Psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis

Abstract

The invention provides a psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis. The psychological intervention system comprises a multi-modal data acquisition module, an emotion quantitative analysis module, a virtual scene generation module and an intervention effect feedback module. Through a closed-loop mechanism of real-time monitoring, dynamic modeling, scene intervention and effect feedback, digitization, individuation and scientization of mental health intervention are realized, and core pain points of a traditional method in efficiency, compliance and accuracy are solved.

Description

Technical Field

[0001] The present application relates to the field of virtual reality technology and mental health, and in particular to a psychological intervention system based on quantitative analysis of virtual reality cognitive behavior and emotions. Background Art

[0002] In modern society, people face a variety of stressors, such as work pressure, daily chores, and interpersonal relationships. These stressors can easily lead to negative emotions such as anxiety, depression, and anger. Being in a negative emotional state for a long time not only affects a person's mental health but can also lead to a range of physical illnesses. Traditional methods of emotion regulation, such as psychological counseling, meditation, and exercise, while widely adopted, each have significant limitations. For example, psychological counseling often requires the establishment of a deep trusting relationship and takes a long time to be effective; meditation relies on high self-discipline and long-term practice and is easily disturbed by the external environment; and exercise requires physical support and a continuous investment of time. For this reason, these methods may not achieve the desired results in practice for some people.

[0003] Virtual reality (VR), an emerging technology, can effectively overcome the limitations of traditional emotion regulation methods, such as high self-discipline, long-term commitment, and inconsistent results, by creating highly immersive and personalized virtual scenes. It can not only simulate various emotional regulation methods, such as psychological counseling, meditation, and exercise, but also provide users with a more efficient, flexible, and engaging approach to emotional management through real-time feedback and interactive experiences. Summary of the Invention

[0004] In order to solve at least one of the technical problems in the background technology, the embodiment of the present application provides a psychological intervention system based on virtual reality cognitive behavior emotion quantification analysis, including a multimodal data acquisition module, an emotion quantification analysis module, a virtual scene generation module and an intervention effect feedback module;

[0005] The multimodal data acquisition module is used to collect the user's physiological signals and behavioral data;

[0006] The physiological signals include heart rate data HR(t), skin electrical data GSR(t), and brain wave data EEG(t), where t represents time;

[0007] The behavioral data includes action frequency and questionnaire responses;

[0008] The emotion quantification analysis module includes physiological data normalization processing and multi-dimensional emotion index calculation;

[0009] The physiological data normalization processing includes normalizing each physiological signal;

[0010] The normalized heart rate data is:

[0011]

[0012] The normalized skin electrical data is:

[0013]

[0014] The normalized EEG data is:

[0015]

[0016] Among them, HR max , HR min are the minimum and maximum values ​​of the heart rate data, GSR max , GSR min are the minimum and maximum values ​​of skin electrical data, EEG max , EEG min are the minimum and maximum values ​​of the EEG data respectively;

[0017] The emotional quantification index is obtained by weighted summation:

[0018] EI(t)=α×HR n (t)+β×GSR n (t)+γ×EEG n (t)

[0019] Among them, α, β, γ are weight coefficients, 且 α+β+ γ =1;

[0020] The multi-dimensional emotion index calculation includes calculating the anxiety tendency score AS, calculating the depression tendency score DS and calculating the mania tendency score MS;

[0021] The virtual scene generation module includes dynamically constructing an immersive scene based on emotion quantification indicators;

[0022] The intervention effect feedback module includes calculating the improvement rate through scale retesting and changes in physiological indicators.

