensional feature vector comprises a combination of at least two types of features, including:

[0064] resting heart rate, time / frequency domain indicators of heart rate variability (HRV);

[0065] blood pressure trend features and postural change response features; the postural change response features are generated from three-axis acceleration data of the first wearable device in combination with the blood pressure data;

[0066] skin conductance response amplitude, rise / fall time, and baseline skin conductance level;

[0067] The model training generation subunit is configured to input the first multi-dimensional feature vector into a machine learning model to train a personalized physiological baseline model corresponding to the user.

[0068] The second multi-dimensional feature vector extraction subunit is configured to extract a second feature vector of the same dimension as the first multi-dimensional feature vector from the second physiological data during the questionnaire answering;

[0069] The matching score calculation subunit is configured to calculate a matching degree score of the second feature vector and the personalized physiological baseline model.

[0070] The judgment subunit is configured to determine that the current wearing user and the registered user are the same person when the matching degree score exceeds a preset threshold.

[0071] Optionally, the first multi-dimensional feature vector generation subunit is specifically configured to:

[0072] extracting resting heart rate mean, post-exercise heart rate recovery slope from the heart rate data;

[0073] extracting low frequency power (LF) to high frequency power (HF) ratio (LF / HF), root mean square of successive differences (RMSSD) from the heart rate variability analysis;

[0074] combining acceleration data to identify specific behavior events, and extracting EDA response peak and blood pressure fluctuation amplitude within a preset time window before and after the event;

[0075] The personalized physiological baseline model in the model training generation subunit is constructed using any of the following algorithms:

[0076] Hidden Markov Model (HMM) is used to model the time series transition probability of physiological state;

[0077] Support Vector Data Description (SVDD) is used to establish a high-dimensional hypersphere boundary of user physiological features;

[0078] Long Short-Term Memory Network (LSTM) is used to learn multi-parameter joint time series dependency pattern;

[0079] The matching degree calculation of the matching score calculation subunit uses one of the following methods:

[0080] Calculate the Dynamic Time Warping (DTW) distance between the new feature vector and the baseline model feature vector;

[0081] Output the probability value belonging to the registered user through the pre-trained binary classifier;

[0082] Based on Mahalanobis Distance, measure the deviation of new data from the baseline model distribution.

[0083] Optionally, the system further comprises a data acquisition module, a content score obtaining module, a behavior parameter obtaining module, a behavior score obtaining module, a consistency score obtaining module, and an evaluation module;

[0084] The data acquisition module is used to acquire the answer content data of the first user to the mental health assessment questionnaire and the answer behavior of the first user in the answering process; wherein the answer behavior at least includes mouse trajectory, answer time length and option modification times;

[0085] The content score obtaining module is used to obtain the content score of the first user according to the answer content data; wherein the content score is positively or negatively correlated with the mental health assessment of the first user;

[0086] The behavior parameter obtaining module is configured to: obtain a mouse track complexity and an invalid movement proportion of the first user according to the mouse track; obtain a cognitive load number of the first user according to the answering time length; obtain a decision conflict degree according to the option modification times; and obtain an attention dispersion degree of the first user according to the mouse track, the answering time length and the option modification times.

[0087] The behavior score obtaining module is configured to: obtain a behavior score of the first user according to the mouse track complexity, the invalid movement proportion, the cognitive load number, the decision conflict degree and the attention dispersion degree; and wherein the behavior score is positively or negatively correlated with the mental health assessment of the first user.

[0088] The consistency score obtaining module is configured to: obtain a consistency score of the first user according to the content score and the behavior score; and wherein the content score and the behavior score are positively correlated with the mental health assessment of the first user.

[0089] The evaluation module is configured to: obtain a mental health assessment result of the first user according to the content score, the behavior score and the consistency score.

[0090] Optionally, the behavior score obtaining module is specifically configured to:

[0091] According to

[0092]

[0093] obtain the behavior score of the first user; wherein, the behavior score is, 4, the mouse track complexity, a weight of the mouse track complexity, the invalid movement proportion, a weight of the invalid movement proportion, the cognitive load number, a weight of the cognitive load number, the decision conflict degree, a weight of the decision conflict degree, the attention dispersion degree, a weight of the attention dispersion degree.

[0094] Optionally, the evaluation module is specifically configured to: according to

[0095]

[0096] obtaining the mental health assessment result; wherein a score corresponding to the mental health assessment result, a content score, a weight of the content score, a behavior score, a weight of the behavior score, a consistency score, a weight of the consistency score.

[0097] Optionally, the system further comprises a first weight allocation module,

[0098] The first weight allocation module is configured to determine the weights corresponding to the content score, the behavior score and the consistency score according to the specific values of the content score, the behavior score and the consistency score; wherein the weight corresponding to the consistency score is positively correlated with the content score and the behavior score.

[0099] Optionally, the system further comprises a cheating judgment module,

[0100] The cheating judgment module is configured to determine that the first user has a cheating behavior in response to the consistency score being less than a first consistency threshold; wherein the cheating behavior includes concealing the mental health condition of the first user.

[0101] Optionally, the system further comprises a question answering behavior judgment module,

[0102] The question answering behavior judgment module is configured to determine that the first user has a random answering behavior in response to the answering time being less than a minimum answering time threshold.

