An emotion intervention and psychological monitoring data analysis method and system based on AI and VR technology

Abstract

The application discloses an emotion intervention and psychological monitoring data analysis method and system based on AI and VR technologies, relates to the technical field of data processing, and comprises the following steps: acquiring psychological test data of a tester, and generating a first analysis result; analyzing and processing face feature data of the tester based on an emotion recognition model to obtain an emotion recognition result of the tester; processing the emotion recognition result of the tester based on an emotion analysis model to generate a second analysis result; and performing data fusion on the first analysis result and the second analysis result through a data fusion model to generate a psychological test analysis result of the tester. Through the algorithms of the two core technologies of AI and VR, the application can realize listening to the confession of a testee, communicating and interacting with the testee, and completing screening of patients with mental disorders, and then performing health detection analysis and intervention treatment.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a multi-dimensional assessment system for attention deficit hyperactivity disorder (ADHD), and particularly to a data analysis method and system for emotion intervention and psychological monitoring based on AI and VR technologies. Background Technology

[0002] Currently, traditional psychological analysis and assessment are mostly based on human experience. Professionals conduct face-to-face questioning with test subjects. During the test, the professionals comprehensively analyze the test subjects' answers to various questions on the psychological test scale and their performance during the questioning process to obtain the test subjects' psychological analysis results. This psychological testing method relies heavily on human experience, and the testing efficiency is low when the number of test subjects is large, and the accuracy of the psychological test will also decrease.

[0003] Meanwhile, society is increasingly emphasizing the development of students' mental health, and with societal progress and development, the monitoring of students' mental health is receiving growing attention. Currently, mental health testing for students often relies on psychological testing scales, with counselors conducting face-to-face interviews and analyzing the students' expressions and states during these interviews using their experience and knowledge. This analysis of psychological monitoring data is overly dependent on manual intervention, and the lack of qualified professionals hinders the widespread adoption of mental health monitoring. As can be seen from the above, for the general public and students, reducing the reliance on professional personnel for psychological monitoring data is one of the key factors for the widespread adoption of psychological monitoring technology. Traditional assessments, such as those for attention deficit hyperactivity disorder (ADHD), mainly rely on the standards defined in the Diagnostic and Statistical Manual of Mental Disorders (DSM), the patient's subjective information, the results of relevant assessment scales, and the comprehensive judgment of psychiatrists. This assessment process still heavily depends on professional medical personnel, must be conducted in a hospital clinic, and requires a professional environment and the participation of professionals. At the same time, traditional assessments are subjective, and the results are qualitative, lacking rigorous and unified quantitative evidence. Furthermore, the results are highly influenced by the patient's ability and willingness to describe the problem, their level of self-awareness, and the professional competence of the assessing physician. Patients are very likely to provide distorted information due to their cognitive abilities, level of expression, and social approval. On the other hand, the scarcity of psychiatric resources and the uneven professional skills of doctors in my country may also lead to diagnostic errors in ADHD.

[0004] Previously, the medical community had attempted a few objective assessment techniques for attention deficit hyperactivity disorder (ADHD), the most typical of which was a method combining continuous performance tasks with body motion tracking technology to comprehensively determine the condition by analyzing the patient's task performance and limb movement intensity. However, this type of method has relatively simple task content, assessment indicators, and assessment dimensions, and can only judge the condition based on the individual's state in a specific scenario of a single test, lacking representativeness and failing to comprehensively reflect the full picture of the condition. Moreover, this type of technology requires specially arranged venues to eliminate image interference and highly specialized equipment for data collection, resulting in poor ecological sustainability in terms of results and feasibility. Therefore, there is an urgent need for a psychological monitoring data analysis method that is less dependent on professional personnel to improve the efficiency of psychological monitoring data analysis and processing, and to achieve standardized and accurate assessment.

[0005] Against the above background, this invention application provides a method and system for emotion intervention and psychological monitoring data analysis based on AI and VR technology. This method and system, combined with an intelligent assessment booth, realizes the replacement of manual psychological testing of test subjects with AI and VR technology, and analyzes the psychological monitoring data of test subjects during the testing process, with low dependence on professionals. Summary of the Invention

[0006] This application provides a method and system for emotion intervention and psychological monitoring data analysis based on AI and VR technologies. It uses AI and VR technologies to conduct psychological tests and data analysis on test subjects, thereby improving the efficiency of psychological monitoring data analysis and achieving standardized and accurate assessment.

[0007] First, this application discloses a data analysis method for emotion intervention and psychological monitoring based on AI and VR technologies, including: The test subjects' psychological test data are obtained. The psychological test data includes the test subjects' question and answer data on the psychological test scale and the test subjects' facial feature data during the psychological test. The test subjects' facial feature data is recorded based on VR devices. The psychological test scale includes multiple test angles, and each test angle includes at least one test question. The test taker's question-and-answer data on the psychological test scale is input into the question-and-answer analysis model to generate the first analysis result, which includes the score for each test angle. The facial feature data of the tester is analyzed and processed based on the emotion recognition model to obtain the emotion recognition result of the tester. The emotion recognition model is a deep learning model, and the emotion recognition result includes the emotional features of the tester in the process of answering each test question. The emotion recognition results of the test subjects are processed based on the emotion analysis model to generate a second analysis result; The data fusion model is used to fuse the first and second analysis results to generate the psychological test analysis results of the test taker.

[0008] Furthermore, the construction of sentiment analysis models includes: Historical test data of multiple users who have taken psychological tests based on psychological test scales were obtained. The scores of each user under each test perspective and the emotional characteristics of each test question under each test perspective were extracted. Multiple test datasets were constructed based on the test perspectives. Each test dataset is classified based on multiple test result types corresponding to each test perspective. Each test dataset is divided into multiple data subsets, and each data subset corresponds to a type of test result. Each type of test result corresponds to a scoring reference range. For any subset of data in any test dataset, count the frequency of occurrence of each emotion feature under each test question in the subset of data, select the reference emotion feature under each test question, and obtain the reference emotion set of the subset of data. Multiple reference emotion sets are determined for each test dataset, where each reference emotion set in each test dataset corresponds to a rating reference range. An emotion analysis model is constructed based on multiple reference emotion sets under each test perspective.

