Method and system for determining psychological relieving scheme, equipment and medium

Abstract

The invention discloses a psychological relieving scheme determination method, system, equipment and medium, and relates to the technical field of psychological analysis, and the method comprises the steps: obtaining psychological health assessment data, collecting voice and video information of a user, and carrying out the text and video combined processing of the voice and video information, thereby obtaining a psychological relieving scheme; and analyzing a change relationship among the facial expressions, actions and emotion features, combining the change relationship with the psychological health assessment data, analyzing to obtain psychological stress information of the user, and determining a psychological soothing scheme based on the psychological stress information. According to the method, the pre-filled psychological health evaluation data is combined with the voice text collected in real time, so that earlier-stage historical data acquisition and user state acquisition in evaluation are realized, psychological pressure information of the user can be more accurately shown, and a psychological relieving scheme can be better selected.

Description

Technical Field

[0001] The present invention relates to the technical field of psychological analysis, and in particular to a method, system, device and medium for determining a psychological relief program. Background Art

[0002] Mental health is a common concern in modern society. Many people feel exhausted and helpless due to abnormal mental health, especially in a busy pace of life and stressful environment. The traditional psychological counseling model usually requires users to go to the hospital in person or communicate with psychologists face to face, which will cause inconvenience and obstacles for some people with social disorders or who are inconvenient to go to the hospital.

[0003] With the rapid development of artificial intelligence technology, its application in the field of mental health has received increasing attention. Existing artificial intelligence technology has made certain progress in the field of mental health.

[0004] When generating personalized catharsis recommendation plans, the existing technology directly uploads the extracted user's physical signs and expression information, selects a predetermined score option through the directly uploaded single information, and then obtains the recommendation plan through the score. However, the score judgment of a single information can only obtain the user's partial psychological emotions, and the result obtained is relatively one-sided, resulting in the final psychological relief plan being inaccurate, reducing the relief effect. Summary of the invention

[0005] The purpose of the present invention is to address the deficiencies of the above-mentioned prior art and provide a method, system, device and medium for determining a psychological relief plan, so as to solve the problem that the recommended plans in the prior art are limited fixed plans with contents prepared in advance, the information processing method generated during the user's psychological relief is single, and the single information acquisition can only obtain partial information about the user's psychology, the relief method is not good, and the psychological relief results obtained are not effective.

[0006] The present invention specifically provides the following technical solutions:

[0007] A method for determining a psychological relief program comprises the following steps:

[0008] Obtain the user's mental health assessment data, and collect the user's voice and video information during the test;

[0009] Convert the voice information into text, obtain the user's emotional characteristics by performing natural language processing on the text, extract the user's facial features from the video information, and classify the emotions based on the facial features;

[0010] The emotion classification results and emotional features are correlated and analyzed to obtain the changing relationship between facial features and emotional features;

[0011] The source of the user's psychological stress is determined through the psychological health assessment data and the change relationship, and a psychological relief plan is determined based on the source of the psychological stress.

[0012] Preferably, the obtaining of the user's mental health assessment data is specifically as follows:

[0013] Different scales are set according to the user's age, social background and corresponding psychological condition; the social background information includes the user's educational level and family members, and the scale includes text, pictures and video elements;

[0014] Users input information into the corresponding scale to generate mental health assessment data.

[0015] Preferably, before classifying emotions by facial features, the following steps are also included:

[0016] Establishing a training data set for a convolutional neural network model; the training data set includes sample data of a specific population representing different psychological states, each sample data including feature data representing the psychological state and a corresponding category label;

[0017] The training data set is input into the convolutional neural network model, and the connection weights and errors of the convolutional neural network model are continuously adjusted until the training error of the convolutional neural network model reaches a predetermined threshold.

[0018] Preferably, a psychological relief plan is determined based on the source of psychological stress, including:

[0019] Determine psychological stress information based on the source of psychological stress and process the psychological stress information, specifically:

[0020] Define a virtual environment M. Users can relieve their psychological stress by interacting with the virtual scene. The state s of the environment M contains the user's emotional state and interaction history information.

[0021] Obtain the reward function R(s,a) based on user feedback and emotional status; specifically:

[0022]

[0023] Construct a reinforcement learning model (S, A, P, π, R), and obtain a psychological relief plan by training the reinforcement learning model (S, A, P, π, R); specifically:

[0024] According to the optimal strategy π* learned by the reinforcement learning model, personalized scenes, plots and task difficulties are generated to adapt to the user's psychological needs and emotional state; the specific expression of the optimal strategy π* is:

[0025]

[0026] Among them, s∈S, S is the state set, action a∈A, A is the action set, P is the transition probability, π is the strategy set, R is the reward, r + is a positive reward, r - is a negative reward, G t To maximize the cumulative reward, E is the relevant function.

[0027] Preferably, after learning the optimal strategy π* according to the reinforcement learning model, the following steps are also included:

[0028] According to the user's immediate feedback (s t ,r t ) Real-time dynamic adjustment of virtual scenes and interactive plots, the specific expression is:

[0029]

[0030] Among them, s t is the state at time t, r t is the reward at time t, π t is the strategy at time t, π t+1 is the strategy at time t+1.

