Mental health screening method based on immersive game task and EEG (electroencephalogram) analysis

Abstract

The invention discloses a psychological health screening method based on immersive game tasks and EEG (electroencephalogram) analysis, and relates to the technical field of psychological health screening, comprising the following steps: determining a psychological health dimension to be evaluated, and integrating VR (virtual reality) and EEG equipment; extracting corresponding data from different psychological dimension game data of VR to obtain emotion stability, cognitive ability and social skill evaluation values; and constructing a psychological health evaluation model based on linear regression through the EEG feature data, outputting score results on different psychological dimensions, and automatically generating a psychological health evaluation report in combination with corresponding evaluation values. The method solves the problems that it is difficult to establish a player attention state classification model to determine the attention concentration duration of players in different game stages, it is difficult to quantify evaluation values constructed by feature data in different psychological dimensions, and it is lack of automatically generating a psychological health evaluation report by combining VR games and EEG electroencephalogram signals.

Description

Technical Field

[0001] The present invention relates to the technical field of mental health screening, and particularly to a mental health screening method based on immersive game tasks and electroencephalogram (EEG) analysis. Background Art

[0002] With the increase of social pressure, mental health problems have received increasing attention. Traditional mental health screening methods mainly rely on questionnaires and interviews, but they have disadvantages such as strong subjectivity and low efficiency. Immersive VR games, as a new form of entertainment, can provide a highly interactive and immersive experience, while EEG technology can monitor brain activities in real time, providing a new way for mental health assessment. The data collected by combining immersive VR games and EEG devices are automatically generated into a physiological health assessment report.

[0003] Although existing mental health screening methods have achieved mental health screening to a certain extent, they have disadvantages such as strong subjectivity and low efficiency. It is difficult to combine the power changes of multiple waveforms with the player's behavior performance as the characteristics for evaluating the degree of attention concentration, establish a classification model for the player's attention state, and it is difficult to train a tone classification model using machine learning algorithms. There is a lack of obtaining the corresponding evaluation values of players in different mental dimension games through comprehensively combining the data obtained from the constructed model and sensors, and there is a lack of adding the scores of specific mental dimensions predicted by the constructed model to the values on the corresponding dimensions to obtain the comprehensive scores on each mental dimension. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art; for this purpose, the present invention proposes a mental health screening method based on immersive game tasks and EEG analysis to solve the following technical problems:

[0005] Although existing mental health screening methods have achieved mental health screening to a certain extent, they have disadvantages such as strong subjectivity and low efficiency. It is difficult to combine the power changes of multiple waveforms with the player's behavior performance as the characteristics for evaluating the degree of attention concentration, establish a classification model for the player's attention state, and it is difficult to train a tone classification model using machine learning algorithms. There is a lack of obtaining the corresponding evaluation values of players in different mental dimension games through comprehensively combining the data obtained from the constructed model and sensors, and there is a lack of adding the scores of specific mental dimensions predicted by the constructed model to the values on the corresponding dimensions to obtain the comprehensive scores on each mental dimension.

[0006] To solve the above problems, the present invention provides a mental health screening method based on immersive game tasks and EEG analysis, including the following steps:

[0007] S1: Determine the mental health dimensions to be evaluated, design corresponding immersive VR game scenarios, select a high-precision head-mounted EEG acquisition device, and integrate the head-mounted EEG acquisition device with the VR device;

[0008] S2: Extract the corresponding feature data of players from the game data of different mental dimensions in VR. In emotion regulation games, an emotional stability evaluation value can be constructed by monitoring various physiological indicators of players. In cognitive challenge games, a cognitive ability evaluation value can be constructed by recording the attention concentration duration rate and memory accuracy data of players. In social interaction games, a social skill evaluation value can be constructed by analyzing their communication frequency and tone recognition data;

[0009] S3: After preprocessing the collected EEG signals, extract the feature indicators related to mental health, construct a mental health evaluation model based on linear regression to output the score results on different mental dimensions, automatically generate a mental health education measurement and problem screening report in combination with the corresponding evaluation values, and give feedback after the player finishes the game.

[0010] Furthermore, step S1 includes the following steps:

[0011] According to psychological theories and clinical studies, the dimensions of mental health include the emotional health dimension, the cognitive health dimension, and the social health dimension. Based on these three dimensions of mental health, design corresponding emotion regulation games, cognitive challenge games, and social interaction games. During the game process, players need to wear VR devices to collect the corresponding game event data in the three types of games. Determine the head-mounted EEG device, and before the game starts, introduce the test process and precautions to the players, turn on the head-mounted EEG device, perform device calibration, and when the players are ready, record the EEG data.

[0012] Furthermore, the integration of the head-mounted EEG acquisition device with the VR device includes the following steps:

[0013] During system initialization, create a clock source as the time reference for the entire system. When the VR game starts running, record the current timestamp and send the timestamp to the EEG acquisition device as the starting time point for its data acquisition. When a key event occurs during the running of the VR game, obtain the corresponding timestamp and record the type and description information of the key event. The EEG acquisition device continuously collects EEG signals and marks each sampling point with a timestamp, and stores the VR game data and EEG data with timestamps in the database respectively.