[0009] Furthermore, based on the emotion analysis model, the emotion recognition results of the test subjects are processed to generate a second analysis result, including: Based on the tester's initial analysis results, determine the reference emotion set corresponding to the tester from each testing perspective; For any test angle, denoted as the target test angle, the tester's emotional feature vector under the target test angle is determined based on multiple emotional features of the tester under the target test angle and a reference emotional set. Determine the emotional feature vector of the tester under each testing angle, calculate the similarity value between the emotional feature vector under each testing angle and the preset reference feature vector, and obtain the second analysis result.

[0010] Furthermore, reference emotional features were selected for each test question, including: For any test question, all emotional features under the test question are traversed in descending order of frequency of occurrence. For the i-th emotional feature, if the difference between the i-th emotional feature and the (i+1)-th emotional feature is greater than the preset difference threshold, the traversal operation is stopped, and all emotional features with a sorting index not greater than i are selected as reference emotional features. The difference between the i-th emotion feature and the (i+1)-th emotion feature is represented as follows:

[0011] In the formula, Indicates the difference value. This represents the frequency of the i-th emotional characteristic. This represents the frequency of occurrence of the (i+1)th emotional feature.

[0012] Furthermore, the first and second analysis results are fused using a data fusion model to generate the test subject's psychological test analysis results, including: Based on the tester's initial analysis results, determine the similarity reference threshold for the tester at each test angle; If the similarity value of the tester is greater than the corresponding similarity reference threshold in every test angle, the first analysis result will be used as the tester's psychological test analysis result.

[0013] Furthermore, the data fusion model also includes: The data fusion model records the similarity reference thresholds corresponding to each type of test result under each test perspective; The determination of multiple similar reference thresholds for each testing angle includes: For any subset of data in any test dataset, extract the emotional feature vector of each user in the subset based on the reference emotional set of the subset, and calculate the reference value of each user, where the reference value is the similarity value between the user's emotional feature vector and the preset reference feature vector; Based on the reference value of each user in the data subset, a similarity reference threshold is determined for the data subset, resulting in multiple similarity reference thresholds for each test angle. Among these, a preset proportion of users in the data subset have reference values ​​greater than the similarity reference threshold.

[0014] On the other hand, this application provides a psychological monitoring data analysis system based on AI and VR technologies, including: The data acquisition module is used to acquire the test subject's psychological test data, which includes the test subject's question and answer data on the psychological test scale and the test subject's facial feature data during the psychological test. The question-and-answer analysis module is used to analyze the test taker's question-and-answer data on the psychological test scale based on the question-and-answer analysis model and generate the first analysis result; The emotion recognition module is used to analyze and process the facial feature data of the test subject based on the emotion recognition model to obtain the emotion recognition result of the test subject; The emotion analysis module is used to process the emotion recognition results of the tester based on the emotion analysis model and generate a second analysis result; The data fusion module uses a data fusion model to fuse the first and second analysis results to generate the psychological test analysis results for the test taker.

[0015] Furthermore, it also includes: The model building module is used to analyze multiple reference emotion sets under each test perspective in the psychological test scale, and to build an emotion analysis model based on the multiple reference emotion sets under each test perspective.

[0016] Furthermore, the present invention also discloses an intelligent assessment cabin evaluation system, comprising: The cabin as a whole provides a closed environment; The human-computer interaction panel is installed inside the cabin and is used to display test information, test questions and results, and to obtain information input by the tester through the human-computer interaction panel, so as to enable the tester to interact with the system. The testing chair, in which the test subject sits; VR smart glasses, the tester wears VR smart glasses and sits on the testing chair and is provided with a unique virtualized testing environment; The media player is installed on the testing chair. At least one camera is installed inside the cabin, and the camera acquires data on the test subject's facial expressions and body movements. The controller connects to and controls the human-computer interaction panel, VR smart glasses, camera, and player. The AI ​​and VR-based emotion intervention and psychological monitoring data analysis methods described above are embedded in the controller. The AI ​​and VR-based psychological monitoring data analysis system described above is also embedded in the controller.

[0017] This application has the following advantages: The method of this invention, based on AI and VR technology, can acquire and analyze the psychological test data of the test taker from multiple perspectives. By combining the test taker's emotional characteristics during the test, the results of the test taker's question and answer process can be comprehensively analyzed, which can improve the accuracy of the analysis of the test taker's psychological test data. The entire data analysis process has low dependence on human intervention and improves the efficiency of analyzing and processing relevant psychological monitoring data during the psychological test.

[0018] The method and system of this invention, combined with an intelligent assessment booth, uses AI and VR technology to replace manual psychological communication and testing of test subjects, as well as to analyze the psychological monitoring data of test subjects during the assessment process. The specific application of this invention can first achieve the following four stages: In the first stage, participants interact with the AI-powered digital human, allowing them to release their emotions. Through this interaction, their emotions are effectively released and alleviated. During this process, the intelligent assessment cabin system of this invention can play music, release scents, and facilitate interaction to help the participants calm down as much as possible.

[0019] In the second stage, the AI ​​intelligent digital human can ask questions to the test subject and enter the testing phase through the self-learning process of the AI ​​intelligent digital human. That is, the evaluation system constructed by the method of this invention analyzes the voice, micro-expression and body movements during the communication process between the AI ​​intelligent digital human and the venter, and selects corresponding questions from the test scale to achieve early warning screening and assess whether the subject has a mental illness.

[0020] In the third stage, based on the screening results, an assessment report can be generated for patients with mental disorders, and those with mental disorders can be identified. This report can be provided to teachers, schools, parents, etc., either through the back-end system or directly. The report provides information on a person's emotional state and mental health.

[0021] In the fourth stage, this system can provide effective intervention and treatment for people with mental disorders. This intervention and treatment can be carried out in a virtual environment provided by VR devices, which can help patients lower their guard, protect their privacy, and improve the effectiveness of intervention and treatment. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating a data analysis method for emotion intervention and psychological monitoring based on AI and VR technologies, as described in this application.