[0031] Preferably, the psychological relief method includes using holographic projection technology to enable users to experience personalized scenes, music, animal and plant elements and interactive games in the real world, and interacting with users during the implementation of the solution, adjusting the scenes and the difficulty of interactive games according to the users' emotional state.

[0032] The present invention provides a system for determining a psychological relief program, comprising:

[0033] The collection module is used to obtain the user's mental health assessment data and collect the user's voice and video information during the test;

[0034] A processing module, used to convert the voice information into text, obtain the user's emotional characteristics by performing natural language processing on the text, extract the user's facial features from the video information, and classify emotions based on the facial features;

[0035] The joint analysis module is used to perform correlation feature analysis on the emotion classification results and the emotional features to obtain the changing relationship between facial features and emotional features;

[0036] The stress processing module is used to determine the source of the user's psychological stress through the psychological health assessment data and the change relationship, and to determine the psychological relief plan based on the source of the psychological stress.

[0037] The present invention provides a computer device, comprising a memory and a processor, wherein a program is stored in the memory, and when the program is executed by the processor, the processor executes the steps of the method for determining a psychological relief program.

[0038] The present invention provides a storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the method for determining a psychological relief program are implemented.

[0039] Compared with the prior art, the present invention has the following significant advantages:

[0040] The method of the present invention obtains psychological health assessment data and collects voice and video information of the user, and performs text and video joint processing on the voice and video information to achieve the acquisition of the user's early psychological health assessment data, and obtains the user's real-time emotional characteristics and emotion classification through voice and video analysis, thereby obtaining more detailed psychological change information, and performing psychological stress analysis based on this. The psychological change information is used to refine and supplement the pre-filled psychological health assessment data, which can more accurately show the user's psychological stress information, thereby better selecting a psychological relief plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 is a schematic diagram of a process of psychological relief for a user by a psychological relief system provided by an embodiment of the present invention;

[0042] Figure 2 The present invention is a flowchart of a method for determining a psychological relief program. DETAILED DESCRIPTION

[0043] The following is a clear and complete description of the technical solutions of the embodiments of the present invention in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0044] like Figure 2 As shown, the present invention provides a method for determining a psychological relief program, which specifically includes the following steps:

[0045] Step S1: Obtain the user's mental health assessment data and collect the user's voice and video information during the test.

[0046] Data Collection:

[0047] The system first collects the user's basic information through the login module, and the user further completes the mental health scale through the scale test module. The scale provides personalized test content based on the user's age, social background and psychological condition, such as questions in the form of text, pictures, videos, etc., that is, different scales are set according to the user's age stage, social background and corresponding psychological condition. Among them, social background information includes the user's educational level and family members.

[0048] Users input information into the corresponding scale to generate mental health assessment data.

[0049] Preliminary evaluation:

[0050] After the scale test is completed, the system will generate preliminary mental health assessment results based on the user's filled content and provide corresponding suggestions. This preliminary result is processed and generated by the scale test module and is mainly used to identify the user's current mental state, such as whether there is anxiety, depression, etc.

[0051] The specific evaluation process is as follows: according to the options selected by the user in each test question, the scores of each question are counted separately, and the scores of all test questions are accumulated to obtain the total score of the test scale. By comparing the total score with the preset mental health status comparison table in the system, the user's preliminary mental health assessment results are obtained. For example, a score of 0-30 points indicates normal mental health; a score of 31-60 points indicates possible anxiety; a score of 61-90 points indicates possible mild depression.

[0052] Step S2: Convert the voice information into text, obtain the user's emotional characteristics by performing natural language processing on the text, extract the user's facial features from the video information (through computer vision technology), classify emotions based on the facial features, and obtain emotion classification results.

[0053] Natural language processing (NLP) and computer vision technology use the behavioral analysis unit to analyze the user's immediate emotional fluctuations, facial micro-expressions and movement changes, and combine the preliminary evaluation results in the scale test module to conduct a real-time assessment of the user's mental health status.

[0054] The specific steps of the real-time evaluation process are as follows:

[0055] 1. Data collection: The basic information collected from the user is recorded as Q1, the preliminary mental health assessment results of the user are recorded as Q2, and the conversation and interaction data with the user in the voice consultation submodule and the video consultation submodule are recorded as Q3.

[0056] 2. Data preprocessing: Preprocess and clean the collected data to ensure the accuracy and consistency of the data.

[0057] 3. Feature extraction: The behavior analysis unit in the AI ​​robot module converts the voice information in the data Q3 into text, and uses natural language processing (NLP) to perform sentiment analysis on the text information to obtain the user's current emotional feature T1. At the same time, the convolutional neural network (CNN) is used to extract the user's facial features, such as timbre, eyes, corners of the mouth, etc. from the video information in Q3. Further, these facial features and as input parameters are input into the trained convolutional neural network model for emotion classification. The classification result is recorded as T2. It should be noted that T1 and T2 here are stage-specific feature sets. For example, when a user communicates about the topic of the i-th specific topic or answers the i-th specific question, it is recorded as T1i and T2i, so as to more accurately analyze the user's true emotions facing a specific topic.