[0103] The beneficial effects of the present application are: 1. The first physiological data collected by the first wearable device in daily life and the second physiological data collected when answering questions can determine whether the user is the same person. This can effectively avoid cheating or wearing errors, and effectively improve the evaluation accuracy. 2. The present application effectively combines the physiological data and the answer data of the user in daily life, and based on the appearance, the potential psychological state of the user is captured for comprehensive evaluation, which improves the accuracy of the evaluation. 3. The present application collects the answering behavior of the user during the answering process, and analyzes the answering behavior. The present application captures the potential psychological state of the user through the answering behavior, evaluates the psychological state of the user, obtains a behavior score, and then judges the consistency according to the behavior score and the content score to obtain a consistency score. The present application effectively combines the content score, the behavior score and the consistency score to comprehensively evaluate the mental health of the user, avoids the one-sidedness of single content score evaluation, effectively reduces the subjective bias, and improves the accuracy of mental health evaluation. 4. The consistency score in the present application evaluates the matching degree of the content score and the behavior score, avoids the deviation of the comprehensive evaluation caused by the too much deviation of the two, and further improves the accuracy of the mental health evaluation. 5. The present application judges that the first user has cheating behavior in response to the consistency score being less than a first consistency threshold. When the consistency score is too small, it means that the user wants to hide the mental health condition of the user, and the present application can timely find and report for review. 6. The present application judges that the first user has a random answering behavior in response to the answering time being less than a minimum answering time threshold. Through this way, the behavior of the user randomly answering in order to make a pretense is avoided. 7. The present application also combines the reading speed of the individual to obtain the cognitive load number, so that the cognitive load number is more accurate, and a more accurate behavior score is obtained.

[0104] In summary, the present application improves the accuracy of mental health evaluation. BRIEF DESCRIPTION OF DRAWINGS

[0105] Fig. 1 is a flowchart of a mental health evaluation method based on AI provided by a specific embodiment of the present application;

[0106] Fig. 2 is a structural schematic diagram of a mental health evaluation system based on AI provided by a specific embodiment of the present application;

[0107] Fig. 3 is a support vector description model diagram of user identification based on a wearable device. DETAILED DESCRIPTION

[0108] The application discloses a mental health assessment system and method based on AI, and a person skilled in the art can improve technical details according to the content of the application. It should be particularly pointed out that all similar substitutions and changes are obvious to a person skilled in the art, and they are regarded as being included in the application. The method and application of the application have been described by the preferred embodiments, and relevant personnel can obviously modify or appropriately change and combine the method and application described herein without departing from the content, spirit and scope of the application, to realize and apply the application technology.

[0109] The applicant has found that in the questionnaire mental health assessment, the potential psychological state of the user can be captured through the behavior parameters (such as the mouse track, the answer time length, the option modification times), and if the psychological state expressed by the behavior parameters and the psychological state expressed by the questionnaire answers can be combined, the accuracy of the mental health assessment can be effectively improved.

[0110] Therefore, the embodiment of the application provides a mental health assessment method based on AI, as shown in the figure, the method comprises the following steps: Figs. 1-3

[0111] Step S1, authenticating a first wearable device worn by a user, and acquiring first physiological data of the user in the recent N consecutive days collected by the first wearable device worn by the user.

[0112] The first physiological data comprises heart rate data, blood pressure data, sleep quality data and skin electricity reaction data.

[0113] It should be noted that the first physiological data can reflect the mental health state of the user in daily life.

[0114] Step S2, acquiring first answer data filled by the user on an electronic terminal for a mental health assessment questionnaire; and acquiring second physiological data of the user collected by the wearable device in real time and generating real-time data features during the answering of the questionnaire.

[0115] It should be noted that the user needs to continuously wear the first wearable device, so that the first wearable device can collect the first physiological data and the second physiological data.

[0116] Step S3, comparing the first physiological data with the real-time data features generated by the second physiological data, and judging whether the first physiological data and the second physiological data are of the same user; if not, terminating the assessment and alarming.

[0117] In the specific embodiment, step S3 comprises the following steps:

[0118] Step S31, pre-processing and feature extraction are performed on the first physiological data to generate a first multi-dimensional feature vector. ​

[0119] wherein the first multi-dimensional feature vector comprises a combination of at least two of the following types of features:

[0120] time / frequency domain indicators of resting heart rate, heart rate variability (HRV) related to heart rate data;

[0121] blood pressure trend features and postural change response features related to blood pressure data; the postural change response features are generated from three-axis acceleration data of the first wearable device in combination with blood pressure data;

[0122] skin conductance response amplitude, rise / fall time, and baseline skin conductance level related to skin conductance response data;

[0123] Step S32, inputting the multi-dimensional feature vector into a machine learning model to train a personalized physiological baseline model corresponding to the user;

[0124] Step S33, extracting a second feature vector of the same dimension as the first multi-dimensional feature vector from the second physiological data during the answering of the questionnaire;

[0125] Step S34, calculating a matching score of the second feature vector and the personalized physiological baseline model;

[0126] Step S35, when the matching score exceeds a preset threshold, determining that the current wearer and the registered user are the same person.

[0127] It is worth mentioning that, in order to avoid misjudgment, in this technical solution, three-axis acceleration data is also considered in combination with resting heart rate and heart rate variability; in fact, the state of the user before answering the questionnaire (whether in motion) is considered. Based on the data used in the first physiological data when the motion state is, it is determined whether the user is in motion before answering the questionnaire, effectively avoiding misjudgment.