[0023] Figure 2 This is a schematic diagram of the structure of a psychological monitoring data analysis system based on AI and VR technology in an embodiment of this application.

[0024] Figure 3 This is a schematic diagram of the application of the intelligent assessment booth formed by the assessment system constructed by the method of the present invention in specific practice. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this application clearer, some embodiments of this application will be further described in detail below with reference to the accompanying drawings and examples.

[0026] Example 1, see Figure 1 The present application provides a method for emotion intervention and psychological monitoring data analysis based on AI and VR technologies, which specifically includes: S01. Obtain the psychological test data of the test subject; Specifically, the psychological test data includes the test taker's answers to the psychological test scale and the test taker's facial feature data during the psychological test. The test taker's facial feature data can be recorded using VR devices. The psychological test scale includes multiple test angles, such as perception test, thinking test, language expression test, language comprehension test, anxiety test, and health attention test. Each test angle includes at least one test question. Psychological testing based on the psychological test scale is a technique well known to those skilled in the art and will not be elaborated here.

[0027] It is worth noting that the psychological tests conducted on test subjects in this application can be used in the following scenarios: in a smart assessment booth set up for psychological monitoring of different users, such as students, a chair is provided for the test subject. During the psychological test, the test subject can sit in the chair and wear a VR headset. The VR headset integrates multiple camera modules to collect the test subject's facial feature data. A virtual simulation scene is constructed based on VR technology, and AI technology and AI intelligent digital human interact with the test subject. AI technology replaces human interaction and question-and-answer sessions with the test subject, and the test subject's question-and-answer data on the psychological test scale is recorded.

[0028] S02. Input the test subject's question and answer data on the psychological test scale into the question and answer analysis model to generate the first analysis result; Specifically, the question-and-answer analysis model can be constructed based on psychological test scales. Each usable psychological test scale has undergone relevant testing and meets the standards of psychological testing. A qualified psychological test scale generally includes tests on multiple test angles for the test taker. Each test angle includes a certain number of test questions. Different answers to test questions correspond to different scores. Taking a percentage system as an example, each test angle corresponds to multiple test results, and each test result corresponds to a scoring reference range of 0 to 100 minutes. Therefore, the question-and-answer analysis model can be constructed based on the psychological test scale. The question-and-answer analysis model processes the test taker's question-and-answer data on the psychological test scale, and the first analysis result generated includes the test taker's score under each test angle.

[0029] S03. Analyze and process the facial feature data of the test subject based on the emotion recognition model to obtain the emotion recognition results of the test subject; Specifically, the emotion recognition model built and trained based on deep learning technology can identify the user's emotional characteristics, such as happiness, anger, fear, sadness, and neutrality, by analyzing the user's facial feature data, such as the feature information contained in facial images. This application builds an emotion recognition model based on deep learning technology in artificial intelligence, which is used to identify the emotions of the tester during the question-and-answer process of each test question. The generated emotion recognition results include the emotional characteristics of the tester during the question-and-answer process of each test question.

[0030] S04. Process the emotion recognition results of the tester based on the emotion analysis model to generate a second analysis result; Specifically, the emotion analysis model stores multiple reference emotion sets for each test perspective. These multiple reference emotion sets for each test perspective are obtained by analyzing historical test data from multiple users who have taken psychological tests based on psychological test scales.

[0031] In one exemplary implementation, the emotion recognition results of the test subject are processed based on an emotion analysis model, specifically including: Based on the tester's initial analysis results, determine the reference emotion set corresponding to the tester from each testing perspective; Taking any test angle as an example, the tester's score under that test angle is extracted from the first analysis result. Based on the score, the reference emotion set corresponding to the tester's score is selected from multiple reference emotion sets stored in the emotion analysis model for each test angle. Specifically, each reference emotion set under each test angle corresponds to a score reference range.

[0032] For any test angle, denoted as the target test angle, the tester's emotional feature vector under the target test angle is determined based on multiple emotional features of the tester under the target test angle and a reference emotional set. Specifically, taking any test angle as an example, for ease of description, we denote it as the target test angle. The reference emotion set records the reference emotion features of each test question under the target test angle. The dimension of the emotion feature vector is specifically the number of test questions under the target test angle. For example, the first element of the emotion feature vector corresponds to the first test question under the target test angle. Specifically, for any test question under the target test angle, if the tester's emotion feature under that test question belongs to one of the reference emotion features of that test question under the target test angle, then the value of the element corresponding to that test question in the emotion feature vector is recorded as 1, otherwise it is recorded as 0. The tester's emotion feature vector under the target test angle can be extracted in the aforementioned way.

[0033] Determine the emotional feature vector of the tester under each test angle, calculate the similarity value between the emotional feature vector under each test angle and the preset reference feature vector, and obtain the second analysis result; Specifically, each element in the preset reference feature vector has a value of 1. The similarity value between the emotion feature vector and the preset reference feature vector can be calculated using a similarity calculation formula, such as using the cosine similarity calculation formula. The second analysis result includes the similarity value of the tester under each test angle.

[0034] S05. The first and second analysis results are fused using a data fusion model to generate the psychological test analysis results for the test taker.

[0035] The role of the data fusion model is to determine the reliability of the first analysis result by analyzing the second analysis result, which records the emotional information of the tester.

[0036] In one exemplary implementation, the first analysis result and the second analysis result are fused using a data fusion model to generate the test subject's psychological test analysis result, specifically including: Based on the tester's initial analysis results, determine the similarity reference threshold for the tester at each test angle; The data fusion model records multiple similarity reference thresholds for each test angle. Each similarity reference threshold corresponds to a test result. Based on the tester's first analysis result, the similarity reference threshold corresponding to the tester for each test angle can be determined.

[0037] If the similarity value of the tester is greater than the corresponding similarity reference threshold in every test angle, the first analysis result will be used as the tester's psychological test analysis result.

[0038] It is worth noting that this application uses a similarity reference threshold to measure whether the testee's facial expression features conform to the test context. If the similarity value of the testee under each test angle is greater than the corresponding similarity reference threshold, it indicates that the testee has a relatively serious attitude in answering the questions in the psychological test scale. Therefore, the first analysis result is used as the testee's psychological test analysis result. If the relevant conditions are not met, manual intervention can be selected, such as remote communication between professionals and testees, or manual analysis of the testee's psychological test data by professionals, to further determine the testee's psychological test analysis result.