[0058] S2 is specifically:

[0059] Convert speech into text, perform semantic understanding and intent recognition on the text, and identify user emotions using behavioral analysis.

[0060] The facial expressions and body movements in the video are analyzed and extracted through computer vision technology, and the relationship between facial expressions, body movements and emotional fluctuations is obtained by combining the recognized emotion evaluation.

[0061] Before the facial features are fed into the convolutional neural network for emotion classification, it also includes:

[0062] A training data set for a convolutional neural network model is established; the training data set contains sample data of a specific population representing different psychological states, and each sample data includes feature data representing the psychological state and a corresponding category label.

[0063] The training data set is input into the convolutional neural network model, and the connection weights and errors of the convolutional neural network model are continuously adjusted until the training error of the convolutional neural network model reaches a predetermined threshold.

[0064] Step S3: Perform correlation feature analysis on the emotion classification results and the emotional features to obtain the changing relationship between facial features and emotional features.

[0065] Step S4: Determine the source of the user's psychological stress through the psychological health assessment data and the change relationship, and determine the psychological relief plan based on the source of psychological stress.

[0066] Determine psychological relief plans based on the source of psychological stress, including:

[0067] Psychological stress information is determined based on the source of psychological stress, and the psychological stress information is processed using a reinforcement learning algorithm, specifically:

[0068] Define a virtual environment M, where users can relieve psychological stress by interacting with virtual scenes. The state s∈S of the environment M contains the user's emotional state and interaction history information. M includes interactive game characters, scenes, tasks, and interaction methods.

[0069] Based on user feedback and emotional status, we obtain the reward function R(s,a) to evaluate the performance of the AI ​​agent in the game; specifically:

[0070]

[0071] The AI ​​agent refers to the part of the system that is responsible for playing the virtual character, making decisions, and interacting with the user. This AI agent is an intelligent agent based on a reinforcement learning algorithm. It acts as a virtual character in the game environment, makes decisions and performs corresponding actions based on the current state representation and the learned strategy.

[0072] Among them, in order to determine the user's emotional state in real time, the AI ​​processing sub-module obtains the user's psychological state and emotional feedback through questions and answers, dialogues, and capturing user movements and micro-expressions. Through sentiment analysis technology, it analyzes the user's tone, speaking speed, and word usage characteristics in the conversation, and captures the user's facial expressions and body language through the camera to determine whether the user is currently feeling happy and relaxed or anxious and uncomfortable.

[0073] Construct a reinforcement learning model (S, A, P, π, R), and obtain a psychological relief plan by training the reinforcement learning model (S, A, P, π, R).

[0074] Generate personalized scenarios, plots, and task difficulties based on the optimal strategy π* learned by the reinforcement learning model to adapt to the user's psychological needs and emotional state;

[0075]

[0076] Among them, s∈S, S is a state set. In the stress relief module, state s∈S can include information such as the user's emotional state (such as pleasure, relaxation, anxiety, discomfort), the current game plot, and the user's interaction history. For example, state s can represent the user's overall psychological state and the virtual environment at a specific moment. Action a∈A, A is an action set. In the stress relief module, action a∈A can be various interactive behaviors performed by the system on the user in the virtual environment, such as adjusting the difficulty of the interactive game, changing the scene, playing different music, proposing new tasks or challenges, etc. Each action a corresponds to a specific decision made by the AI ​​agent in a certain state s. P is the transition probability, which describes the probability that the system transitions from one state s∈S to another state s′∈S through a certain action a∈A. Specifically, P(s′|s,a) represents the probability of transitioning to state s′ after taking action a in state s. In stress relief, the state transition probability can describe the changes in the user's emotional state in different interactive situations. π is a set of strategies, which defines the probability distribution of the system selecting action a∈A in each state s∈S. Specifically, π(a|s) represents the probability of selecting action a in state s. In stress relief, strategy π aims to select the optimal interaction method and content to maximize the user's soothing effect. R is the reward, which defines the immediate reward obtained by the system for taking action a∈A in each state s∈S. Specifically, R(s,a) represents the reward obtained for taking action a in state s. In the stress relief module, the reward function R is defined based on the user's feedback and emotional state. For example, if the user feels happy and relaxed in the game, R(s,a) will give a positive reward r+; if the user feels anxious or uncomfortable, a negative reward r- will be given. In order to judge the user's emotional state, the system can obtain the user's psychological state and emotional feedback through questions and answers, dialogues, capturing user actions and micro-expressions, etc., r + is a positive reward, r - is a negative reward. t To maximize the cumulative reward, E is the relevant function. The goal is to maximize the cumulative reward through strategy π G t represents the cumulative reward at the beginning of time step t. It is the sum of rewards from time step t to all future time steps. γ t Represents the power exponent of the discount factor, which indicates the degree to which the reward at time step t is discounted. As the time step increases, the impact of future rewards on the current cumulative reward will gradually decrease. R(s t ,a t ) means that at time step t, the system is in state s t and take action a t The immediate reward obtained. This reward function R reflects the user's feedback and emotional state under specific states and actions.