[0014] Furthermore, the construction of an emotional stability evaluation value by monitoring various physiological indicators of players in emotion regulation games includes the following steps:

[0015] For emotion regulation games, the game can monitor the player's physiological indicators, including heart rate changes, respiratory rate and blood pressure changes, and record the data of the corresponding physiological indicators. Through the heart rate sensor, respiratory rate sensor and blood pressure sensor, the player's heart rate value, respiratory rate value, systolic blood pressure and diastolic blood pressure value can be obtained in real time, and the multiple values ​​can be combined to obtain the emotional stability evaluation value;

[0016] Emotional stability evaluation value calculation formula:

[0017]

[0018] Among them, E stability represents the emotional stability evaluation value, H 0 represents the initial heart rate, H t Indicates the current heart rate, B 0 represents the initial respiratory rate, B t Indicates the current respiratory rate, P s represents the initial systolic blood pressure, P sys-t Indicates the current systolic blood pressure, P d represents the initial diastolic pressure, P dia-t represents the current diastolic pressure, ω 1 ,ω 2 and ω 3 Represents the corresponding weight coefficient.

[0019] Furthermore, in the cognitive challenge game, the cognitive ability evaluation value is constructed by recording the player's attention concentration duration and memory accuracy data, including the following steps:

[0020] Multiple electrodes are placed on the player's scalp according to specific standard positions to collect the player's EEG signals during the game, and preprocess them. According to a preset time window, the preprocessed EEG signals are fast Fourier transformed, and the power values ​​of α waves, θ waves and β waves are calculated. Then, combined with the player's behavioral performance records, the attention state of each time period is annotated, and a training set for training is constructed. The training set is classified according to the attention state label, and the classified training data is trained using a machine learning algorithm to establish a player's attention state classification model. The player's EEG data during the game and the corresponding attention state label are input into the model as training data. The model outputs the attention state prediction result. By analyzing the continuous EEG data, the player's concentration time at different game stages is determined. When the model predicts that the player is in a focused state, the timing starts, and the time until the predicted result changes to a non-focused state is used as the player's concentration time at that game stage.

[0021] Set a recording interface in the game. After the game ends, the player inputs the text they remember in the text box in sequence to record the content of the text. The game program compares it with the content of the original text and calculates the memory accuracy rate of the player;

[0022] Weight and add the ratio value of the player's attention concentration duration to the total duration in this game stage and the player's memory accuracy rate to obtain the cognitive ability evaluation value;

[0023] Cognitive ability evaluation value calculation formula:

[0024]

[0025] Among them, C ability represents the cognitive ability evaluation value, T focus represents the player's attention concentration duration in this game stage, T all represents the total duration of the player in this game stage, M accuracy represents the player's memory accuracy rate, ω 4 and ω 5 represent the corresponding weight coefficients.

[0026] Furthermore, analyzing the communication frequency and tone recognition data in the social interaction game to construct the social skill evaluation value includes the following steps:

[0027] Set a fixed observation time window. Within this time window, count the number of effective communications between players. Effective communication is defined as sending meaningful text messages, voice messages, or performing specific game-related activity operations. Calculate the average communication frequency of multiple time windows. After adding up the number of communications in each time window, divide by the number of time windows to obtain the player's average communication frequency;

[0028] Use a machine learning algorithm to train a tone classification model: collect chat record data from the social interaction game, determine the rules for annotating positive and negative tones, use the multiple annotation method to annotate the same chat record, perform preprocessing operations on the collected chat record data, including cleaning and using the combination of the bag-of-words model and term frequency-inverse document frequency to convert the data into a feature vector based on word frequency. Use the training set to train the machine learning algorithm based on the long short-term memory network architecture, deploy the trained model to the game. During the game, obtain the player's chat record and input it into the trained model. The model outputs the probability that the message belongs to different tone categories. According to the probability output by the model, convert the probability into the corresponding tone value according to the rules, and perform weighted summation on different tone values according to the set weights to obtain the comprehensive tone value;

[0029] Comprehensive tone value calculation formula:

[0030] Z moods = (1 × G positive ) × ω 6 + (-1 × G negative ) × ω 7

[0031] where Z moods represents the comprehensive tone value, G positive represents the positive tone probability, G negative represents the negative tone probability, ω 6 and ω 7 represent the corresponding weight coefficients;

[0032] The weighted sum of the average communication frequency of players and the comprehensive tone value gives the social skill evaluation value;

[0033] Social skill evaluation value calculation formula:

[0034] S ability = F avg × ω 8 + Z moods × ω 9

[0035] where S ability represents the social skill evaluation value, F avg represents the average communication frequency of players, Z moods represents the comprehensive tone value, ω 8 and ω 9 represent the corresponding weight coefficients.

[0036] Furthermore, after performing preprocessing operations on the collected EEG signals, extracting feature indicators related to mental health, and constructing a mental health assessment model based on linear regression to output score results on different psychological dimensions, the steps are as follows:

[0037] Import the collected EEG signal data into data analysis software for preprocessing operations. According to the different types and task requirements of the game, divide the data into corresponding segments, mark each segment, use signal processing methods to extract the spectral features of the EEG signals and the functional connection features between brain regions from the processed EEG data, perform statistical analysis on the extracted feature indicators respectively, calculate the mean, median, and standard deviation. According to the statistical analysis results, use the feature indicators after statistical analysis processing as independent variables and the scores of players on the psychological dimension as dependent variables, and use the training set data to train the mental health assessment model based on linear regression to predict the scores of players on specific psychological dimensions.