[0128] Further, step S31 comprises:

[0129] extracting the mean resting heart rate and the heart rate recovery slope after exercise from the heart rate data;

[0130] extracting the low frequency power (LF) to high frequency power (HF) ratio (LF / HF) and the root mean square of successive differences (RMSSD) of adjacent R-R intervals from the heart rate variability analysis;

[0131] combining acceleration data to identify specific behavior events and extract EDA response peaks and blood pressure fluctuation amplitudes within a preset time window before and after the event;

[0132] The personalized physiological baseline model of step S32 is constructed using any of the following algorithms:

[0133] Hidden Markov Model (HMM) for modeling the time sequence transition probability of physiological state;

[0134] Support Vector Data Description (SVDD) is used to establish a high-dimensional hyper-sphere boundary of the user's physiological characteristics;

[0135] Long Short-Term Memory Network (LSTM) is used to learn multi-parameter joint time series dependency patterns;

[0136] The matching degree calculation in step S34 adopts one of the following methods:

[0137] Calculate the Dynamic Time Warping (DTW) distance between the new feature vector and the baseline model feature vector;

[0138] Output the probability value belonging to the registered user through the pre-trained binary classifier;

[0139] Based on the Mahalanobis Distance, measure the deviation of the new data from the baseline model distribution.

[0140] In addition, the personalized physiological baseline model also includes a model adaptive updating mechanism:

[0141] When the verification is successful for M consecutive times, the new feature vector is merged into the original baseline model with a decay weight;

[0142] When the matching degree score is lower than the warning threshold but higher than the rejection threshold, trigger short-time high-frequency data acquisition to recalibrate the model.

[0143] It should be noted that the user can be accurately judged by artificial intelligence, effectively avoiding the occurrence of cheating phenomenon.

[0144] Step S4, in response to the first physiological data and the second physiological data being the same user, input the following data into the AI evaluation model: the first answer data, the first physiological data, and output the user's mental health state evaluation result and personalized intervention suggestion.

[0145] In this specific embodiment, step S2 further comprises:

[0146] Step S21, collect the first user's answer content data to the mental health assessment questionnaire and the first user's answer behavior during the answering process.

[0147] Among them, the answer behavior at least includes mouse trajectory, answer time length and option modification times.

[0148] It should be noted that when the user is impatient or inattentive, the track complexity and invalid movement proportion in the mouse track will increase. When the user has a psychological problem, the thinking may be delayed, thereby significantly prolonging the time spent on answering, or the time spent on answering negative emotion-related questions may be significantly prolonged due to psychological factors. When the user group lacks accurate awareness of their own emotions, the options may be modified multiple times. Therefore, the user's mental health status can be captured by analyzing these behaviors.

[0149] In this specific embodiment, after step S21, the method further comprises:

[0150] In response to the answering time being less than the minimum answering time threshold, it is determined that the first user has a careless answering behavior.

[0151] It should be noted that the questions of the questionnaire generally need to be read before answering, but some users may randomly select in order to quickly and superficially complete the task, thereby causing the content score to be too distorted. Therefore, this embodiment can effectively prevent the occurrence of such a situation by determining whether the user has such a behavior through the answering time.

[0152] Step S22, obtaining a content score of the first user according to the answering content data.

[0153] The content score is positively or negatively correlated with the mental health assessment of the first user.

[0154] It should be noted that depending on the content of the questionnaire, some questionnaires may be more psychologically healthy with a higher score, while some questionnaires may be less psychologically healthy with a higher score, depending on the actual situation. However, there is monotonicity between the score and the mental health situation in any case.

[0155] Step S23, obtaining a mouse track complexity and an invalid movement proportion of the first user according to the mouse track; obtaining a cognitive load number of the first user according to the answering time; obtaining a decision conflict degree according to the number of option modifications; and obtaining an attention dispersion degree of the first user according to the mouse track, the answering time, and the number of option modifications.

[0156] It should be noted that the greater the answering time, the greater the decline in the user's cognitive ability (possibly caused by psychological factors), thereby the greater the cognitive load number, which is defined by a specific numerical value. The greater the number of option modifications, the greater the decision conflict degree, which is defined by a specific numerical value. The above behaviors may all be caused by attention, and the greater the inattention, the greater the answering time, mouse track, and number of option modifications, and psychological problems are also likely to cause greater inattention.

[0157] In this embodiment, according to the mouse trajectory, the mouse trajectory complexity and the invalid movement proportion of the first user are obtained, including:

[0158] According to the mouse trajectory, the mouse trajectory complexity of the first user is obtained.

[0159] According to the initial position and the final position of the mouse, the effective movement distance is determined; according to the mouse trajectory, the total movement distance is obtained; and according to the effective movement distance and the total movement distance, the invalid movement proportion is obtained.

[0160] It should be noted that when the user has problems such as impatience, impetuosity, and lack of concentration, the mouse trajectory complexity and the invalid movement proportion are likely to increase. In the embodiment of the application, the mouse trajectory complexity and the effective movement distance are defined by specific numerical values.

[0161] In this embodiment, according to the answer time length, the cognitive load number of the first user is obtained in step S23, including:

[0162] The reading speed of the first user is obtained.

[0163] According to the reading speed of the first user and the answer time length, the cognitive load number of the first user is obtained.

[0164] It should be noted that different users have different reading speeds, so by collecting the reading speed and comprehensively considering the reading speed, the cognitive load number can be more accurate.

[0165] In this embodiment, a simple common sense question can be set before the questionnaire to let the user answer to collect the reading speed of the user.