[0077] After learning the optimal strategy π* according to the reinforcement learning model, the following steps are also included:

[0078] According to the user's immediate feedback (s t ,r t ) Real-time dynamic adjustment of virtual scenes and interactive plots, the specific expression is:

[0079]

[0080] Among them, s t is the state at time t, r t is the reward at time t, π t is the strategy at time t, π t+1 is the strategy at time t+1.

[0081] Psychological relief methods include using holographic projection technology to enable users to experience personalized scenes, music, animal and plant elements, and interactive games in the real world, and interacting with users during the implementation of the plan, adjusting the scenes and difficulty of interactive games based on the users' emotional state.

[0082] During real-time evaluation: Q1, Q2, T1, and T2 are used as input parameters, and the trained convolutional neural network is used to conduct a real-time evaluation of the user's mental health status to obtain the current real-time evaluation results.

[0083] Based on the above method, the present invention provides a system for determining a psychological relief program, including: a collection module, a processing module, a joint analysis module and a stress processing module.

[0084] The acquisition module is used to obtain the user's mental health assessment data and collect voice and video information during the user test; the processing module is used to convert voice information into text, obtain the user's emotional characteristics by performing natural language processing on the text, and extract the user's facial features from the video information, and classify emotions based on facial features; the joint analysis module is used to perform correlation feature analysis on the emotion classification results and emotional features, and obtain the changing relationship between facial features and emotional features; the stress processing module is used to determine the source of the user's psychological stress through mental health assessment data and the changing relationship, and determine the psychological relief plan based on the source of psychological stress.

[0085] like Figure 1As shown in the figure, the acquisition module includes a scale test module and an AI robot module. The scale test module provides users with personalized scale selection, collects the scale information filled in by users to generate the preliminary psychological health evaluation results and suggestions for users; the AI ​​robot module adopts natural language processing (NLP), dialogue management and learning and optimization functions to achieve more intelligent human-computer interaction functions. Through voice and video communication, the user's immediate emotional fluctuations, facial micro-expressions and movement changes are analyzed. Combined with the data information and behavior analysis and sound adjustment units in the scale test module, the user's psychological health status is evaluated in real time, and personalized and targeted measures are taken in time to guide and clear the abnormal psychological health status of the user through the evaluation results; the processing module and the joint analysis module complete the collection of user data information and analysis results of each module, and further calculate and analyze the user's psychological health status. The analysis results include the final evaluation results and suggestions for the user's psychological health. According to the analysis results, the system opens corresponding access permissions for different user types; the pressure processing module calls the AI ​​robot module and the statistical analysis module. After the collection of user data information is completed, the system generates personalized relief plans for users through intelligent analysis. Through the optical principles of holographic projection technology, a safe and realistic three-dimensional virtual imaging environment is created for users, providing a three-dimensional personalized stress relief experience.

[0086] It also includes a login module and a consultation module. The login module is used to provide a variety of login methods, including face recognition, voice recognition, fingerprint authentication and account password authentication. User login account types include students, parents, and teachers; the system supports users to adjust and control operation permissions according to login methods and account types. The consultation module includes an online consultation submodule and a file management submodule. The online consultation submodule is used to build a platform for communication with professional psychologists to enable users to conduct real-time online consultations or make offline appointments for consultations. The file management submodule will fully record the consultation information of each user, aiming to provide users with a professional and integrated network platform.

[0087] In the login module, the system will match different operation permissions according to different user types and login methods. For example, students and parents who log in to the system with their account password can only view assessment suggestions, and teachers who log in to the system with their account password can only view assessment results. The student and parent login modules provide a variety of login methods, including face recognition, voice recognition, fingerprint authentication, and account passwords, to improve user convenience and flexibility and increase system security.

[0088] The AI ​​robot module includes a voice consultation submodule, a video consultation submodule, a behavior analysis unit, and a sound adjustment unit. When receiving negative psychological and emotional changes from the user, the AI ​​robot can quickly process the user's behavior information through behavior analysis and make corresponding positive guidance and communication in words or text. The AI ​​robot learns and optimizes by continuously collecting user feedback and interaction data to continuously improve its answers and performance and provide a better user experience. The following is the process of AI robot learning and optimization:

[0089] 1. Data collection: AI robots will record the conversations and interactions between users and them, including user input, robot responses, user feedback, etc. This data is used to analyze user behavior and needs.

[0090] 2. Data preprocessing: Before learning, the collected data needs to be preprocessed and cleaned to ensure the accuracy and consistency of the data.

[0091] 3. Feature extraction: From the collected data, the AI ​​robot extracts features, which can include the user’s intentions, emotions, conversation flow, etc., in order to analyze and model the data.

[0092] 4. Model training: AI robots use technologies such as machine learning and deep learning to train models using preprocessed data and extracted features. The model training process will be iterated and optimized based on user feedback and data accuracy.