[0023] In one possible embodiment, the psychological intervention system based on virtual reality cognitive behavioral emotion quantitative analysis provided by the present application, the calculation of the anxiety tendency score AS includes:

[0024] The anxiety tendency assessment questionnaire is used. The questionnaire contains m questions, and each question is assigned an initial weight according to its importance in judging the degree of anxiety. and Assume that the user's answer score to question i is a i, the answer options are divided into k levels; calculate the adjustment weight w for each question i , the calculation of weight adjustment takes into account the correlation between problems; let the correlation coefficient matrix of problem i and other problems be R ij (j=1, 2, ..., m), where R ij Indicates the degree of relevance between question i and question j. Its value range is [-1, 1]. The adjustment weight is calculated using the following formula:

[0025]

[0026] Calculate the anxiety tendency score:

[0027]

[0028] In one possible embodiment, the psychological intervention system based on virtual reality cognitive behavioral emotion quantitative analysis provided by the present application includes calculating the anxiety tendency score to ensure a more accurate judgment effect, thereby predicting the user's future development status and achieving positive intervention;

[0029] For a given training data set (x1, y1), (x2, y2), (x3, y3), ..., (x n ,y n );

[0030] where x i is the input feature, y i is the corresponding target value; θ0 is the baseline anxiety value; θ1 is the anxiety sensitivity; by continuously adjusting these two parameters, the model can better fit the relationship between input and output, thereby providing support for the prediction or evaluation of anxiety symptoms;

[0031] Establish the fitting straight line equation of the regression model: h θ (x) = θ0 + θ1x, find the appropriate parameters θ0 and θ1 so that the predicted value h θ (x) = θ0 + θ1 The error between x and the true value y is the smallest;

[0032] The optimal value J(θ) of θ0 and θ1 is obtained by the following calculation formula:

[0033]

[0034] The optimal θ value is solved by taking the partial derivative of the loss function and setting it to 0. The partial derivative calculation formula is as follows:

[0035]

[0036] In one possible embodiment, the psychological intervention system based on virtual reality cognitive behavioral emotion quantitative analysis provided by the present application, the calculation of the depression tendency score DS includes:

[0037] The depression tendency assessment questionnaire is used. The questionnaire has n questions. Let the severity weight of the jth question be u j , whose value range is from 0 to 1, and The score of the user's answer to question j is d j , the answer options are divided into l levels; calculate the weighted answer score s for each question j =d j ×u j The weighted scores of the questions are summed to get the original total score representing the depressive tendency

[0038] Introducing nonlinear adjustment factors The original total score is standardized; the final depression tendency score is obtained: DS = f(x); x is substituted into the nonlinear adjustment factor f(x) to perform nonlinear adjustment on the comprehensive score.

[0039] In one possible embodiment, the psychological intervention system based on virtual reality cognitive behavioral emotion quantitative analysis provided by the present application, the calculation of the mania tendency score MS includes:

[0040] The manic tendency assessment questionnaire is used. The questionnaire has p questions, set p = 20; set the activity weight of the qth question to be v q , with a value range of 0 to 1, and The score of the user's answer to question q is e q , the answer options are divided into r levels;

[0041] Calculate the weighted activity score t for each question q =e q ×v q ;

[0042] Calculating comprehensive activity indicators

[0043] Manic mood tendency score MS = g(y), introducing correction factor

[0044] In one possible implementation, the psychological intervention system based on virtual reality cognitive behavioral emotion quantification analysis provided by the present application is calculated by minimizing a loss function in order to make the predicted value close to the real emotion score;

[0045] Calculate the total original activity score S = ∑a j ·u j ;

[0046] Among them, a j is the weight coefficient of the jth question in the assessment of manic tendency, u j is the score of the user’s answer to the jth question;

[0047] Calculate model predictions where θ0 is the baseline mania value.

[0048] In one possible implementation, the psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis provided by the present application is to achieve the predicted value. A supervised learning framework is constructed for the progressive approximation of the true manic tendency score M, and parameter optimization is performed by minimizing the regularized mean squared error loss function.