[0166] Step S24, according to the mouse trajectory complexity, the invalid movement proportion, the cognitive load number, the decision conflict degree, and the attention dispersion degree, the behavior score of the first user is obtained.

[0167] The behavior score is positively or negatively correlated with the psychological health assessment of the first user.

[0168] It should be noted that the behavior score and the content score are the same, and there is monotonicity between the score and the psychological health condition.

[0169] Step S25, according to the content score and the behavior score, the consistency score of the first user is obtained.

[0170] The score of the content score and the behavior score is positively correlated with the psychological health assessment of the first user.

[0171] In this embodiment, after step S25, the method further includes:

[0172] In response to the consistency score being less than the first consistency threshold, it is determined that the first user has a cheating behavior; wherein the cheating behavior includes concealing the mental health condition of the first user.

[0173] It should be noted that when the consistency is too small, it indicates that the first user has a problem of concealing the mental health condition of the first user, which can be discovered in time and a plan can be made.

[0174] It should be noted that the same direction correlation is that when the content score is higher, the user's mental health is better, and when the behavior score is higher, the user's mental health is better. Or when the content score is higher, the user's mental health is worse, and when the behavior score is higher, the user's mental health is worse.

[0175] Step S26, obtaining a mental health assessment result of the first user according to the content score, the behavior score and the consistency score.

[0176] In the first specific embodiment, step S24 includes:

[0177] According to

[0178]

[0179] obtaining a behavior score of the first user; wherein, the behavior score is, 4, a mouse trajectory complexity, a weight of the mouse trajectory complexity, an invalid movement proportion, a weight of the invalid movement proportion, a cognitive load number, a weight of the cognitive load number, a decision conflict degree, a weight of the decision conflict degree, an attention dispersion degree, a weight of the attention dispersion degree.

[0180] In the first specific embodiment, step S26 includes

[0181] According to

[0182]

[0183] obtaining a mental health assessment result; wherein, the mental health assessment result corresponds to a score, the content score, a weight of the content score, the behavior score, a weight of the behavior score, the consistency score, weighting the consistency score.

[0184] In a first specific embodiment, the method further comprises

[0185] determining weights corresponding to the content score, the behavior score and the consistency score according to specific values of the content score, the behavior score and the consistency score; wherein the weight corresponding to the consistency score is positively correlated with the content score and the behavior score.

[0186] The first physiological data collected by the first wearable device in daily life and the second physiological data collected by the first wearable device when answering questions can be used to determine whether the user is the same person, so that cheating or wearing errors can be effectively avoided, and the evaluation accuracy is effectively improved.

[0187] It is found through the above embodiments that the embodiments effectively combine the physiological data of the user in daily life and the answering data, and based on the appearance, the potential psychological state of the user can be captured for comprehensive evaluation, and the evaluation accuracy is improved.

[0188] The embodiments collect the answering behavior of the user during the answering process, and analyze the answering behavior. The embodiments capture the potential psychological state of the user through the answering behavior, evaluate the psychological state of the user, obtain a behavior score, and then determine the consistency according to the behavior score and the content score to obtain a consistency score. The embodiments effectively combine the content score, the behavior score and the consistency score to comprehensively evaluate the mental health of the user, avoid the one-sidedness of single content score evaluation, effectively reduce subjective bias, and improve the accuracy of mental health evaluation.

[0189] The consistency score in the embodiments evaluates the matching degree of the content score and the behavior score, avoids the deviation of the comprehensive evaluation caused by the too large deviation of the two, and further improves the accuracy of the mental health evaluation.

[0190] The embodiments determine that the first user has cheating behavior in response to the consistency score being less than a first consistency threshold. When the consistency score is too small, it indicates that the user wants to conceal the mental health condition, and the embodiments can timely discover and report for review.

[0191] The embodiments determine that the first user has random answering behavior in response to the answering time being less than a minimum answering time threshold. In this way, the user's random answering behavior for expediency is avoided.

[0192] The embodiments further combine the reading speed of the individual to obtain the cognitive load number, so that the cognitive load number is more accurate, and a more accurate behavior score is obtained.

[0193] In summary, the embodiment of the present application improves the accuracy of mental health assessment.

[0194] Based on the above-mentioned mental health assessment method, the embodiment of the present application further provides an AI-based mental health assessment system, as shown in the figure, the system comprises a first physiological data acquisition unit 201, a second physiological data acquisition unit 202, a user comparison unit 203 and an evaluation unit 204. Fig. 2

[0195] The first physiological data acquisition unit 201 is used for authenticating the first wearable device worn by the user and acquiring the first physiological data of the user collected by the first wearable device worn by the user in the recent N consecutive days; the first physiological data comprises heart rate data, blood pressure data, sleep quality data and skin electric response data.

[0196] The second physiological data acquisition unit 202 is used for acquiring the first answer data filled by the user on the electronic terminal for the mental health assessment questionnaire; and during the answering of the questionnaire, the real-time physiological data of the user collected by the wearable device is acquired and real-time data features are generated.

[0197] The user comparison unit 203 is used for comparing the first physiological data with the real-time data features generated by the second physiological data to determine whether the first physiological data and the second physiological data are of the same user; if they are not of the same user, the evaluation is terminated and an alarm is given.

[0198] The evaluation unit 204 is used for inputting the first answer data and the first physiological data into the AI evaluation model in response to the first physiological data and the second physiological data being of the same user, and outputting the mental health state evaluation result and the individualized intervention suggestion of the user.