[0038] Furthermore, automatically generate a mental health education measurement and problem screening report in combination with the corresponding evaluation values, and give feedback after the player finishes the game, including the following steps:

[0039] Add the emotional stability evaluation value, cognitive ability evaluation value, social skill evaluation value obtained from the game in different psychological dimensions and the score result predicted by the model to obtain the scores in different psychological dimensions, generate a mental health assessment report for the player, and list the names, score values and corresponding explanatory notes of each dimension in the player's mental health assessment report;

[0040] According to the score situation of the player in each psychological dimension, combine the preset norm reference, analyze the psychological problems existing in the player, and provide personalized suggestions and guidance for the player according to the discovered psychological problems. Review the generated mental health assessment report and deliver the approved mental health assessment report to the player in an appropriate way.

[0041] Advantages of the present invention:

[0042] In the present invention, multiple electrodes are placed on the player's scalp according to specific standard positions, and power spectral density analysis is performed on the electroencephalogram signals. It shows that when the attention is concentrated, the power of the α wave and the θ wave decreases, while the power of the β wave increases. The three power changes are used as the characteristics for evaluating the degree of attention concentration. The machine learning algorithm is used to learn and train the extracted characteristics, establish a player attention state classification model, the model outputs the attention state prediction result, and determines the attention concentration duration of the player in different game stages, realizing the potential application prospect of health monitoring;

[0043] In the present invention, a tone classification model is trained by using a machine learning algorithm: collect chat record data from social interaction games, determine the rules for annotating positive and negative tones, use the multiple annotation method to annotate the same chat record, use the bag-of-words model and term frequency-inverse document frequency combination to convert the data into a feature vector based on word frequency, and deploy the trained model to the game. The model outputs the probability that the message belongs to different tone categories, realizing that when players communicate, it can immediately judge whether the message tone is positive or negative, which helps other players to respond appropriately;

[0044] In the present invention, add the emotional stability evaluation value, cognitive ability evaluation value, social skill evaluation value obtained from the game in different psychological dimensions and the score result predicted by the model to obtain the scores in different psychological dimensions, generate a mental health assessment report for the player. Through the mental health assessment report in the game, more players can come into contact with the concepts and knowledge of mental health, and improve the public's awareness and attention to mental health problems. This helps to break the social prejudice and misunderstanding of mental problems and create a more understanding and supportive social atmosphere for mental health. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flowchart of the implementation method of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0046] The following combines Figure 1 to make a detailed description of the implementation manner and specific operation process of the present invention. However, the protection scope of the present invention is not limited to the following embodiments.

[0047] Please refer to Figure 1 As shown, the present invention is a mental health screening method based on immersive game tasks and electroencephalogram (EEG) analysis, including the following steps:

[0048] S1: Determine the mental health dimensions to be evaluated, design corresponding immersive VR game scenarios, and select a high-precision head-mounted EEG acquisition device, and integrate the head-mounted EEG acquisition device with the VR device;

[0049] S2: Extract the corresponding characteristic data of the player from the VR game data of different psychological dimensions. In the emotion regulation games, the emotional stability evaluation value can be constructed by monitoring various physiological indexes of the player. In the cognitive challenge games, the cognitive ability evaluation value can be constructed by recording the attention concentration duration rate and memory accuracy data of the player. In the social interaction games, the social skill evaluation value can be constructed by analyzing the communication frequency and tone recognition data;

[0050] S3: After preprocessing the collected EEG signals, extract the characteristic indexes related to mental health, construct a mental health evaluation model based on linear regression to output the score results on different psychological dimensions, automatically generate a mental health education measurement and problem screening report in combination with the corresponding evaluation value, and give feedback after the player finishes the game.

[0051] Specifically, the purpose of the study was to explore the relationship between different mental health dimensions and player performance and EEG data through specially designed VR games. VR equipment and head-mounted EEG equipment with stable performance were selected, and corresponding games related to the three mental health dimensions were designed. A process guide was written. After the players arrived at the laboratory, the purpose, process, and precautions of the test were introduced in detail to the players. The basic demographic information of the players was collected. After ensuring that the VR equipment and EEG equipment were ready, the VR game and EEG data acquisition software were started at the same time to start recording the player's game data and EEG signal data. In emotion regulation games, the heart rate value, respiratory rate value, systolic blood pressure and diastolic blood pressure values ​​were combined to obtain the emotional stability evaluation value. In cognitive challenge games, a player's attention state classification model was constructed to determine the emotional stability evaluation value. The attention span rate is determined and combined with the memory accuracy to obtain the cognitive ability evaluation value. In social interactive games, the tone classification model outputs the probability that the message belongs to different tone categories. The probability is converted into corresponding tone values ​​and weighted addition is performed to obtain a comprehensive tone value. The comprehensive tone value is combined with the average communication frequency of the player to obtain the social skill evaluation value. From different psychological dimension games, the player's emotional stability evaluation value, cognitive ability evaluation value and social skill evaluation value are obtained. The relevant features of the player's EEG data in the game are input into the pre-built model to obtain the score of the specific psychological dimension predicted by the model. After standardizing the evaluation value and the model prediction score, the evaluation value and the model prediction score on different psychological dimensions are added to obtain the comprehensive score on each psychological dimension, and a mental health assessment report is generated for the player.