[0039] It should be added that during the psychological testing process using the smart testing booth, virtual scenes suitable for each testing angle can be constructed based on human experience and VR technology, giving the test taker an immersive experience. Audio playback devices, such as those installed on the chairs, can be set up in the smart testing booth to play soothing music during the psychological test. Smart aromatherapy devices can also be installed to emit pleasant scents during the psychological test. Through various auxiliary methods, the test taker can be integrated into the testing scenario, ensuring that the collected facial feature data closely approximates the test taker's true inner emotions.

[0040] This application provides a data analysis method for emotion intervention and psychological monitoring based on AI and VR technologies. It acquires and analyzes the psychological test data of the test subjects from multiple perspectives using AI and VR technologies, and combines the emotional characteristics of the test subjects during the test to conduct a comprehensive analysis of the results of the test subjects in the question-and-answer process, thereby improving the accuracy of the analysis of the test subjects' psychological test data. The entire data analysis process has low dependence on human intervention and improves the efficiency of analyzing and processing relevant psychological monitoring data during the psychological test process.

[0041] In one alternative implementation, the sentiment analysis model in step S04 is constructed based on the following steps: Historical test data of multiple users who have taken psychological tests based on psychological test scales were obtained. The scores of each user under each test perspective and the emotional characteristics of each test question under each test perspective were extracted. Multiple test datasets were constructed based on the test perspectives. Specifically, historical test data of multiple users who have undergone psychological testing based on psychological test scales can be extracted from data collected in the past based on manual testing of different processes, such as video data and document data. For video data, the emotional characteristics of users can be extracted from it through emotion analysis models. Based on the testing perspective, the historical test data of multiple users can be classified to construct a test data set under each testing perspective.

[0042] Each test dataset is classified based on multiple test result types corresponding to each test perspective. Each test dataset is divided into multiple data subsets, and each data subset corresponds to a type of test result. Specifically, considering that there are multiple test results for each test perspective, a data subset is constructed for each type of test result based on the different test results.

[0043] For any subset of data in any test dataset, count the frequency of occurrence of each emotion feature under each test question in the subset of data, select the reference emotion feature under each test question, and obtain the reference emotion set of the subset of data. Specifically, based on the frequency of occurrence of each emotional feature under each test question, representative emotional features of each test question are selected from multiple emotional features, namely reference emotional features. The number of reference emotional features for each test question is at least one. The reference emotional features corresponding to multiple test questions under the data subset are summarized to construct a set of reference emotional features for each data subset.

[0044] In one exemplary implementation, the selection of reference emotional features for each test question specifically includes: Taking any test question as an example, all emotional features under that test question are traversed in descending order of frequency of occurrence; Specifically, during the traversal process, for the i-th emotion feature, if the difference between the i-th emotion feature and the (i+1)-th emotion feature is greater than the preset difference threshold, the traversal operation is stopped, and all emotion features with a sorting index not greater than i are selected as reference emotion features. In this embodiment, the difference between the i-th emotion feature and the (i+1)-th emotion feature is calculated based on the following formula:

[0045] In the formula, Indicates the difference value. This represents the frequency of the i-th emotional characteristic. This represents the frequency of occurrence of the (i+1)th emotional feature.

[0046] The above methods can be used to selectively obtain reference sentiment features for each test question, thereby determining multiple reference sentiment sets for each test dataset. Specifically, based on the rating reference range to which each data subset belongs, a rating reference range is determined for each reference emotion set in each test data set, and an emotion analysis model is constructed based on multiple reference emotion sets under each test perspective.

[0047] In one exemplary implementation, for the data fusion model in step S05, determining multiple similarity reference thresholds for each test angle specifically includes: For any subset of data in any test dataset, extract the emotional feature vector of each user in the subset based on the reference emotional set of the subset, and calculate the reference value for each user; Specifically, the reference value is the similarity between the user's emotional feature vector and the preset reference feature vector, which can be calculated using a similarity calculation formula, such as the cosine similarity calculation formula.

[0048] Based on the reference value of each user in the data subset, the similarity reference threshold of the data subset is determined, and multiple similarity reference thresholds are obtained for each test angle. Specifically, a similarity reference threshold is determined by a preset ratio. In this embodiment, the preset ratio is 90% as an example. For each data subset, the reference value of users in the data subset is greater than the corresponding similarity reference threshold.

[0049] Example 2, see Figure 2 This application also provides a psychological monitoring data analysis system based on AI and VR technologies, including: The data acquisition module is used to acquire the test subject's psychological test data, which includes the test subject's question and answer data on the psychological test scale and the test subject's facial feature data during the psychological test. Specifically, the test subject's facial feature data can be recorded using VR devices, and the psychological test scale includes multiple testing angles, with each testing angle including at least one test question; The question-and-answer analysis module is used to analyze the test taker's question-and-answer data on the psychological test scale based on the question-and-answer analysis model, and generate the first analysis result, which includes the score for each test angle. The emotion recognition module is used to analyze and process the facial feature data of the tester based on the emotion recognition model to obtain the emotion recognition result of the tester. The emotion recognition model is a deep learning model, and the emotion recognition result includes the emotional features of the tester in the process of answering each test question. The emotion analysis module is used to process the emotion recognition results of the tester based on the emotion analysis model and generate a second analysis result; The data fusion module uses a data fusion model to fuse the first and second analysis results to generate the psychological test analysis results for the test taker. The model building module is used to analyze multiple reference emotion sets under each test perspective in the psychological test scale, and to build an emotion analysis model based on the multiple reference emotion sets under each test perspective.

[0050] Example 3, see Figure 3This embodiment provides an assessment system that uses the emotion intervention and psychological monitoring data analysis method based on AI and VR technology described in Embodiment 1, and the psychological monitoring data analysis system based on AI and VR technology described in Embodiment 2.