[0093] 5. Feedback loop: The trained model will be applied to actual conversations to interact with users. At the same time, the AI ​​robot will continue to collect user feedback and add this feedback as new training data to the data set.

[0094] 6. Model optimization: Through the data collected through the feedback loop, the AI ​​robot can identify the weaknesses and deficiencies of the model and make corresponding optimizations and improvements.

[0095] 7. Adaptability: During the learning and optimization process, AI robots will also make adaptive adjustments based on the interaction habits and preferences of different users to better meet personalized needs.

[0096] The specific functions and roles of the behavior analysis unit are:

[0097] 1. Behavior analysis: The behavior analysis unit analyzes the user's behavior during the interaction with the AI ​​robot. These behaviors may include voice intonation, expressions, movements, conversation content, etc. By analyzing these behaviors, the behavior analysis unit can obtain information such as the user's emotions and emotional state.

[0098] 2. Emotion recognition: Based on the results of behavioral analysis, the behavioral analysis unit uses emotion recognition technology and algorithms to identify the user's emotional state, such as joy, anger, sadness, anxiety, etc. Emotion recognition can help AI robots better understand the user's emotional needs and provide more appropriate responses.

[0099] 3. Psychological state analysis: The behavior analysis unit will also analyze the user's psychological state by analyzing the user's behavior and conversation content. This includes the user's psychological pressure, emotional fluctuations, self-confidence, etc. Psychological state analysis helps to understand the user's current psychological state and needs, so as to provide better psychological guidance.

[0100] 4. Positive guidance: The behavior analysis unit provides positive guidance to users based on the results of emotion recognition and psychological state analysis, that is, it enhances users’ self-confidence and positive emotions through encouragement and positive words, thereby improving users’ mental health.

[0101] 5. Communication: If the behavior analysis unit identifies that the user has an abnormal mental health state, it will try to help the user release emotions, reduce stress and anxiety through communication. Communication can include listening to the user's problems and feelings, understanding the user's needs, and providing appropriate suggestions based on the user's situation.

[0102] The emotion recognition technology and algorithm can be a Gaussian mixture model algorithm.

[0103] The voice adjustment unit can adjust the AI ​​voice manually or automatically. Users can switch the voice type of the AI ​​robot according to their personal preferences, including male voice, female voice and private customized voice. If the user does not specify the voice system, the AI ​​robot can automatically match the user based on the user's behavior pattern and emotional state. For example, if the user has strong emotional fluctuations during the test, the AI ​​robot will use emotion recognition technology to intelligently adjust the timbre, tone, speed and intonation of the voice to control the user's emotions to remain in a relatively stable and safe state, providing users with higher emotional value.

[0104] Both the voice consultation submodule and the video consultation submodule have natural language processing, psychological consultation, and voice analysis functions. The natural language processing (NLP) function covers many aspects, mainly including the following core steps:

[0105] 1. Speech Recognition: Speech recognition is the process of converting the user's spoken voice input into text. In this step, the AI ​​robot uses speech recognition technology and algorithms to convert the user's voice into computer-readable text for subsequent processing and understanding.

[0106] 2. Semantic understanding: Semantic understanding is the process of semantically parsing and understanding the text input by the user. AI robots use semantic understanding technology and algorithms to analyze the text input by the user, identify the keywords, phrases and sentence structures, and thus understand the user's intentions and needs.

[0107] 3. Intent recognition: Intent recognition is the process of determining the specific intent or requirement expressed in the text input by the user. AI robots use intent recognition technology and algorithms to map the results of semantic understanding to pre-defined intent categories to determine the user's purpose and intention.

[0108] 4. Context maintenance: During the conversation, the AI ​​robot needs to maintain context information in order to better understand the user's continuous conversation. Context maintenance can ensure the coherence and consistency of the conversation, allowing the AI ​​robot to better answer the user's questions.

[0109] 5. Dialogue generation: Dialogue generation is the process of converting the AI ​​robot's answers or instructions to the user into natural language text. The AI ​​robot uses natural language generation technology and algorithms to generate appropriate natural language replies or instructions based on the results of intent recognition and contextual information. When implementing these functions, various NLP technologies and algorithms, such as deep learning, natural language understanding models, intent classifiers, etc., can be used to improve the accurate understanding of user intent and the quality of dialogue.

[0110] Natural language processing technologies include text segmentation algorithms (such as word segmenters and word vector representations), semantic analysis algorithms (such as semantic understanding models based on deep learning), and speech recognition algorithms (such as end-to-end speech recognition models based on deep learning).

[0111] In the voice consultation submodule, the AI ​​robot has natural language processing (NLP) and voice adjustment and behavior analysis functions, and plays the role of a psychologist to provide users with professional responses. Users can have natural conversations with the AI ​​robot through voice, and express their psychological feelings and needs verbally. Specifically, users can express questions, feelings or needs to the AI ​​robot through a microphone. The AI ​​robot will use the NLP function to convert the user's voice into text, then perform semantic understanding and intent recognition on the text, and provide appropriate responses.