[0199] Optionally, the user comparison unit 203 comprises a first multi-dimensional feature vector generation subunit, a model training generation subunit, a second multi-dimensional feature vector extraction subunit, a matching score calculation subunit and a judgment subunit.

[0200] The first multi-dimensional feature vector generation subunit is used for pre-processing and feature extraction according to the first physiological data to generate a first multi-dimensional feature vector.

[0201] The first multi-dimensional feature vector comprises a combination of at least two types of features, including:

[0202] Resting heart rate, time / frequency domain indicators of heart rate variability (HRV) related to heart rate data;

[0203] Blood pressure trend features and body position change response features related to blood pressure data; the body position change response features are generated by combining the three-axis acceleration data of the first wearable device with the blood pressure data. ​

[0204] Skin conductance response amplitude, rise / fall time, and baseline skin conductance level associated with the skin conductance response data;

[0205] The model training generation subunit is configured to input the multi-dimensional feature vector into a machine learning model to train and generate a personalized physiological baseline model corresponding to the user;

[0206] The second multi-dimensional feature vector extraction subunit is configured to extract a second feature vector of the same dimension as the first multi-dimensional feature vector from the second physiological data during the answering of the questionnaire;

[0207] The matching score calculation subunit is configured to calculate a matching degree score of the second feature vector and the personalized physiological baseline model;

[0208] The judgment subunit is configured to determine that the current wearer and the registered user are the same person when the matching degree score exceeds a preset threshold.

[0209] Optionally, the first multi-dimensional feature vector generation subunit is specifically configured to:

[0210] extract the mean resting heart rate, post-exercise heart rate recovery slope from the heart rate data;

[0211] extract the low frequency power (LF) and high frequency power (HF) ratio (LF / HF), and the root mean square of successive differences (RMSSD) of adjacent R-R intervals from the heart rate variability analysis;

[0212] combine the acceleration data to identify specific behavior events, and extract the EDA response peak value and blood pressure fluctuation amplitude within a preset time window before and after the event;

[0213] The personalized physiological baseline model in the model training generation subunit is constructed using any of the following algorithms:

[0214] Hidden Markov Model (HMM) for modeling the time series transition probability of physiological state;

[0215] Support Vector Data Description (SVDD) for establishing a high-dimensional hyper-sphere boundary of user physiological features;

[0216] Long Short-Term Memory Network (LSTM) for learning multi-parameter joint time series dependency patterns;

[0217] The matching degree calculation of the matching score calculation subunit uses one of the following methods:

[0218] Calculate the Dynamic Time Warping (DTW) distance between the new feature vector and the baseline model feature vector;

[0219] Output the probability value belonging to the registered user through a pre-trained binary classifier;

[0220] based on mahalanobis distance to measure the deviation of new data from the baseline model distribution.

[0221] Optionally, the system further comprises a data collection module, a content score obtaining module, a behavior parameter obtaining module, a behavior score obtaining module, a consistency score obtaining module, and an evaluation module.

[0222] The data collection module is configured to collect the answer content data of the first user to the mental health assessment questionnaire and the answer behavior of the first user in the answering process, wherein the answer behavior at least includes the mouse trajectory, the answer duration, and the option modification times.

[0223] The content score obtaining module is configured to obtain the content score of the first user according to the answer content data, wherein the content score is positively or negatively correlated with the mental health assessment of the first user.

[0224] The behavior parameter obtaining module is configured to obtain the mouse trajectory complexity and the invalid movement proportion of the first user according to the mouse trajectory, obtain the cognitive load number of the first user according to the answer duration, obtain the decision conflict degree according to the option modification times, and obtain the attention dispersion degree of the first user according to the mouse trajectory, the answer duration, and the option modification times.

[0225] The behavior score obtaining module is configured to obtain the behavior score of the first user according to the mouse trajectory complexity, the invalid movement proportion, the cognitive load number, the decision conflict degree, and the attention dispersion degree, wherein the behavior score is positively or negatively correlated with the mental health assessment of the first user.

[0226] The consistency score obtaining module is configured to obtain the consistency score of the first user according to the content score and the behavior score, wherein the scores of the content score and the behavior score are positively correlated with the mental health assessment of the first user.

[0227] The evaluation module is configured to obtain the mental health assessment result of the first user according to the content score, the behavior score, and the consistency score.

[0228] Optionally, the behavior score obtaining module is specifically configured to:

[0229] obtain the behavior score of the first user according to the mouse trajectory complexity, the invalid movement proportion, the cognitive load number, the decision conflict degree, and the attention dispersion degree.

[0230]

[0231] obtain the behavior score of the first user according to the mouse trajectory complexity, the invalid movement proportion, the cognitive load number, the decision conflict degree, and the attention dispersion degree. the behavior score is the sum of the mouse trajectory complexity, the invalid movement proportion, the cognitive load number, the decision conflict degree, and the attention dispersion degree. the mouse trajectory complexity is 4, the mouse trajectory complexity is the mouse trajectory complexity, the weight of the mouse trajectory complexity is 0.2, the invalid movement proportion is the invalid movement proportion, is a weight of the invalid mobile proportion, is a number of cognitive loads, is a weight of the number of cognitive loads, is a decision conflict degree, is a weight of the decision conflict degree, is an attention dispersion degree, is a weight of the attention dispersion degree.