[0049] Original activity score By the dynamic weight coefficient a j and the discretized answer score u j The linear combination of

[0050] Prediction Model In the equation, θ0 represents the baseline manic tendency, and θ1 quantifies the marginal effect of activity on the predicted value;

[0051] The true value is corrected nonlinearly To accommodate the diminishing nature of manic symptoms;

[0052] Loss Function The gradient descent method is used to optimize the model to ensure the generalization performance of the model under limited samples.

[0053] In one possible embodiment, the psychological intervention system based on virtual reality cognitive behavior emotion quantification analysis provided by the present application includes generating a suitable virtual reality scene based on the emotion quantification index EI(t) obtained by the emotion monitoring module and information in the user feature database;

[0054] Assuming the immersion of the virtual reality scene is I, the scene complexity is C, and the matching degree with the user's emotional tendency is M, the scene generation function can be expressed as:

[0055] VRScene(EI(t),A,S,AS,DS,MS,···)=f(I,C,M,…);

[0056] The mathematical model of immersion I is: I = δ × ER + ∈ × CC + ζ × SS;

[0057] Among them, the richness of scene elements is ER, the coordination of scene colors is CC, the three-dimensional sense of sound is SS, and δ,∈,ζ are the corresponding weight coefficients;

[0058] The mathematical model of scene element richness ER is: ER = λ × N + μ × K;

[0059] Among them, the number of objects in the scene is N, the type of objects is K, and λ and μ are weight coefficients.

[0060] In one possible implementation, the psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis provided by the present application, the intervention effect feedback module includes:

[0061] Calculation formula and parameters: Use the formula E = C × (I1-I2);

[0062] Where E is the emotion improvement score, C is the effectiveness coefficient of the intervention, I1 is the emotion intensity before the intervention, and I2 is the emotion intensity after the intervention;

[0063] The formula for calculating mood improvement is:

[0064] The beneficial effects of this application: Through the closed-loop mechanism of "real-time monitoring-dynamic modeling-scenario intervention-effect feedback", the digitalization, personalization and scientificization of mental health intervention are realized, and the core pain points of traditional methods in efficiency, compliance and accuracy are solved. By integrating multimodal physiological data (heart rate, skin electricity, brain waves) with psychological assessment questionnaires, multi-dimensional emotional quantification indicators (anxiety tendency AS, depression tendency DS, mania tendency MS) are constructed. Compared with traditional static questionnaires or single physiological indicator monitoring, it can capture emotional change trends in real time, and achieve precise intervention by dynamically adjusting virtual scene parameters, breaking through the limitations of traditional methods with strong lag. It can also design differentiated scenario intervention strategies for different emotional characteristics, without drug side effects, and avoid the risk of privacy leakage in traditional psychological counseling. DETAILED DESCRIPTION

[0065] The embodiment of the present application provides a psychological intervention system based on virtual reality cognitive behavior emotion quantification analysis, including a multimodal data acquisition module, an emotion quantification analysis module, a virtual scene generation module and an intervention effect feedback module;

[0066] The multimodal data acquisition module is used to collect the user's physiological signals and behavioral data;

[0067] Physiological signals include heart rate data HR(t), skin electrical data GSR(t), and brain wave data EEG(t), where t represents time;

[0068] Behavioral data include action frequency and questionnaire responses;

[0069] The emotion quantification analysis module includes physiological data normalization processing and multi-dimensional emotion index calculation;

[0070] Physiological data normalization processing includes normalization processing of each physiological signal;

[0071] The normalized heart rate data is:

[0072]

[0073] The normalized skin electrical data is:

[0074]

[0075] The normalized EEG data is:

[0076]

[0077] Among them, HR max , HR min are the minimum and maximum values ​​of the heart rate data, GSR max , GSR min are the minimum and maximum values ​​of skin electrical data, EEG max , EEG min are the minimum and maximum values ​​of the EEG data respectively;

[0078] The emotional quantification index is obtained by weighted summation:

[0079] EI(t)=α×HR n (t)+β×GSR n (t)+γ×EEG n (t)

[0080] Where α, β, γ are weight coefficients, and α+β+γ=1;

[0081] The calculation of multidimensional emotional indicators includes the calculation of anxiety tendency score AS, the calculation of depression tendency score DS and the calculation of mania tendency score MS;

[0082] The virtual scene generation module includes dynamically building immersive scenes based on emotion quantification indicators;

[0083] The intervention effect feedback module includes calculating the improvement rate through scale retesting and changes in physiological indicators.