[0112] The video consultation submodule adds video transmission and micro-expression and motion capture functions to the voice consultation submodule. Users can make video calls with the AI ​​robot through the camera, making the consultation process closer to a real face-to-face communication experience. At the same time, the AI ​​robot uses computer vision technology to analyze the user's facial expressions and movements to assess the user's mental health in real time and provide more personalized and targeted guidance. Computer vision technology includes facial detection algorithms (such as Haar Cascade, MTCNN, etc.), facial expression recognition algorithms (such as CNN methods based on deep learning), and action recognition algorithms (such as action analysis based on optical flow).

[0113] The function of capturing micro-expressions and movements is as follows: during the video consultation, the user's facial expressions and movements will be captured in real time through the camera. Micro-expressions are usually instantaneous and imperceptible changes in expression. They can reflect the user's true emotions and inner state. AI robots use computer vision technology to analyze and identify these micro-expressions and movements. By analyzing the user's micro-expressions and movements, the AI ​​robot can more accurately identify the user's emotional state, mood swings, anxiety level and other emotional conditions. This helps to understand the user's actual psychological needs in order to guide the user. The AI ​​robot will make corresponding guidance measures based on the micro-expressions and movements captured during the video consultation, combined with previous interaction data and emotional analysis results. Measures include a more concerned tone, emotional comfort, positive psychological guidance, etc., to help users better understand their emotions and needs, and provide timely psychological counseling.

[0114] For the AI ​​robot module, when receiving the voice output by the user, the AI ​​robot uses natural language processing technology to intelligently detect the voice output by the user, and can play the role of a psychologist to intelligently switch between different languages ​​according to the voice language expressed by the user and the questions consulted. Optionally, during the communication process, the AI ​​robot will combine the behavior analysis unit and the sound adjustment unit to evaluate the user's mental health status in real time according to the scale test results and voice analysis and take personalized measures. The measures taken include changing the timbre, tone, speed, intonation, playing background music, and telling stories to guide and dredge the user. The video consultation submodule includes user micro-expressions and motion capture. During the communication process, the AI ​​robot will combine the behavior analysis unit and the sound adjustment unit to evaluate the user's mental health status in real time according to the scale test results and voice analysis, captured micro-expressions and movements, and take personalized measures. The measures taken include changing the timbre, tone, intonation, playing background music, playing videos, and telling stories to guide and dredge the user. The AI ​​robot module has natural language processing and behavior analysis functions, which can realize natural dialogue with users and provide positive guidance and communication to users through behavior analysis functions. This enables the system to understand user needs more real-time and intelligently and take personalized measures.

[0115] The specific implementation methods of personalized measures are:

[0116] 1. Behavior analysis: AI robots analyze users’ interactive behaviors in the system through behavior analysis units to understand users’ usage habits, interests, and reactions and answers to different questions. These behavior patterns can reveal users’ personality traits and psychological tendencies, laying the foundation for providing personalized services.

[0117] 2. Emotional state recognition: Through voice, text and other interactive methods, AI robots use emotion recognition technology to perceive the user's emotional state, such as joy, anger, sadness, tension and relaxation. According to the changes in emotions, AI robots can adjust the tone, attitude and speed of the reply to better adapt to the user's emotional needs and provide more considerate replies.

[0118] 3. Comprehensive evaluation: The AI ​​robot combines the data information in the scale test module with the behavior analysis and voice adjustment unit to conduct a real-time evaluation of the user's mental health status. Through the comprehensive evaluation results, the AI ​​robot can gain an in-depth understanding of the user's mental health status and problems, and thus answer the user's questions in a targeted manner and provide personalized suggestions. The data information includes the user's age, social background, consulting questions, etc.

[0119] 4. Personalized responses: Combining behavioral patterns, emotional states, and scale test results, AI robots can provide customized responses based on the user's personalized needs and psychological state. For example, for users who are in a low mood, AI robots can comfort and encourage them with a warm tone; for anxious users, AI robots can provide calm and rational advice;

[0120] Furthermore, the statistical analysis module will collect the user characteristics and analysis results collected in the login module, scale test module, AI robot module, stress relief module, and consultation module, and further calculate and analyze the user's mental health status. The analysis results include the final evaluation results and suggestions for the user's mental health. Different user permissions can view different contents. For example, students and parents who use their account passwords to complete the system login can only view the final evaluation suggestions, and teachers who use their account passwords to complete the system login can only view the final evaluation results.

[0121] The computational analysis of user data information includes a method for evaluating mental health by weighting the different user characteristics collected.

[0122] Users can selectively enter the stress relief module, which includes an AI processing submodule. The AI ​​processing submodule is the core part of the stress relief module. The AI ​​processing submodule will call the AI ​​robot module and the statistical analysis module to collect user data information and analyze it. At the same time, data mining technology is used or users are directly interacted with to infer the specific source of psychological stress in users, such as emotional experience, social environment, or even a specific type of thing or event.