[0052] In one embodiment of the present invention, the step S1 comprises the following steps:

[0053] According to psychological theory and clinical research, the dimensions of mental health include emotional health, cognitive health, and social health. Based on the three dimensions of mental health, corresponding emotion regulation games, cognitive challenge games, and social interaction games are designed. During the game, players need to wear VR devices to collect corresponding game event data in the three categories of games; determine the head-mounted EEG device, and before the game starts, introduce the test process and precautions to the players, turn on the head-mounted EEG device, calibrate the equipment, and record the EEG data when the players are ready.

[0054] Specifically, the research aims to explore the relationship between different mental health dimensions, player performance, and EEG data through specifically designed VR games. Select VR devices and head-mounted EEG devices with stable performance. Based on the three dimensions of mental health, design corresponding games related to mental health dimensions and compile a process guide, including the reception process after players arrive at the laboratory, the method of wearing the device, game operation teaching, rest arrangements during the test, and data recording. After players arrive at the laboratory, the reception staff guides the players to sign in and introduces the purpose, process, and precautions of the test to the players in detail again, collecting the players' basic demographic information, including age, gender, occupation, and education level. Help the players wear the VR device, turn on the Bluetooth connection of the EEG device, and pair it with the acquisition device. After determining that both the VR device and the EEG device are ready, start the VR game and the EEG data acquisition software simultaneously to record the players' game data and electroencephalogram signal data.

[0055] In one embodiment of the present invention, the integration of the head-mounted EEG acquisition device and the VR device includes the following steps:

[0056] During system initialization, create a clock source as the time reference for the entire system; when the VR game starts running, record the current timestamp and send the timestamp to the EEG acquisition device as the starting time point for its data acquisition; when a key event occurs during the running of the VR game, obtain the corresponding timestamp and record the type and description information of the key event. The EEG acquisition device continuously collects electroencephalogram signals and marks the timestamp for each sampling point, and stores the VR game data and EEG data with timestamps in the database respectively.

[0057] Specifically, the high-precision clock service provided by the operating system is used to create a clock source. For example, in many computer systems, a specific API of the operating system can be called to obtain a clock signal accurate to the microsecond or even nanosecond level. This can be used as the source of the time base for the entire system, and the clock source can be calibrated to lay the foundation for subsequent accurate timestamp recording. The VR game software is installed and configured, and the EEG acquisition device is connected and started to be in standby mode, waiting to receive the start timestamp signal and other related instructions from the VR game. When the player starts the VR game, the system immediately obtains the current timestamp from the clock source and calculates it to the millisecond level. This timestamp is quickly sent to the EEG acquisition device as the starting time point for the start of data collection. The VR game system itself also records the timestamp. During the operation of the VR game, the key events in the game are monitored in real time. Key events include but are not limited to players completing tasks, making important decisions, and voice commands. When a key event occurs, the system obtains the current timestamp from the clock source again and records the type of the key event. The key event data with timestamp, event type and description information is temporarily stored in the data cache, and then synchronized and stored with the EEG data. After receiving the starting timestamp, the EEG acquisition device continuously collects EEG signals at the set acquisition rate of 1000Hz. Within a certain time interval, the VR game data and EEG data with timestamps are taken out from the system cache and stored in the database.

[0058] In one embodiment of the present invention, in the emotion regulation game, the emotion stability evaluation value can be constructed by monitoring multiple physiological indicators of the player, including the following steps:

[0059] For emotion regulation games, the game can monitor the player's physiological indicators, including heart rate changes, respiratory rate and blood pressure changes, and record the data of the corresponding physiological indicators. Through the heart rate sensor, respiratory rate sensor and blood pressure sensor, the player's heart rate value, respiratory rate value, systolic blood pressure and diastolic blood pressure value can be obtained in real time, and the multiple values ​​can be combined to obtain the emotional stability evaluation value;

[0060] Emotional stability evaluation value calculation formula:

[0061]

[0062] Among them, E stability represents the emotional stability evaluation value, H 0 represents the initial heart rate, H t Indicates the current heart rate, B 0 represents the initial respiratory rate, B t Indicates the current respiratory rate, P s represents the initial systolic blood pressure, P sys-tRepresents the current systolic blood pressure, P d Represents the initial diastolic blood pressure, P dia-t Represents the current diastolic blood pressure, ω 1 、ω 2 And ω 3 Represents the corresponding weight coefficient.

[0063] Specifically, before the game starts, the heart rate sensor, respiratory rate sensor, and blood pressure sensor are respectively bound to appropriate positions on the player's body, and automatic calibration is performed after startup. When the emotion regulation game starts running, the time intervals of the three sensors are respectively set for acquisition, and the acquired results are transmitted to the data acquisition system. The data acquisition system records the received heart rate value, respiratory rate value, systolic blood pressure, and diastolic blood pressure values according to the corresponding timestamps. The relative change in heart rate is obtained by dividing the absolute difference between the current heart rate value and the initial value by the initial heart rate value. The relative changes in respiratory rate, diastolic blood pressure, and systolic blood pressure are obtained in the same way. When the value of the relative change is larger, it indicates that the emotional stability is lower and the emotional fluctuation is larger. After subtracting the relative change of the corresponding physiological index by 1 respectively, the weighted sum is used to obtain the emotional stability evaluation value. The larger the value of the emotional stability evaluation value, the more stable the emotion. Among them, the value of ω 1 is 0.4, and the values of ω 2 and ω 3 are both 0.3.