[0084] Specifically, the calculation of the anxiety tendency score AS includes:

[0085] The anxiety tendency assessment questionnaire is used. The questionnaire contains m questions, and each question is assigned an initial weight according to its importance in judging the degree of anxiety. and Assume that the user's answer score to question i is a i , the answer options are divided into k levels; calculate the adjustment weight w for each questioni , the calculation of weight adjustment takes into account the correlation between problems; let the correlation coefficient matrix of problem i and other problems be R ij (j=1, 2, ..., m), where R ij Indicates the degree of relevance between question i and question j. Its value range is [-1, 1]. The adjustment weight is calculated using the following formula:

[0086]

[0087] Calculate the anxiety tendency score:

[0088]

[0089] Specifically, the calculation of the anxiety tendency score includes predicting the user's future development status in order to ensure a more accurate judgment effect, thereby achieving positive intervention;

[0090] For a given training data set (x1, y1), (x2, y2), (x3, y3), ..., (x n ,y n );

[0091] where x i is the input feature, y i is the corresponding target value; θ0 is the baseline anxiety value; θ1 is the anxiety sensitivity; by continuously adjusting these two parameters, the model can better fit the relationship between input and output, thereby providing support for the prediction or evaluation of anxiety symptoms;

[0092] Establish the fitting straight line equation of the regression model: h θ (x) = θ0 + θ1x, find the appropriate parameters θ0 and θ1 so that the predicted value h θ (x) = θ0 + θ1 The error between x and the true value y is the smallest;

[0093] The optimal value J(θ) of θ0 and θ1 is obtained by the following calculation formula:

[0094]

[0095] The optimal θ value is solved by taking the partial derivative of the loss function and setting it to 0. The partial derivative calculation formula is as follows:

[0096]

[0097] Specifically, calculating the depression tendency score DS includes:

[0098] The depression tendency assessment questionnaire is used. The questionnaire has n questions. Let the severity weight of the jth question be u j , whose value range is from 0 to 1, and The score of the user's answer to question j is d j , the answer options are divided into l levels; calculate the weighted answer score s for each question j =d j ×u j The weighted scores of the questions are summed to get the original total score representing the depressive tendency

[0099] Introducing nonlinear adjustment factors The original total score is standardized; the final depression tendency score is obtained: DS = f(x); x is substituted into the nonlinear adjustment factor f(x) to perform nonlinear adjustment on the comprehensive score.

[0100] Specifically, calculation of the mania tendency score (MS) includes:

[0101] The manic tendency assessment questionnaire is used. The questionnaire has p questions, set p = 20; set the activity weight of the qth question to be v q , with a value range of 0 to 1, and The score of the user's answer to question q is e q , the answer options are divided into r levels;

[0102] Calculate the weighted activity score t for each question q =e q × vq ;

[0103] Calculating comprehensive activity indicators

[0104] Manic mood tendency score MS = g(y), introducing correction factor

[0105] Specifically, in order to make the predicted value close to the true sentiment score, it is calculated by minimizing the loss function;

[0106] Calculate the total original activity score S = ∑a j ·u j ;

[0107] Among them, a j is the weight coefficient of the jth question in the assessment of manic tendency, u j is the score of the user’s answer to the jth question;

[0108] Calculate model predictions where θ0 is the baseline mania value.

[0109] Specifically, to achieve the predicted value Score for true manic tendency M iA supervised learning framework is constructed by gradually approximating the regularized mean square error loss function to complete parameter optimization.