[0123] Data mining technology means that the AI ​​processing submodule can use machine learning algorithms (such as reinforcement learning algorithms) to conduct in-depth mining of user data and find out the characteristics related to the user's psychological state. Direct interaction with users means that the AI ​​processing submodule can directly interact with users to obtain the user's psychological state and emotional feedback through questions and answers, dialogues, and capturing user movements and micro-expressions. Such an interactive process can obtain user information more timely and flexibly.

[0124] Through data mining techniques or direct interaction with users, AI can better identify users’ psychological needs and sources of stress to generate personalized relief solutions.

[0125] The steps of the reinforcement learning algorithm in the AI ​​processing submodule are shown in step S4 below.

[0126] Based on the user data information obtained by the AI ​​processing submodule, the AI ​​processing submodule uses a reinforcement learning algorithm to customize a personalized starting virtual scene and interactive task main line for each user, and obtains the user's psychological state and emotional feedback through questions and answers, dialogues, and capturing user movements and micro-expressions during the interaction with the user. Through sentiment analysis technology, it analyzes the user's tone, speaking speed, and word usage characteristics in the conversation, and captures the user's facial expressions and body language through the camera to determine whether the user is currently feeling happy and relaxed or anxious and uncomfortable, and then adjusts the virtual scene and interactive tasks in real time to generate a personalized psychological relief plan.

[0127] The AI ​​processing submodule customizes personalized relief solutions for each user based on the determined user data information and interaction information. These relief solutions specifically use holographic projection technology to generate personalized scenes, such as projecting beautiful sunsets, gorgeous starry skies, or soothing beach views, while playing soft music. Such scenes can help users relax and relieve emotional stress. Virtual games can also be customized through holographic projection technology to relieve psychological pressure. In this customized game experience, users will immerse themselves in virtual characters and participate in puzzle solving, adventure, or relaxing atmosphere plots. For users with lower emotions, the game plot and difficulty can be adjusted, and some simple and easy-to-implement tasks can be provided to help users soothe their mood; for users with more tense emotions, some challenging tasks can be provided to help users divert their attention to achieve the effect of psychological relief. In the game, users can enter the game character, temporarily break away from real life, remove the pressure and worries of daily life, and enjoy interaction and communication with the virtual world.

[0128] The customized virtual game themes are:

[0129] 1. Self-exploration: Design a virtual world adventure game to meet the needs of users in terms of self-cognition and exploration. Users can play the role of an adventurer in the interactive game and embark on a journey of self-exploration. In this virtual world, users will face various obstacles and challenges, and will gain a sense of accomplishment and positive feedback every time they complete a task. Psychologists or mentors can be added to the interactive game to provide guidance and advice to users, help them recognize their strengths and weaknesses, and explore their deepest feelings and emotions.

[0130] 2. Emotional liberation: Design a virtual emotional world to meet the needs of users in terms of emotional release and emotional expression. Users can play a role in the interactive game and express their emotions freely, whether it is joy, anger, sadness or fear. The environment provided in the game is relatively safe and secure, allowing users to release their emotions and relieve their inner pressure and burden by releasing emotional pressure.

[0131] 3. Social training: Design a virtual social training game to address users’ anxiety and uneasiness in social situations. Users can play a virtual social role in the interactive game and participate in various social activities and interactions. The game can simulate various social scenes, such as parties, speeches, interviews, etc. Users can simulate communicating with others in virtual reality through the game, strengthen their social skills, thereby increasing their self-confidence in dealing with social situations and reducing their social fear or anxiety.

[0132] 4. Historical Reappearance: Design a virtual historical reappearance game to relieve the psychological pressure of users in past experiences. Users can play a role, go back to a certain period in the past, and re-experience past experiences. In the interactive game, users can deal with past events in different ways and attitudes, thereby achieving a new understanding and guidance of past experiences.

[0133] Such personalized role-playing games not only meet the entertainment needs of users, but more importantly, provide a unique psychological relief program. Users can try different roles and experience different plots in interactive games, thereby releasing emotions in the virtual world, exploring their inner feelings, and improving their self-confidence and ability to cope with stress through the challenges and sense of achievement in interactive games. Through such role-playing game experience, users can get deeper psychological relief and relaxation, realize the cognition and regulation of their own emotions and behaviors, and achieve better psychological relief effects.

[0134] In summary, the AI ​​processing submodule calls the AI ​​robot module and the statistical analysis module to collect and analyze user data information, obtain the specific source of psychological pressure on users by data mining or directly asking users, and then combine the reinforcement learning algorithm to generate personalized relief plans based on specific sources and interactive information. The relief plan includes generating personalized scenes using holographic projection. Optionally, the scene includes personalized music, plants and animals, and scene color temperature. Holographic projection can also be used to generate personalized scenes and enable users to participate in virtual game experiences in the real world.

[0135] The stress relief module uses holographic projection technology to generate personalized relief solutions. At the same time, compared with virtual reality technology, the use of holographic projection technology makes it less likely for users to experience dizziness and other discomfort. The holographic projection virtual game experience provides users with a novel and immersive entertainment method, providing users with a better relief experience and better relief effect.