[0064] In one embodiment of the present invention, constructing a cognitive ability evaluation value by recording the attention concentration duration rate and memory accuracy data of the player in the cognitive challenge game includes the following steps:

[0065] Place multiple electrodes on the player's scalp according to specific standard positions, collect the electroencephalogram signals of the player during the game, perform preprocessing operations on them, perform fast Fourier transform on the preprocessed electroencephalogram signals according to a preset time window, calculate the power values of α waves, θ waves, and β waves, and then combine the player's behavioral performance records to label the attention state of each time period, construct a training set for training, and classify the training set according to the attention state labels. Use machine learning algorithms to train the classified training data, establish a player attention state classification model, input the electroencephalogram data of the player during the game and the corresponding attention state labels as training data into the model, and the model outputs the attention state prediction result. By analyzing the continuous electroencephalogram data, determine the attention concentration duration of the player in different game stages. When the model predicts that the player is in a focused state, start timing until the prediction result becomes a non-focused state, which is used as the attention concentration duration of the player in this game stage;

[0066] Set a recording interface in the game. After the game ends, the player inputs the text they remember in the text box in sequence to record the content of the text. The game program compares it with the content of the original text and calculates the memory accuracy rate of the player.

[0067] Weight and add the ratio of the attention concentration duration of the player in this game stage to the total duration and the memory accuracy rate of the player to obtain the cognitive ability evaluation value.

[0068] Cognitive ability evaluation value calculation formula:

[0069]

[0070] Among them, C ability represents the cognitive ability evaluation value, T focus represents the attention concentration duration of the player in this game stage, T all represents the total duration of the player in this game stage, M accuracy represents the memory accuracy rate of the player, ω 4 and ω 5 represent the corresponding weight coefficients.

[0071] Specifically, multiple electrodes are placed on the player's scalp at specific standard positions to cover different brain regions. An electroencephalogram recorder is connected to the electrodes to collect the player's electroencephalogram (EEG) signals during the game. After filtering and removing artifacts from the EEG signals, power spectral density analysis is performed on the EEG signals to obtain the power distribution of different frequency components. According to research data, when attention is concentrated, the power of alpha waves and theta waves decreases, while the power of beta waves increases. Combining with the player's behavioral performance records, the attention state of each time period is labeled to construct a training set for training. The training set is classified according to the attention state labels, and a machine learning algorithm is used to train the classified training data to establish a player attention state classification model. The player's EEG data during the game and the corresponding attention state labels are used as training data and input into the model for training. The attention state labels include the focused state and the non-focused state. During the actual game process, the real-time collected player EEG data is input into the trained model, and the model outputs the prediction result of the player's current attention state or attention concentration level. By analyzing the continuous EEG data, the attention concentration duration of the player in different game stages can be determined. For example, when the model predicts that the player is in the focused state, the timing starts until the prediction result changes to the non-focused state. This period of time is the attention concentration duration of the player in this game stage. Dividing the concentration duration by the total duration gives the attention concentration duration rate, which can reflect the player's ability to maintain attention when facing complex tasks. A longer attention concentration duration usually means that the player has better attention control ability and can process game information more deeply. Before the game starts, a set of original text content is preset by the program and presented to the player in a suitable way, such as being displayed on the game interface for a certain period of time to allow the player enough time to read and memorize. After the specified time ends, the original text disappears from the game interface, and a text box is generated on the game interface for the player to input the text content they remember. The game program receives the player's input in real time and stores it in a temporary data structure for subsequent comparison with the original text. The text content input by the player is compared with the original text one by one. During the comparison process, the number of successfully matched words and the total number of words are recorded. The accuracy rate of the player's memory is calculated based on the ratio of the number of successfully matched words to the total number of words, and it is weighted and added to the attention concentration duration rate to obtain the cognitive ability evaluation value. The larger the value of the cognitive ability evaluation value, the higher the player's cognitive ability, where ω 4 and ω 5 both have a value of 0.5.

[0072] In one embodiment of the present invention, analyzing the communication frequency and tone recognition data in a social interaction game to construct a social skill evaluation value includes the following steps:

[0073] Set a fixed observation time window. Within this time window, count the number of effective communications among players. Effective communication is defined as sending meaningful text messages, voice messages, or performing specific game-related activity operations. Calculate the average communication frequency over multiple time windows. After adding up the number of communications in each time window, divide by the number of time windows to obtain the average communication frequency of the player;

[0074] Train a tone classification model using a machine learning algorithm: Collect chat record data from social interaction games, determine the rules for annotating positive and negative tones, use multiple annotation methods to annotate the same chat record, perform preprocessing operations on the collected chat record data, including cleaning and converting the data into feature vectors based on word frequencies using a combination of the bag-of-words model and term frequency-inverse document frequency. Use the training set to train a machine learning algorithm based on the long short-term memory network architecture, deploy the trained model into the game. During the game, obtain the player's chat record and input it into the trained model. The model outputs the probabilities that the message belongs to different tone categories. According to the probabilities output by the model, convert the probabilities into corresponding tone values according to the rules, and perform weighted summation on the different tone values according to the set weights to obtain the comprehensive tone value;

[0075] Comprehensive tone value calculation formula:

[0076] Z moods =(1×G positive )×ω 6 +(-1×G negative )×ω 7

[0077] Where, Z moods represents the comprehensive tone value, G positive represents the positive tone probability, G negative represents the negative tone probability, ω 6 and ω 7 represent the corresponding weight coefficients;

[0078] Assign weights to and add the average communication frequency of the player and the comprehensive tone value to obtain the social skill evaluation value;

[0079] Social skill evaluation value calculation formula:

[0080] S ability =F avg ×ω 8 +Z moods ×ω 9

[0081] Where, S ability represents the social skill evaluation value, F avg represents the average communication frequency of the player, Zmoods Represents the comprehensive tone value, ω 8 and ω 9 represent the corresponding weight coefficients.