[0051] The assessment system described in this embodiment is implemented in conjunction with an intelligent assessment cabin. Specifically, the intelligent assessment cabin includes a relatively complete enclosed cabin 100. Inside the cabin 100, there is a human-computer interaction panel 103. The human-computer interaction panel 103 can display relevant test information, question and answer information, and result information. At the same time, the tester can also input relevant information and interact with the system or the AI ​​intelligent digital human in the system through the human-computer interaction panel 103. The intelligent assessment cabin also includes a testing chair 105. When the tester is allowed to sit on the testing chair 105, a VR smart glasses are also provided. By wearing the VR smart glasses, the tester can complete the test and express emotions in a specific testing environment. Meanwhile, the testing chair 105 is also compatible with a player that plays soothing music 101, allowing the test subject to relax. Furthermore, the testing chair 105 can also release specific fragrances or odors 102, similarly achieving the purpose of relaxing the test subject. In addition, the smart testing cabin is equipped with multiple cameras 104, which can acquire data such as the test subject's facial expressions and body movements from multiple angles. It may also include a wearable watch, which can obtain data on the test subject's arm and arm movement (such as acceleration). Of course, the aforementioned human-computer interaction panel 103, VR smart glasses, cameras 104, player, wearable watch, etc. are all connected to a controller and controlled by the controller. The emotion intervention and psychological monitoring data analysis method based on AI and VR technology described in Embodiment 1 and the psychological monitoring data analysis system based on AI and VR technology described in Embodiment 2 are embedded in the controller.

[0052] Therefore, in this embodiment, a relatively complete enclosed cabin can be built using the Internet and artificial intelligence (AI) technology. The cabin's atmosphere is designed to create a comfortable and relaxing environment through sight, sound, and smell. During testing, AI-powered digital humans, VR virtual environments, and AI psychological mentors are integrated to achieve multiple functions, including efficient communication and privacy protection. This allows test takers to more willingly and completely express their feelings, enabling psychological counselors to more easily and clearly provide correct and effective prevention or intervention methods based on the test taker's condition. Ultimately, this achieves the goals of early detection, early prevention, early intervention, and disease prevention, thereby improving the overall health of the population.

[0053] The evaluation system provided in this embodiment, combined with the intelligent assessment booth, is implemented mainly in the following four stages: In the first stage, the test takers enter the intelligent testing booth and interact with the AI ​​intelligent digital human, either standing or sitting in the testing chair. This process mainly allows the test takers to vent their emotions through communication. By interacting with the AI ​​intelligent digital human, their emotions can be effectively released and alleviated. During this process, the interaction with the AI ​​intelligent digital human can be achieved through the human-computer interaction panel or through the AI ​​intelligent digital human in the virtual environment provided by VR smart glasses.

[0054] After the test taker has vented their emotions, the system can play soothing music, allow the test taker to choose their favorite scent (such as jasmine), and release it to help the test taker relax and alleviate their emotions as much as possible, so as to facilitate the subsequent testing phase.

[0055] In the second stage, through the self-learning process of the AI ​​intelligent digital human, combined with the communication and interaction content with the test subject in the first stage, further questions can be asked to the test subject and enter the testing stage. In this process, through the analysis of the voice content, facial micro-expression changes and body movements during the communication between the AI ​​intelligent digital human and the venter, corresponding questions are selected from the test scale for targeted assessment, early warning screening, and evaluation of whether the subject has a mental illness.

[0056] In the third stage, the screening results can be combined to generate an assessment report for patients with mental disorders and identify those who actually have mental disorders. This report can be provided to teachers, schools, parents, etc. through the back-end system or directly. The report provides information on a person's emotional state and mental health.

[0057] In the fourth stage, this system can provide effective intervention and treatment for people with mental disorders. This intervention and treatment can be carried out in a virtual environment provided by VR devices, which can help patients lower their guard, protect their privacy, and improve the effectiveness of intervention and treatment.

[0058] The following examples (not all are provided) illustrate the specific test scenarios, test table contents, and data involved in the method and system of this invention in specific application environments.

[0059] (I) In the specific implementation of the above embodiments, one of the test scenarios may be: Sequence Test: The screen displays a matrix with the same dimensions (e.g., 2*2, 4*4, 5*5, etc.). Each element of the matrix consists of an outer outline and an internal number. The numbers within a matrix are not repeated. The test subject must click all the elements as quickly as possible in a certain order (from smallest to largest or from largest to smallest). During the test, the system collects four indicators: screen pressure at each click (data 1), acceleration data of the wearable device (data 2), completion time (data 3), and number of correct clicks (data 4).

[0060] Target Tracking: The screen displays multiple elements with identical shapes but different positions. Before the test begins, the test subject is prompted that one or more of these elements are the target elements to be tracked. After the test begins, all elements start moving according to their pre-set trajectories. After all elements have finished moving, the test subject is required to use the finger on the side wearing the wearable module to click on the elements to indicate which of the initially designated target elements are correct. The test collects the time required for the test subject to point out the elements (data 5), ​​the number of correct points (data 6), the click pressure (data 7), and the acceleration data of the wearable module (data 8).

[0061] Grasshopper Catching: Images of grass and grasshoppers were displayed on the screen. During the test, the test subject needed to slowly approach and hold the grasshopper with the hand wearing the wearable module. If the approach was too fast, the grasshopper would fly away. The test subject was required to catch as many grasshoppers as possible within a fixed time (e.g., 5 minutes). The test collected the number of grasshoppers finally caught (data 9) and the number of failed attempts (data 10).

[0062] Water Feeding Game: The screen displays multiple water pipes from different directions, with only one pipe supplying water at a time. Test subjects were instructed to control the input side of the faucet by rotating their wrist (the side wearing the wearable module), thus continuously supplying water to a drinking animal. A successful input required a relatively fast rotation followed by a slower return to center. The test results were ultimately quantified by the time the water supply stopped (Data 11) and the number of invalid wrist movements (Data 12).

[0063] Limb Response: The screen displays an image of a child, with one of the child's limbs highlighted. There are two highlighting methods: one indicates the test subject needs to move the corresponding limb, and the other indicates the test subject should not move that limb. During the test, the test subject needs to stand in front of the main unit's camera, allowing the camera to capture a full-body image. The image data is input into a general human motion recognition model, which determines whether the test subject has performed the correct action. The test collects the number of correct limb movements (data 13), the number of correct non-limb movements (data 14), and the number of consecutive correct responses (data 15).