[0110] Original activity score By the dynamic weight coefficient a j and the discretized answer score u j The linear combination of

[0111] Prediction Model In the equation, θ0 represents the baseline manic tendency, and θ1 quantifies the marginal effect of activity on the predicted value;

[0112] The true value is corrected nonlinearly To accommodate the diminishing nature of manic symptoms;

[0113] Loss Function The gradient descent method is used to optimize the model to ensure the generalization performance of the model under limited samples.

[0114] Specifically, the virtual scene generation module includes generating a suitable virtual reality scene based on the emotion quantification index EI(t) obtained by the emotion monitoring module and the information in the user feature database;

[0115] Assuming the immersion of the virtual reality scene is I, the scene complexity is C, and the matching degree with the user's emotional tendency is M, the scene generation function can be expressed as:

[0116] VRScene(EI(t),A,S,AS,DS,MS,···)=f(I,C,M,…);

[0117] The mathematical model of immersion I is: I = δ × ER + ∈ × CC + ζ × SS;

[0118] Among them, the richness of scene elements is ER, the coordination of scene colors is CC, the three-dimensional sense of sound is SS, and δ,∈,ζ are the corresponding weight coefficients;

[0119] The mathematical model of scene element richness ER is: ER = λ × N + μ × K;

[0120] Among them, the number of objects in the scene is N, the type of objects is K, and λ and μ are weight coefficients.

[0121] Specifically, the intervention effect feedback module includes:

[0122] Calculation formula and parameters: Use the formula E = C × (I1-I2);

[0123] Where E is the emotion improvement score, C is the effectiveness coefficient of the intervention, I1 is the emotion intensity before the intervention, and I2 is the emotion intensity after the intervention;

[0124] The formula for calculating mood improvement is:

[0125] The purpose of this invention is to provide a psychological intervention method based on supervised emotion recognition, which effectively helps users regulate their emotions, relieve stress and improve their mental health by constructing personalized virtual scenes and interactive experiences.

[0126] Step 1: Building the Emotion Monitoring Data Collection Module

[0127] This system collects basic user information, initial emotional data (e.g., emotional values ​​obtained through psychological scale tests), and physiological data (e.g., heart rate, galvanic skin response, etc.). It uses a variety of physiological sensors, such as heart rate sensors, galvanic skin sensors, and EEG sensors, to collect the user's physiological data in real time. Let HR(t) represent heart rate data (where t represents time), GSR(t) represent galvanic skin data, and EEG(t) represent EEG data.

[0128] First, normalize the body data.

[0129] Assume that the normalized heart rate data is:

[0130] The skin electrical data is normalized as follows:

[0131] After normalization of EEG data

[0132] Among them, HR max , HR min are the minimum and maximum values ​​of the heart rate data, GSR max , GSR min are the minimum and maximum values ​​of skin electrical data, EEG max , EEG min are the minimum and maximum values ​​of the EEG data, respectively).

[0133] Then the weighted summation method is used to obtain the sentiment quantification index:

[0134] EI(t)=α×HR n (t)+β×GSR n (t)+ γ ×EEG n (t), where α, β, and γ are weight coefficients determined based on a large amount of experimental data, and α+β+γ=1

[0135] Step 2: User feature data construction module construction

[0136] Stores the user's basic information, past emotion regulation history data, personality characteristics data, etc. For example, the user's age A, gender S, anxiety tendency score AS, depression tendency score DS and other information.

[0137] Calculation of the anxiety tendency score AS:

[0138] A specially designed anxiety tendency assessment questionnaire is used, which contains m questions, for example, m = 20. Each question is assigned an initial weight based on its importance in judging the level of anxiety. and

[0139] Assume that the user's answer score to question i is a i , the answer options are divided into k levels, for example, k = 5, corresponding to scores from 0 to 4 (0 means completely inconsistent, 1 means slightly consistent, 2 means partially consistent, 3 means relatively consistent, and 4 means completely consistent).