[0082] Specifically, set a fixed observation time window. For example, every ten minutes is used as a statistical unit. Within this time window, count the number of effective communications between players. Effective communication can be defined as sending meaningful text messages. Meaningful text messages exclude simple greetings such as "Hello", "Are you there", etc., voice messages, or performing specific game-related activity operations such as sharing resources and jointly attacking enemies. For example, in a multiplayer online cooperative adventure game, within the time period from the 10th to the 20th minute after the game starts, there are five effective text communications between player A and player B. The communication frequency is recorded as 5 times within this time period. To more comprehensively evaluate the communication frequency of players throughout the game, calculate the average communication frequency of multiple time windows. Assume that a game lasts for 60 minutes in total. With a 10-minute time window, there are 6 time windows. Add up the number of communications in each time window and then divide by the number of time windows to obtain the average communication frequency; collect chat record data from social interaction games, determine the rules for annotating positive and negative tones. For example, positive tones may include emotional expressions such as praise, encouragement, and excitement; negative tones cover emotional expressions such as complaints, criticism, and anger. At the same time, provide detailed examples for annotators to accurately judge. Multiple annotators annotate the same chat record and then review the results. Perform preprocessing operations on the collected chat record data, including cleaning and converting the data into a feature vector based on word frequency using a combination of the bag-of-words model and term frequency-inverse document frequency. Use the training set to train a machine learning algorithm based on the long short-term memory network architecture. Deploy the trained model to the game. During the game, obtain the chat record of the player and input it into the trained model. The model outputs the probability that the message belongs to different tone categories. The tone values are multiplied by the corresponding weights and added together to obtain the final tone value. Among them, the tone rule means that the positive and negative tone probabilities output by the model are multiplied by the corresponding tone maximum values to obtain the tone values. The positive tone maximum value is set to 1, and the negative tone maximum value is set to -1. This comprehensive value reflects the overall emotional state of the player in the chat record on that day. A positive value close to 1 indicates that the player's mood is relatively positive; if the value is negative, it means that the player's mood is relatively negative. The closer the value is to -1, the higher the degree of negativity; if the value is close to 0, it means that the player's mood is in a relatively neutral state. The average communication frequency of the player and the comprehensive tone value are weighted and added together to obtain the social skill evaluation value. Among them, ω 6 has a value of 0.6, ω 7 has a value of 0.4, ω 8 and ω 9 are both 0.5.

[0083] In one embodiment of the present invention, after preprocessing the collected EEG signals, feature indicators related to mental health are extracted, and a mental health assessment model based on linear regression is constructed to output score results on different psychological dimensions, including the following steps:

[0084] Import the collected EEG signal data into data analysis software for preprocessing operations. According to different types and task requirements of the game, the data is divided into corresponding segments, and each segment is marked. Signal processing methods are used to extract the spectral features of the EEG signals and the functional connection features between brain regions from the processed EEG data. Statistical analysis is performed on the extracted feature indicators respectively, calculating the mean, median, and standard deviation. According to the statistical analysis results, the feature indicators after statistical analysis processing are used as independent variables, and the scores of the players on the psychological dimension are used as dependent variables. The training set data is used to train the mental health assessment model based on linear regression to predict the scores of the players on specific psychological dimensions.

[0085] Specifically, import the collected EEG signal data into data analysis software for preprocessing operations, including removing electrooculogram, motion artifacts, and filtering. According to different types and task requirements of the game, the data is divided into corresponding segments, and each segment is marked. The marked content includes game type, task type, segment start time, end time, and key events included in the segment. Perform a fast Fourier transform on each marked data segment to convert the EEG signals in the time domain into frequency domain signals, obtaining the amplitude and phase information of different frequency components, calculating the power spectral density of each frequency band, and performing statistical analysis on the power spectral features of each frequency band, including calculating the mean, variance, peak frequency statistic, and using various functional connection analysis methods to calculate the functional connection strength between each pair of electrodes. For each data segment, construct a functional connection matrix for analysis to obtain the functional network features between different brain regions. Statistical analysis is performed on the extracted feature indicators respectively, calculating the mean, median, and standard deviation. For example, calculate the average and standard deviation of the power spectral mean of a certain frequency band at each game stage of the player to evaluate the spectral feature level of the group at this stage. According to the statistical analysis results, the feature indicators after statistical analysis processing are used as independent variables, and the scores of the players on the psychological dimension are used as dependent variables. The training set data is used to train the mental health assessment model based on linear regression to predict the scores of the players on specific psychological dimensions.