[0064] Reading Aloud: A text or pinyin passage will be displayed on the screen, and test takers are required to read it aloud correctly. The test taker's reading voice will be captured through a microphone, and the sound signal will be input into a general Chinese speech recognition model to determine whether the test taker has read the content correctly. The reading material is content that participants of the participant's age can understand. For school students, the reading material is designed to be similar in difficulty to the textbooks for their grade level. The test will use the number of errors (data 16) as a quantitative indicator of performance.

[0065] Random Spatial Order Test: 5-15 circles of the same size, randomly positioned and non-overlapping, will be presented on the screen. Before the test begins, several circles will be highlighted on the screen in a specific order, and the test taker will be asked to try to remember this order and select them in that order in subsequent steps. The length of the test sequence increases after each correct answer given by the test taker, and the test stops when the test taker gives an incorrect answer. The maximum length reached by the test taker (Data 17) is used as the quantitative indicator of performance.

[0066] Shape-Color Conflict Test: The screen will display holes in basic shapes (e.g., circles, squares, triangles, etc.) of several colors (e.g., yellow, blue, white, black, red). The test will continuously present participants with a block of a specific color and shape, and participants will be asked to quickly match the given block to the correct hole according to either color or shape. The test's performance will be quantified by the number of completed blocks (Data 18) and the number of errors (Data 19) within a fixed time.

[0067] Bug-catching test: A speaker randomly plays several monosyllabic animal names, including the word "bug" (e.g., dog, chicken, rabbit, insect). An apple is displayed on the screen. A bug may or may not appear on the apple when the speaker plays the animal name. Test subjects are required to touch the screen only when a bug is displayed on the screen and the speaker plays the pronunciation of "bug." They should not touch the screen if this condition is not met. The final performance quantification is based on the percentage of correct touches (data 20), incorrect touches (data 21), correct no-touches (data 22), and incorrect no-touches (data 23) within a certain time period.

[0068] Maintaining Balance: A ball is displayed on the screen and slides in the direction the screen is tilted, the degree of which is detected by a built-in gyroscope and accelerometer. Test subjects are instructed to hold the main unit horizontally to ensure the ball remains as centered on the screen as possible while continuing to move forward. The test subject's forward movement is monitored by the built-in accelerometer. The test results are ultimately quantified by the total time the test subject maintains balance (Data 24) and the average distance the ball travels from the center of the screen during the walking period (Data 25).

[0069] After obtaining the data from data 1 to data 25, the following scoring criteria can be used: First, calculate the scores for each of the 10 tests, and then categorize them according to their characteristics into the attention, hyperactivity, and impulsivity problem dimensions. Sequential test: Result 1: (Data 4 / Data 3) * (-1), Result 2 (Impulsiveness): Standard deviation of Data 1, Result 3 (Hyperactivity): Standard deviation of Data 2; Target tracking: Result 4: (Data 6 / Data 5) * (-1), Result 5 (impulsiveness): standard deviation of Data 7, Result 6 (hyperactivity): standard deviation of Data 8; Catching grasshoppers: Result 7: Data 9 * (-1), Result 8 (Impulsive): Data 10; Water feeding game: Result 9: Data 11, Result 10 (hyperactivity, impulsivity): Data 12; Limb responses: Result 11: Data 13 * (-1), Result 12 (Attention, Impulsivity): Data 14 * (-1), Result 13: Data 15 * (-1) Read aloud: Result 14: Data 16; Random Spatial Order Test: Results 15: Data 17 * (-1); Shape-Color Conflict Test: Result 16: Data 18 * (-1), Result 16 (Attention, Impulsivity): Data 19; Bug-catching test: Result 18: Data 20 * (-1) + Data 23; Result 19 (Impulse): Data 21 + Data 22 * ​​(-1); Maintaining balance: Result 20 (hyperactivity): Data 24 * (-1), Result 21: Data 25.

[0070] Scoring Preset Data Preparation: Data was collected from normal test subjects aged 6, 7, 8...13, 14, 15, with 60 (or more) participants in each age group and 60 (or more) participants aged 15 and above. The mean and standard deviation of each result for each age group were calculated as the norms for subsequent scoring.

[0071] Scoring implementation steps: (1) Obtain the test subject's results 1-21; (2) Find the mean and standard deviation of each result that matches their age group, and use this data to convert the score into a standard score (i.e., subtract the mean and divide by the standard deviation); (3) Calculate the mean of the results belonging to the same dimension.

[0072] Results evaluation method: When the mean of a certain dimension of a test subject is between [-∞, 1.5), the test subject is considered to have no obvious problems in that dimension; when the mean of a certain dimension of a test subject is between [1.5-2.0), the test subject is considered to have a relatively likely problem in that dimension; when the mean of a certain dimension of a test subject is between [2.0, ∞), the test subject is considered to have a very likely problem.