[0232] Optionally, the evaluation module is specifically configured to: according to

[0233]

[0234] obtain a mental health evaluation result; wherein, is a score corresponding to the mental health evaluation result, is a content score, is a weight of the content score, is a behavior score, is a weight of the behavior score, is a consistency score, is a weight of the consistency score.

[0235] Optionally, the system further comprises a first weight allocation module,

[0236] The first weight allocation module is configured to determine weights corresponding to the content score, the behavior score and the consistency score according to specific values of the content score, the behavior score and the consistency score; wherein the weight corresponding to the consistency score is positively correlated with the content score and the behavior score.

[0237] Optionally, the system further comprises an cheating judgment module,

[0238] The cheating judgment module is configured to judge that the first user has cheating behavior in response to the consistency score being less than a first consistency threshold; wherein the cheating behavior includes concealing the mental health condition of the first user.

[0239] Optionally, the system further comprises a test behavior judgment module,

[0240] The test behavior judgment module is configured to judge that the first user has random answering behavior in response to the test duration being less than a minimum test duration threshold.

[0241] The first physiological data collected by the first wearable device in daily life and the second physiological data collected by the first wearable device during the test can be used to determine whether the user is the same person. This can effectively avoid cheating or wearing errors, and effectively improve the evaluation accuracy.

[0242] The embodiment of the application effectively combines physiological data and answer data of the user in daily life, and based on the appearance, the potential psychological state of the user can be captured for comprehensive evaluation, and the accuracy of the evaluation is improved.

[0243] The embodiment of the application collects the answering behavior of the user in the answering process, and analyzes the answering behavior. The embodiment of the application captures the potential psychological state of the user through the answering behavior, evaluates the psychological state of the user, and obtains a behavior score; and then, according to the behavior score and the content score, consistency is judged, and a consistency score is obtained. The embodiment of the application effectively combines the content score, the behavior score and the consistency score to comprehensively evaluate the mental health of the user, avoids one-sidedness of single content score evaluation, effectively reduces subjective bias, and improves the accuracy of mental health evaluation.

[0244] The consistency score in the embodiment of the application evaluates the matching degree of the content score and the behavior score, avoids the deviation of the comprehensive evaluation caused by the too much deviation of the two, and further improves the accuracy of the mental health evaluation.

[0245] The embodiment of the application judges that the first user has cheating behavior in response to the consistency score being less than a first consistency threshold. When the consistency score is too small, it indicates that the user wants to conceal the mental health condition of the user, and the embodiment of the application can timely find and report for review.

[0246] The embodiment of the application judges that the first user has random answering behavior in response to the answering time being less than a minimum answering time threshold. Through this mode, the behavior of the user randomly answering in order to make a pretense is avoided.

[0247] The embodiment of the application also combines the reading speed of the individual to obtain the cognitive load number, so that the cognitive load number is more accurate, and a more accurate behavior score is obtained.

[0248] To illustrate the user identity verification process based on heart rate, blood pressure and skin electric reaction data, the following supplementary data input format and examples are given.

[0249] a, original data input format

[0250] The system needs to collect the time sequence physiological data of the user for N days (such as 7 days), and record the time stamp, data type and auxiliary information (such as motion state). The following is a typical input format (for example, in JSON):

[0251] {

[0252] "user_id": "U20250512001", / / User unique identifier

[0253] "device_id": "WATCH-007", / / Wearable device ID

[0254] "data": [

[0255] {

[0256] "timestamp": "2025-05-10 08:00:00", / / Timestamp (Year-Month-Day Hour:Minute:Second)

[0257] "data_type": "heart_rate", / / Data Type (Heart Rate)

[0258] "value": 72, / / Heart Rate Value (BPM, Beats Per Minute)

[0259] "state": "rest" / / Action State (Rest / Sleep / Exercise, etc.)

[0260] },

[0261] {

[0262] "timestamp": "2025-05-10 08:05:00",

[0263] "data_type": "blood_pressure", / / Blood Pressure

[0264] "systolic": 120, / / Systolic Pressure (mmHg)

[0265] "diastolic": 80, / / Diastolic Pressure (mmHg)

[0266] "state": "rest"

[0267] },

[0268] {

[0269] "timestamp": "2025-05-10 08:10:00",

[0270] "data_type": "eda", / / Electrodermal Activity (EDA)

[0271] "value": 1.2, / / Skin Conductance (μS, Microsiemens)

[0272] "state": "rest"

[0273] },

[0274] {

[0275] "timestamp": "2025-05-10 09:30:00",

[0276] "data_type": "accelerometer", / / Three-axis acceleration (aids in motion detection)

[0277] "x": 0.1, "y": 0.2, "z": 9.8 / / Acceleration components (m / s²)

[0278] }]

[0279] }

[0280] b. Specific data example (excerpt of first physiological data for 7 consecutive days)

[0281] The following is a user's resting physiological data for 3 consecutive days (N=7 days, only the first 3 days are shown), used to generate a personalized physiological baseline model:

[0282] 1. Heart rate data (resting state) (Note: The actual data volume is larger than this)

[0283] Table 1. Heart Rate Data Illustration

[0284]

[0285] Feature extraction example:

[0286] Mean resting heart rate: (72+75+68+70+73+71) / 6 ≈ 71.8 BPM;

[0287] Heart rate variability (HRV): The root mean square (RMSSD) of the difference between adjacent RR intervals is 5.2 ms (reflecting the stability of autonomic nervous activity).