[0086] In one embodiment of the present invention, a mental health education measurement and problem screening report is automatically generated in combination with corresponding evaluation values, and feedback is given after the player finishes the game, including the following steps:

[0087] The emotional stability evaluation value, cognitive ability evaluation value, social skill evaluation value obtained from games based on different psychological dimensions and the score results predicted by the model are added together to obtain the scores on different psychological dimensions, and a mental health assessment report is generated for the player. The names of each dimension, the score values, and the corresponding explanatory notes are listed in the player's mental health assessment report;

[0088] Based on the score situation of the player on each psychological dimension, combined with the preset norm reference, analyze the psychological problems existing in the player, and according to the discovered psychological problems, provide personalized suggestions and guidance for the player, review the generated mental health assessment report, and deliver the reviewed mental health assessment report to the player in an appropriate manner.

[0089] Specifically, obtain the emotional stability evaluation value, cognitive ability evaluation value, and social skill evaluation value of the player from games based on different psychological dimensions, input the relevant features of the player's EEG data in the game into a pre-constructed model to obtain the scores of specific psychological dimensions predicted by the model. After standardizing the evaluation values and the model-predicted scores, add the evaluation values and the model-predicted scores on different psychological dimensions to obtain the comprehensive scores on each psychological dimension, and generate a mental health assessment report for the player. The names of each dimension, the score values, and the corresponding explanatory notes of the player are listed in the report; Based on the score situation of the player on each psychological dimension, combined with the preset norm reference, through large-scale sampling surveys or research, select a representative sample group, such as players of different ages, genders, and regions, and conduct evaluations on them on each psychological dimension to obtain the score distribution of each dimension, reflecting the general level and distribution range of normal players on each psychological dimension. According to the norm data, calculate statistical parameters such as the mean and standard deviation. For example, the mean of a certain psychological dimension is 3 and the standard deviation is 0.5, which means that the scores of most normal people are concentrated between 2.5 and 3.5. Compare the score of the player on a certain psychological dimension with the preset norm reference. If the player's score is lower than the norm mean by more than a certain standard deviation, such as lower than the mean by 1 standard deviation, it may indicate that the player's performance on this dimension deviates from the level of the normal population and there are potential psychological risks. Determine the possible types of psychological problems based on the score situation of the player on each dimension and the comparison results with the norm reference, and provide personalized suggestions and guidance for the player.

[0090] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A mental health screening method based on immersive game tasks and EEG analysis, characterized in that: The following steps are involved: S1: Determine the mental health dimensions that need to be evaluated, design the corresponding immersive VR game scenes, select high-precision head-mounted EEG acquisition equipment, and integrate the head-mounted EEG acquisition equipment with the VR equipment; S2: Extract the corresponding characteristic data of players from the game data of different psychological dimensions of VR. In emotion regulation games, the emotional stability evaluation value can be constructed by monitoring the players' various physiological indicators. In cognitive challenge games, the cognitive ability evaluation value can be constructed by recording the players' concentration time and memory accuracy data. In social interaction games, the communication frequency and tone recognition data can be analyzed to construct the social skill evaluation value. S3: After preprocessing the collected EEG signals, characteristic indicators related to mental health are extracted, and a mental health assessment model based on linear regression is constructed to output score results in different psychological dimensions. Combined with the corresponding evaluation values, a mental health education measurement and problem screening report is automatically generated, and feedback is given after the player finishes the game.

2. A mental health screening method based on immersive game tasks and EEG analysis according to claim 1, characterized in that: The step S1 comprises the following steps: According to psychological theory and clinical research, the dimensions of mental health include emotional health, cognitive health, and social health. Based on the three dimensions of mental health, corresponding emotion regulation games, cognitive challenge games, and social interaction games are designed. During the game, players need to wear VR devices to collect corresponding game event data in the three categories of games; determine the head-mounted EEG device, and before the game starts, introduce the test process and precautions to the players, turn on the head-mounted EEG device, calibrate the equipment, and record the EEG data when the players are ready.

3. A mental health screening method based on immersive game tasks and EEG analysis according to claim 1, characterized in that: The integration of the head-mounted EEG acquisition device and the VR device includes the following steps: When the system is initialized, a clock source is created as the time reference for the entire system. When the VR game starts running, the current timestamp is recorded and sent to the EEG acquisition device as the starting time point for its data collection. When a key event occurs during the running of the VR game, the corresponding timestamp is obtained, and the type and description information of the key event are recorded. The EEG acquisition device continuously collects EEG signals and marks each sampling point with a timestamp, and stores the VR game data and EEG data with timestamps in the database respectively.

4. The mental health screening method based on immersive game tasks and EEG analysis according to claim 1, characterized in that: In the emotion regulation game, the emotion stability evaluation value can be constructed by monitoring the player's various physiological indicators, including the following steps: For emotion regulation games, the game can monitor the player's physiological indicators, including heart rate changes, respiratory rate and blood pressure changes, and record the data of the corresponding physiological indicators. Through the heart rate sensor, respiratory rate sensor and blood pressure sensor, the player's heart rate value, respiratory rate value, systolic blood pressure and diastolic blood pressure value can be obtained in real time, and the multiple values ​​can be combined to obtain the emotional stability evaluation value; Emotional stability evaluation value calculation formula: Among them, E stability represents the emotional stability evaluation value, H0 represents the initial heart rate, H t represents the current heart rate, B0 represents the initial breathing rate, B t Indicates the current respiratory rate, P s represents the initial systolic blood pressure, P sys-t Indicates the current systolic blood pressure, P d represents the initial diastolic pressure, P dia-t represents the current diastolic pressure, and ω1, ω2, and ω3 represent the corresponding weight coefficients.