[0073] (II) In the specific implementation of the above embodiments, one of the test scenarios could be: when conducting a potential self-harm analysis and assessment test on middle school students, the following assessment form can be used: 1. Your family encourages you to exercise. 2. Dinner is after 8 pm. 3. Sleep duration meets my needs. 4. The school offers psychological counseling services. 5. Your parents answered your questions very seriously. 6. You envy other people's parents. 7. You can feel your family's love for you. 8. Your parents taught you the importance of self-respect and chastity. 9. Your parents ask for your opinion when buying things for you. 10. Your parents praise and compliment you. 11. Your parents support your hobbies. 12. Your parents taught you to learn to forgive others. 13. Your parents or teachers have told you about the physical changes you should experience during puberty. 14. Playing in unsafe places 15. You don't want to go to school. 16. Your parents believe you should have your own privacy. 17. You chat with your family 18. Your parents answered your questions very seriously. 19. Your parents are very concerned about your growth and development (such as height and weight). 20. Family members or teachers remind you not to play in unsafe places. 21. Be instructed on how to seek help and save yourself in the event of a fire. 22. Your parents respect your opinions. 23. Your family reminds you to protect your eyesight. 24. You have your own personal washing utensils, such as a washbasin and towel. 25. Your family will comfort you when you are sad. 26. Drink milk every day 27. Your parents know where you usually go to play. 28. Wash hands before meals and after using the toilet. 29. Your parents can listen to you patiently. 30. When your request is refused, your parents will explain the reasons and gain your understanding. 31. When you lack confidence, you will receive support and encouragement from your family. 32. The school provides computer labs with internet access. 33. The school conducts health, safety, and hygiene education activities every semester. 34. Your parents' expectations put a lot of pressure on you. 35. The pressure of studying is immense, and I have nowhere to confide in. 36. When your parents break a promise, they will apologize to you. Urban primary school students are being neglected (examples provided) 1. Go to a street stall to buy something to eat. 2. Able to eat fruits and vegetables every day 3. Eating fast food (such as KFC) or drinking carbonated beverages (such as Coca-Cola) 4. Get your vaccines on time. 5. I told you that you need to get your family's permission before going anywhere else. 6. The school offers psychological counseling services. 7. Your parents answered your questions very seriously. 8. Teacher scolds student 9. Your parents take their anger out on you when they are in a bad mood. 10. I was told not to eat or drink food or beverages from strangers. 11. You envy other people's parents. 12. I told you that no one should see or touch the parts of your body covered by your swimsuit. 13. You keep your unhappiness to yourself. 14. You can feel your family's love for you. 15. You're lying at home watching TV. 16. You go to an internet cafe 17. Your parents ask for your opinion when buying things for you. 18. Your parents praise and compliment you. 19. Your parents or teachers remind you to pay attention to traffic safety, electrical safety, fire safety, and gas safety, etc. 20. If you don't do well on your exams, your parents will get angry with you. 21. Your parents support your hobbies. 22. Eating barbecued, fried, or pickled foods 23. Your parents or teachers tell you not to meet with strangers (such as online friends). 24. Your parents taught you to help others. 25. Your parents or teachers have told you that your body will change as you grow up. 26. Playing in unsafe places 27. You don't want to go to school. 28. Replacing lunch or dinner with instant noodles. 29. Chat or play with family. 30. I've been told not to open the door to strangers when I'm home alone. 31. Your home study environment is very noisy. 32. You have medical and / or accident insurance. 33. Your parents answered your question very seriously. 34. Your parents are very concerned about your growth and development (such as height and weight). 35. Someone is playing mahjong at home. 36. Family members or teachers remind you not to play in unsafe places. 37. Be instructed on how to seek help and save yourself in the event of a fire. 38. Your parents force you to learn musical instruments, painting, or other things you don't like. 39. Your family reminds you to protect your eyesight. 40. Your family will comfort you when you are sad. 41. Your parents know where you usually go to play. 42. Wash hands before meals and after using the toilet. 43. Your parents can promptly detect changes in your emotions, such as unhappiness, tension, or fear. 44. Your parents have said things like, "It would have been so much better if I hadn't been born to you." 45. Your parents listen to you very attentively. 46. ​​Your parents believe that as long as you study well, nothing else matters. 47. Difficulty communicating with parents 48. Your parents hit you or point out your shortcomings in front of others. 49. When your request is refused, your parents will explain the reasons and you will get your way. 50. When you lack confidence, you will receive support and encouragement from your family. 51. When parents fail to keep their promises, they will proactively apologize to you. 52. The school conducts health, safety, and hygiene education activities every semester. 53. When family members or teachers are angry, they throw your things. 54. You don't want to go home. 55. Your family or teachers have told you things like, "You won't amount to anything when you grow up." 56. The pressure of studying is immense, and I have nowhere to confide in. 57. Your parents got angry because you didn't do well on the test. Rural primary school students are being neglected (examples provided) 1. Your parents send you blessings and greetings on holidays or your birthday. 2. You can eat animal-based foods such as meat, fish, or eggs. 3. The school has extracurricular books and periodicals available for borrowing. 4. Your parents take their bad mood out on you. 5. Village clinics can provide the healthcare services you need (physical examinations, doctor's visits, vaccinations). 6. When your parents buy clothes for you, they should consider the material and quality of the clothing. 7. When you get poor grades, you are verbally abused or physically punished by your teachers or parents. 8. The school has computer labs, language labs, and other facilities available for student use. 9. Family members keep their word and don't lie to you. 10. Your family members are paying attention to your growth and development (e.g., measuring your height and weight). 11. When you encounter relationship problems, your teachers and parents will promptly identify and actively guide you. 12. The school holds a sports meet every year. 13. Do you know some commonly used emergency numbers (110, 119, etc.)? 14. Your family will allow you to do what you are capable of doing yourself. 15. Schools offer sex education or physiological hygiene courses. 16. You make friends online and meet them in person. 17. Your parents told you to carefully read the instructions before using any medication. 18. Even if you don't do something well, your family will encourage you. 19. When you lack confidence, your family will provide you with support and encouragement. 20. Your parents will soak and wash the new underwear they buy for you before you wear it. 21. Your family taught you to be strong and never give up easily. 22. The village has activity facilities and venues. 23. Your parents will keep track of your school performance from your teacher. 24. Your family thinks you must listen to them in everything. 25. Your parents require that your grades be better than other students'. 26. Your parents will teach you basic hygiene and health knowledge. 27. When you buy food, do you check the production date, shelf life, and ingredients? 28. When you make a mistake, your family will reason with you. 29. Your family will lie to you and won't keep their word. 30. The school conducts health, safety, and hygiene education activities. 31. Your parents are equally concerned about your physical and mental health. 32. Your instructors change frequently. 33. Your family knows the reasons for your happiness or sadness. 34. The desks and chairs in the classroom will be adjusted to your height as you grow. 35. Your parents believe that as long as you study well, nothing else matters. Child Abuse Scale (CAS) (Example) 1. I get beaten for the smallest mistake. 2. They hit me for no reason. 3. He scolded me in front of others. 4. They criticize and ridicule me when they are dissatisfied with my words or actions. 5. When I am sad or angry, my family members make sarcastic or mocking remarks to me. 6. My family members said some harsh or insulting things to me. It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims. Parts not described in detail in this specification are prior art known to those skilled in the art.