[0288] 2. Blood pressure data (resting state)

[0289] Table 2. Blood Pressure Data Illustration

[0290]

[0291] Feature extraction example:

[0292] Blood pressure trend characteristics: mean systolic blood pressure 120 mmHg, standard deviation ±1.5 mmHg (high stability);

[0293] Postural change response (combined with acceleration data): When the user stands up from a sitting position (acceleration z-axis from 9.8→8.5m / s²), systolic blood pressure increases by 5 mmHg (120→125) and diastolic blood pressure increases by 3 mmHg (80→83), which is consistent with the normal postural blood pressure change pattern.

[0294] 3. Electrodermal activity (EDA) data (resting state)

[0295] Table 3. Electrodermal activity data table

[0296]

[0297] Feature extraction example:

[0298] Base electrodermal activity average: 1.3 μS;

[0299] Electrodermal response amplitude (peak value during emotional fluctuation - base value): 1.5-1.2 = 0.3 μS (mild tension) at 15:00;

[0300] Rise time (time from base value to peak value): about 30 seconds (consistent with normal emotional activation speed).

[0301] c. Specific implementation of data comparison

[0302] When the user answers the questionnaire, the system collects the second physiological data in real time (the format is consistent with the first physiological data), extracts the same dimension feature vector (such as resting heart rate average, HRV-RMSSD, blood pressure trend, body position response, and base EDA level), and calculates the matching degree with the personalized baseline model (such as LSTM trained time sequence model).

[0303] Comparison result: true user second physiological data features: resting heart rate 70 BPM (baseline average 71.8, deviation +2.5%), HRV-RMSSD 5.0 ms (baseline 5.2, deviation -3.8%), blood pressure trend 121 / 81 mmHg (baseline 120 / 80, deviation +0.8%), base EDA 1.2 μS (baseline 1.3, deviation -7.7%); Matching degree score (Mahalanobis distance): 0.85 (threshold set to 0.8), determined as the same user. For example Fig. 3 Support vector description model diagram for user recognition based on wearable devices, wherein test point A is within the high-dimensional hyper-sphere boundary of the registered user physiological features, and it is determined that the user wearing the device is the same user as the registered user; and test point B is outside the high-dimensional hyper-sphere boundary of the registered user physiological features, and it is determined as an abnormal user.

[0304] In summary, the embodiment of the present application improves the accuracy of mental health assessment.

[0305] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0306] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0307] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. An AI-based mental health assessment system, characterized in that, The system is configured to perform the following steps: Step S1, authenticating a first wearable device worn by an input user, and obtaining first physiological data of the user collected by the first wearable device worn by the user in the recent N consecutive days; The first physiological data includes heart rate data, blood pressure data, sleep quality data, and skin electrical response data; Step S2, obtaining first answer data filled by the user on an electronic terminal for a psychological health assessment questionnaire; and during the answering of the questionnaire, obtaining second physiological data of the user collected by the wearable device in real time and generating real-time data features; Step S3, comparing the first physiological data with the real-time data features generated by the second physiological data to determine whether the first physiological data and the second physiological data are of the same user; if they are not the same user, terminating the evaluation and alarming; Step S4, in response to the first physiological data and the second physiological data being of the same user, inputting the following data into an AI evaluation model: the first answer data, the first physiological data, and outputting the user's psychological health status evaluation result and personalized intervention suggestion; The step S2 further comprises: Step S21, collecting answer content data of a first user to a psychological health assessment questionnaire and answer behavior of the first user during the answering process; wherein the answer behavior at least includes mouse trajectory, answer duration and option modification times; Step S22, obtaining a content score of the first user according to the answer content data; wherein the content score is positively or negatively correlated with the psychological health assessment of the first user; Step S23, obtaining mouse trajectory complexity and invalid movement proportion of the first user according to the mouse trajectory; obtaining cognitive load of the first user according to the answer duration; obtaining decision conflict degree according to the option modification times; and obtaining attention dispersion degree of the first user according to the mouse trajectory, the answer duration and the option modification times; Step S24, obtaining a behavior score of the first user according to the mouse trajectory complexity, the invalid movement proportion, the cognitive load number, the decision conflict degree and the attention dispersion degree; wherein the behavior score is positively or negatively correlated with the mental health assessment of the first user; comprising: according to obtaining the behavior score of the first user; wherein, the behavior score is, 4, the mouse trajectory complexity is, the weight of the mouse trajectory complexity is, the invalid movement proportion is, the weight of the invalid movement proportion is, the cognitive load number is, the weight of the cognitive load number is, the decision conflict degree is, the weight of the decision conflict degree is, the attention dispersion degree is, the weight of the attention dispersion degree is; Step S25, obtaining a consistency score of the first user according to the content score and the behavior score; wherein the content score and the behavior score are positively correlated with the psychological health assessment of the first user; Step S26, obtaining a mental health assessment result of the first user according to the content score, the behavior score and the consistency score; including obtaining the mental health assessment result according to obtaining the mental health assessment result; wherein, a score corresponding to the mental health assessment result, the content score, a weight of the content score, the behavior score, a weight of the behavior score, the consistency score, a weight of the consistency score. 2.The AI-based mental health assessment system of claim 1, wherein, The step S3 comprises: Step S31, preprocessing and feature extraction of the first physiological data to generate a first multi-dimensional feature vector; The first multi-dimensional feature vector contains a combination of at least two types of features: Resting heart rate, time / frequency domain indicators of heart rate variability HRV; Blood pressure trend features and body position change response features; the body position change response features are generated from three-axis acceleration data of the first wearable device combined with the blood pressure data; Skin electrical response amplitude, rise / fall time and basic skin electrical level; Step S32, inputting the first multi-dimensional feature vector into a machine learning model to train and generate a personalized physiological baseline model corresponding to the user; Step S33, extracting a second feature vector of the same dimension as the first multi-dimensional feature vector from the second physiological data during the answering of the questionnaire; Step S34, calculating a matching score of the second feature vector and the personalized physiological baseline model; Step S35, determining that the current wearing user and the registered user are the same person when the matching score exceeds a preset threshold. 3.The AI-based mental health assessment system of claim 2, wherein, The step S31 includes: extracting the mean resting heart rate, post-exercise heart rate recovery slope from the heart rate data; extracting the low frequency power LF and high frequency power HF ratio LF / HF, the adjacent R-R interval difference root mean square RMSSD from the heart rate variability analysis; combining the acceleration data to identify specific behavior events, and extracting the EDA response peak and blood pressure fluctuation amplitude within a preset time window before and after the event; The personalized physiological baseline model of the step S32 is constructed by any of the following algorithms: Hidden Markov Model HMM, used to model the time sequence transition probability of physiological state; Support Vector Data Description SVDD, used to establish a high-dimensional hyper-sphere boundary of user physiological features; Long Short-Term Memory Network LSTM, used to learn multi-parameter joint time sequence dependence mode; The matching calculation in the step S34 adopts one of the following methods: Calculate the dynamic time warping DTW distance between the new feature vector and the baseline model feature vector; Output the probability value belonging to the registered user through the pre-trained binary classifier; Based on the Mahalanobis Distance, measure the deviation of the new data relative to the baseline model distribution. 4.The AI-based mental health assessment system of claim 1, wherein, The system is also configured to: According to the specific scores of the content score, the behavior score and the consistency score, determine the weights corresponding to the content score, the behavior score and the consistency score; wherein the weight corresponding to the consistency score is positively correlated with the content score and the behavior score. 5.The AI-based mental health assessment system of claim 1, wherein, After the step S25, the system is also configured to: In response to the consistency score being less than a first consistency threshold, determine that the first user has cheating behavior; wherein the cheating behavior includes concealing the user's mental health situation. 6.The AI-based mental health assessment system of claim 1, wherein, After the step S21, the system is also configured to: In response to the answering time being less than a minimum answering time threshold, determine that the first user has a random answering behavior.