5. The mental health screening method based on immersive game tasks and EEG analysis according to claim 1, characterized in that: The method of constructing a cognitive ability evaluation value by recording the player's attention concentration duration and memory accuracy data in a cognitive challenge game includes the following steps: Multiple electrodes are placed on the player's scalp according to specific standard positions to collect the player's EEG signals during the game, and preprocess them. According to a preset time window, the preprocessed EEG signals are fast Fourier transformed, and the power values ​​of α waves, θ waves and β waves are calculated. Then, combined with the player's behavioral performance records, the attention state of each time period is annotated, and a training set for training is constructed. The training set is classified according to the attention state label, and the classified training data is trained using a machine learning algorithm to establish a player's attention state classification model. The player's EEG data during the game and the corresponding attention state label are input into the model as training data. The model outputs the attention state prediction result. By analyzing the continuous EEG data, the player's concentration time at different game stages is determined. When the model predicts that the player is in a focused state, the timing starts, and the time until the predicted result changes to a non-focused state is used as the player's concentration time at that game stage. A recording interface is set up in the game. After the game is over, the player enters the words he remembers in the text box one by one and records the content of the words. The game program compares it with the content of the original text and calculates the player's memory accuracy. The cognitive ability evaluation value is obtained by weighting and adding the ratio of the player's concentration time to the total concentration time in this game stage and the player's memory accuracy rate; Cognitive ability evaluation value calculation formula: Among them, C ability represents the cognitive ability evaluation value, T focus Indicates the player's concentration time at this stage of the game, T all Indicates the total time the player spends in this game phase, M accuracy represents the player’s memory accuracy, ω4 and ω5 represent the corresponding weight coefficients.

6. The mental health screening method based on immersive game tasks and EEG analysis according to claim 1, characterized in that: The method of analyzing the communication frequency and tone recognition data in the social interaction game to construct a social skill evaluation value includes the following steps: Set a fixed observation time window. Within this time window, count the number of effective communications between players. Effective communication is defined as sending meaningful text messages, voice messages, or performing specific game-related activities. Calculate the average communication frequency of multiple time windows. Add the number of communications in each time window and divide it by the number of time windows to get the average communication frequency of players. Use machine learning algorithms to train a tone classification model: collect chat record data from social interaction games, determine the rules for annotating positive and negative tones, annotate the same chat record using multiple annotations, and preprocess the collected chat record data, including cleaning and using the bag-of-words model and term frequency-inverse document frequency to convert the data into feature vectors based on word frequency. Use the training set to train the machine learning algorithm based on the long short-term memory network architecture, deploy the trained model into the game, and while the game is in progress, obtain the player's chat records and input them into the trained model. The model outputs the probability that the message belongs to different tone categories. According to the probability output by the model, the probability is converted into the corresponding tone value according to the rules, and the different tone values ​​are weighted and summed according to the set weights to obtain the comprehensive tone value. The calculation formula of comprehensive tone value is: WITH moods =(1×G positive )×ω6+(-1×G negative )×ω7 Among them, Z moods Indicates the comprehensive tone value, G positive represents the probability of positive tone, G negative represents the probability of negative tone, ω6 and ω7 represent the corresponding weight coefficients; The average communication frequency and comprehensive tone value of the players are weighted and added together to get the social skill evaluation value; Social skill evaluation value calculation formula: S ability =F avg ×ω8+Z moods ×ω9 Among them, S ability represents the social skill evaluation value, F avg represents the average communication frequency of players, Z moods represents the comprehensive tone value, ω8 and ω9 represent the corresponding weight coefficients.

7. The mental health screening method based on immersive game tasks and EEG analysis according to claim 1, characterized in that: The method of extracting characteristic indicators related to mental health after preprocessing the collected EEG signals, and constructing a mental health evaluation model based on linear regression to output score results on different psychological dimensions includes the following steps: The collected EEG signal data is imported into the data analysis software for preprocessing. According to the different types of games and task requirements, the data is divided into corresponding segments, and each segment is marked. The signal processing method is used to extract the spectral characteristics of the EEG signal and the functional connection characteristics between brain regions from the processed EEG data. The extracted characteristic indicators are statistically analyzed, and the mean, median and standard deviation are calculated. According to the statistical analysis results, the characteristic indicators after statistical analysis are used as independent variables, and the player's score on the psychological dimension is used as the dependent variable. The training set data is used to train the mental health assessment model based on linear regression to predict the player's score on a specific psychological dimension.

8. The mental health screening method based on immersive game tasks and EEG analysis according to claim 1, characterized in that: The method of automatically generating a mental health education measurement and problem screening report in combination with corresponding evaluation values ​​and providing feedback to the player after the player finishes the game includes the following steps: The scores of different psychological dimensions are obtained by adding up the emotional stability evaluation values, cognitive ability evaluation values, social skills evaluation values ​​and model prediction scores obtained from the game, and a mental health evaluation report is generated for the player. The name, score value and corresponding explanation of each dimension are listed in the player's mental health evaluation report; Based on the player's scores on various psychological dimensions and combined with the preset normative reference, we analyze the player's psychological problems, provide personalized advice and guidance to the player based on the psychological problems found, review the generated mental health assessment report, and deliver the reviewed mental health assessment report to the player in an appropriate manner.

Images