Claims

1. A data analysis method for emotion intervention and psychological monitoring based on AI and VR technologies, characterized in that, include: The test subjects' psychological test data are obtained. The psychological test data includes the test subjects' question and answer data on the psychological test scale and the test subjects' facial feature data during the psychological test. The test subjects' facial feature data is recorded based on VR devices. The psychological test scale includes multiple test angles, and each test angle includes at least one test question. The test taker's question-and-answer data on the psychological test scale is input into the question-and-answer analysis model to generate the first analysis result, which includes the score for each test angle. The facial feature data of the tester is analyzed and processed based on the emotion recognition model to obtain the emotion recognition result of the tester. The emotion recognition model is a deep learning model, and the emotion recognition result includes the emotional features of the tester in the process of answering each test question. The emotion recognition results of the test takers are processed based on the emotion analysis model to generate a second analysis result. In this step, the construction of the emotion analysis model includes: acquiring historical test data from multiple users who have taken psychological tests based on psychological test scales; extracting the scores of each user under each test perspective, and the emotional characteristics during the question-and-answer process of each test question under each test perspective; constructing multiple test datasets based on the test perspectives; classifying each test dataset based on the multiple test result types corresponding to each test perspective, dividing each test dataset into multiple data subsets, each data subset corresponding to a type of test result; where each type of test result corresponds to a scoring reference range; for any data subset in any test dataset, the scores of each test result in the statistical subset are analyzed. The frequency of occurrence of each emotional feature under the test question is used to select reference emotional features for each test question, resulting in a reference emotional set for the data subset. Multiple reference emotional sets are determined for each test data set, where each reference emotional set in each test data set corresponds to a scoring reference range. An emotional analysis model is constructed based on multiple reference emotional sets for each test perspective. In this step, the selection of reference emotional features for each test question includes: for any test question, all emotional features under the test question are traversed in descending order of frequency of occurrence. For the i-th emotional feature, if the difference between the i-th emotional feature and the (i+1)-th emotional feature is greater than a preset difference threshold, the traversal operation is stopped, and all emotional features with a sorting index not greater than i are selected as reference emotional features. The difference between the i-th emotion feature and the (i+1)-th emotion feature is represented as follows: ; In the formula, Indicates the difference value. This represents the frequency of the i-th emotional characteristic. This represents the frequency of occurrence of the (i+1)th emotional feature; The first and second analysis results are fused using a data fusion model to generate the test subject's psychological test analysis result, including: determining the similarity reference threshold for the test subject at each testing angle based on the first analysis result; if the similarity value of the test subject at each testing angle is greater than the corresponding similarity reference threshold, the first analysis result is used as the test subject's psychological test analysis result; and processing the test subject's emotion recognition result based on an emotion analysis model to generate the second analysis result, including: Based on the tester's initial analysis results, determine the reference emotion set corresponding to the tester from each testing perspective; For any test angle, denoted as the target test angle, the tester's emotional feature vector under the target test angle is determined based on multiple emotional features of the tester under the target test angle and a reference emotional set. Determine the emotional feature vector of the tester under each testing angle, calculate the similarity value between the emotional feature vector under each testing angle and the preset reference feature vector, and obtain the second analysis result.

2. The data analysis method for emotion intervention and psychological monitoring based on AI and VR technology as described in claim 1, characterized in that, For data fusion models, it also includes: The data fusion model records the similarity reference thresholds corresponding to each type of test result under each test perspective; The determination of multiple similar reference thresholds for each testing angle includes: For any subset of data in any test dataset, extract the emotional feature vector of each user in the subset based on the reference emotional set of the subset, and calculate the reference value of each user, where the reference value is the similarity value between the user's emotional feature vector and the preset reference feature vector; Based on the reference value of each user in the data subset, a similarity reference threshold is determined for the data subset, resulting in multiple similarity reference thresholds for each test angle. Among these, a preset proportion of users in the data subset have reference values ​​greater than the similarity reference threshold.

3. A psychological monitoring data analysis system based on AI and VR technology, using the emotion intervention and psychological monitoring data analysis method based on AI and VR technology as described in claim 1 or 2, characterized in that, include: The data acquisition module is used to acquire the test subject's psychological test data, which includes the test subject's question and answer data on the psychological test scale and the test subject's facial feature data during the psychological test. The question-and-answer analysis module is used to analyze the test taker's question-and-answer data on the psychological test scale based on the question-and-answer analysis model and generate the first analysis result; The emotion recognition module is used to analyze and process the facial feature data of the test subject based on the emotion recognition model to obtain the emotion recognition result of the test subject; The emotion analysis module is used to process the emotion recognition results of the tester based on the emotion analysis model and generate a second analysis result; The data fusion module uses a data fusion model to fuse the first and second analysis results to generate the psychological test analysis results for the test taker.

4. The psychological monitoring data analysis system based on AI and VR technology as described in claim 3, characterized in that, Also includes: The model building module is used to analyze multiple reference emotion sets under each test perspective in the psychological test scale, and to build an emotion analysis model based on the multiple reference emotion sets under each test perspective.

5. An intelligent assessment cabin evaluation system, characterized in that, include: The cabin as a whole provides a closed environment; The human-computer interaction panel is installed inside the cabin and is used to display test information, test questions and results, and to obtain information input by the tester through the human-computer interaction panel, so as to enable the tester to interact with the system. The testing chair, in which the test subject sits; VR smart glasses, the tester wears VR smart glasses and sits on the testing chair and is provided with a unique virtualized testing environment; The media player is installed on the testing chair. At least one camera is installed inside the cabin, and the camera acquires data on the test subject's facial expressions and body movements. The controller, the human-computer interaction panel, VR smart glasses, camera and player are all connected to and controlled by the controller, and the emotion intervention and psychological monitoring data analysis method based on AI and VR technology as described in claim 1 or 2 is embedded in the controller; The psychological monitoring data analysis system based on AI and VR technology as described in claim 3 or claim 4 is embedded in the controller.

Images