7. An AI-based mental health assessment system, characterized by, The system includes: a first physiological data acquisition unit, a second physiological data acquisition unit, a user comparison unit, and an evaluation unit; The first physiological data acquisition unit is configured to authenticate a first wearable device worn by an input user, and acquire first physiological data of the user collected by the first wearable device worn by the user for N consecutive days in the near future; the first physiological data includes heart rate data, blood pressure data, sleep quality data, and skin electric response data. The second physiological data acquisition unit is configured to acquire first answering data filled by a user on an electronic terminal for a mental health assessment questionnaire; and during the answering of the questionnaire, the wearable device collects real-time physiological data of the user and generates real-time data features. The user comparison unit is configured to compare the first physiological data with the real-time data feature generated by the second physiological data, and determine whether the first physiological data and the second physiological data are of the same user; if not, terminate the evaluation and alarm; The evaluation unit is configured to, in response to the first physiological data and the second physiological data being of the same user, input the first answer data and the first physiological data into an AI evaluation model, and output a mental health state evaluation result and a personalized intervention suggestion of the user; The second physiological data acquisition unit is configured to: acquire answer content data of a first user to a mental health evaluation questionnaire and answer behavior of the first user in the answering process; wherein the answer behavior at least includes a mouse trajectory, an answer duration and an option modification frequency; obtain a content score of the first user according to the answer content data; wherein the content score is positively or negatively correlated with the mental health evaluation of the first user; obtain a mouse trajectory complexity and an invalid movement proportion of the first user according to the mouse trajectory, obtain a cognitive load of the first user according to the answer duration, obtain a decision conflict degree according to the option modification frequency, and obtain an attention dispersion degree of the first user according to the mouse trajectory, the answer duration and the option modification frequency; According to the mouse trajectory complexity, the invalid movement proportion, the cognitive load number, the decision conflict degree, and the attention dispersion degree, obtain a behavior score of the first user; wherein the behavior score is positively or negatively correlated with the mental health assessment of the first user; comprising: obtain the behavior score of the first user; wherein, the behavior score is, 4, the mouse trajectory complexity is, the weight of the mouse trajectory complexity is, the invalid movement proportion is, the weight of the invalid movement proportion is, the cognitive load number is, the weight of the cognitive load number is, the decision conflict degree is, the weight of the decision conflict degree is, the attention dispersion degree is, the weight of the attention dispersion degree is; obtain a consistency score of the first user according to the content score and the behavior score; wherein the content score and the behavior score are positively correlated with the mental health evaluation of the first user. According to the content score, the behavior score, and the consistency score, obtaining a mental health assessment result of the first user; including according to obtaining the mental health assessment result; wherein, a score corresponding to the mental health assessment result, the content score, a weight of the content score, the behavior score, a weight of the behavior score, the consistency score, a weight of the consistency score.

Images