[0136] The consultation module includes an online consultation submodule and a file management submodule, providing users with a communication platform with professional psychologists. In the online consultation submodule, users can choose a given psychologist for online consultation and appointment consultation. Psychologists can refer to the evaluation results of the scale test module and the statistical analysis module to answer the user's consultation questions, and can also make an appointment with the user for offline consultation at a local hospital. The file management submodule will fully save the complete consultation information of each case and scientifically record the growth process of the user's consultation case.

[0137] The present invention also provides a computer device, including a memory and a processor, wherein a program is stored in the memory, and when the program is executed by the processor, the processor executes the steps of a method for determining a psychological relief program.

[0138] In accordance with the disclosed embodiments, a computing device may communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth communications, etc.), or with any device (e.g., routers, modems, etc.) that enables a computing device to communicate with one or more other computing devices.

[0139] The present invention also provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of a method for determining a psychological relief program are implemented.

[0140] According to the disclosed embodiments, the storage medium may be a non-volatile computer-readable storage medium, such as but not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, the storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0141] The above content is a further detailed description of the present invention in combination with a specific preferred embodiment. For technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as belonging to the protection scope of the present invention.

Claims

1. A method for determining a psychological relief program, characterized in that: The steps include: Obtain the user's mental health assessment data, and collect the user's voice and video information during the test; Convert the voice information into text, obtain the user's emotional characteristics by performing natural language processing on the text, extract the user's facial features from the video information, and classify the emotions based on the facial features; The emotion classification results and emotional features are correlated and analyzed to obtain the changing relationship between facial features and emotional features; The source of the user's psychological stress is determined through the psychological health assessment data and the change relationship, and a psychological relief plan is determined based on the source of the psychological stress.

2. A method for determining a psychological relief program as claimed in claim 1, characterized in that: The obtaining of the user's mental health assessment data is specifically as follows: Different scales are set according to the user's age, social background and corresponding psychological condition; the social background information includes the user's educational level and family members, and the scale includes text, pictures and video elements; Users input information into the corresponding scale to generate mental health assessment data.

3. The method for determining a psychological relief program according to claim 1, characterized in that: Before classifying emotions by facial features, the method further includes: Establishing a training data set for a convolutional neural network model; the training data set includes sample data of a specific population representing different psychological states, each sample data including feature data representing the psychological state and a corresponding category label; The training data set is input into the convolutional neural network model, and the connection weights and errors of the convolutional neural network model are continuously adjusted until the training error of the convolutional neural network model reaches a predetermined threshold.

4. The method for determining a psychological relief program according to claim 1, characterized in that: Determine psychological relief plans based on the source of psychological stress, including: Determine psychological stress information based on the source of psychological stress and process the psychological stress information, specifically: Define a virtual environment M. Users can relieve their psychological stress by interacting with the virtual scene. The state s of the environment M contains the user's emotional state and interaction history information. Obtain the reward function R(s,a) based on user feedback and emotional status; specifically: Construct a reinforcement learning model (S, A, P, π, R), and obtain a psychological relief plan by training the reinforcement learning model (S, A, P, π, R); According to the optimal strategy π* learned by the reinforcement learning model, personalized scenes, plots and task difficulties are generated to adapt to the user's psychological needs and emotional state; the specific expression of the optimal strategy π* is: Among them, s∈S, S is the state set, action a∈A, A is the action set, P is the transition probability, π is the strategy set, R is the reward, r + is a positive reward, r - is a negative reward, G t To maximize the cumulative reward, E is the relevant function.

5. A method for determining a psychological relief program as claimed in claim 4, characterized in that: After the optimal strategy π* is learned according to the reinforcement learning model, the following steps are also included: According to the user's immediate feedback (s t ,r t ) Real-time dynamic adjustment of virtual scenes and interactive plots, the specific expression is: Among them, s t is the state at time t, r t is the reward at time t, π t is the strategy at time t, π t+1 is the strategy at time t+1.

6. The method for determining a psychological relief program according to claim 1, characterized in that: The psychological relief method includes using holographic projection technology to enable users to experience personalized scenes, music, animal and plant elements and interactive games in the real world, and interacting with users during the implementation of the solution, adjusting the scenes and the difficulty of interactive games according to the users' emotional conditions.

7. A system for determining a psychological relief program, characterized in that: include: The collection module is used to obtain the user's mental health assessment data and collect the user's voice and video information during the test; A processing module, used to convert the voice information into text, obtain the user's emotional characteristics by performing natural language processing on the text, extract the user's facial features from the video information, and classify emotions based on the facial features; The joint analysis module is used to perform correlation feature analysis on the emotion classification results and the emotional features to obtain the changing relationship between facial features and emotional features; The stress processing module is used to determine the source of the user's psychological stress through the psychological health assessment data and the change relationship, and to determine the psychological relief plan based on the source of the psychological stress.

8. A computer device, characterized in that: The method comprises a memory and a processor, wherein a program is stored in the memory, and when the program is executed by the processor, the processor executes the steps of a method for determining a psychological relief program as claimed in any one of claims 1 to 6.

9. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a method for determining a psychological relief program according to any one of claims 1 to 6 are implemented.

Images