[0140] Then calculate the adjusted weight w for each question i , the calculation of weight adjustment takes into account the correlation between problems. Let the correlation coefficient matrix between problem i and other problems be R ij (j=1, 2, ..., m), where R ij Indicates the degree of relevance between question i and question j, and its value range is [-1, 1]. The adjustment weight is calculated using the following formula:

[0141] Finally, calculate the anxiety tendency score

[0142] At the same time, in order to ensure a more accurate judgment effect, it is planned to predict the future development status of the user, so as to achieve positive intervention (calculation and prediction intervention are performed in each part).

[0143] For a given training data set (x1, y1), (x2, y2), (x3, y3), ..., (x n ,y n ),

[0144] where x i is the input feature (such as anxiety tendency score), y i is the corresponding target value (anxiety symptom rating). The goal is to find appropriate parameters θ0 and θ1. In this linear model, θ0 is called the "baseline anxiety value"; θ1 is called the "anxiety sensitivity." By continuously adjusting these two parameters, the model can better fit the relationship between input and output, thereby supporting the prediction or assessment of anxiety symptoms.

[0145] Establish the fitting straight line equation of the regression model: h θ(x) = θ0 + θ1x, the goal is to find the appropriate parameters θ0 and θ1 so that the predicted value (future emotion change) h θ The error between (x)=θ0+θ1x and the true value y (anxiety, depression, mania) is the smallest.

[0146] The optimal value J(θ) of θ0 and θ1 can be obtained by the following calculation formula:

[0147]

[0148] The optimal θ value is solved by taking the partial derivative of the loss function and setting it to 0. The partial derivative calculation formula is as follows:

[0149]

[0150] Calculation of depression tendency score DS:

[0151] The depression tendency assessment questionnaire is used. The questionnaire has n questions, for example, n = 20. Let the severity weight of the jth question be u j , whose value range is from 0 to 1, and

[0152] The score of the user's answer to question j is d j , the answer options are divided into l levels, for example l = 4, corresponding to scores from 0 to 3 (0 means no such symptom, 1 means mild symptoms, 2 means moderate symptoms, and 3 means severe symptoms).

[0153] First calculate the weighted answer score s for each question j =d j ×u j ,

[0154] The weighted scores of all questions are summed to get the original total score representing depression tendency:

[0155] Then, by introducing the nonlinear adjustment factor The original total score was standardized.

[0156] Finally, the final depression tendency score is obtained: DS = f(x).

[0157] Substituting x into the nonlinear adjustment factor f(x) causes a nonlinear adjustment to be made to the composite score. For example, if the value of x is very small (indicating a low score for depression symptoms in the user's questionnaire answers), the value of f(x) will approach 0, indicating a low tendency toward depression. If the value of x is very large (indicating a high score for depression symptoms in the user's questionnaire answers), the value of f(x) will approach 1, indicating a high tendency toward depression.

[0158] Calculation of the Manic Mood Tendency Score (MS):

[0159] The manic tendency assessment questionnaire is used. The questionnaire has p questions, and p is set to 20. The activity weight of the qth question is v q , with a value range of 0 to 1, and The score of the user's answer to question q is e q , the answer options are divided into r levels, for example, r = 3, corresponding to scores from 0 to 2 (0 means no such performance, 1 means mild performance, 2 means obvious performance). Calculate the weighted activity score t for each question q =e q ×v q , and then calculate the comprehensive activity index

[0160] Manic mood tendency score MS = g(y), introducing correction factor (The square root function is used as the correction factor here because manic mood is related to overactivity to a certain extent, and as activity increases, the growth rate of its impact on manic tendencies gradually slows down. The square root function can better simulate this relationship).

[0161] In order to make the predicted value close to the real emotion score, we calculate it by minimizing the loss function. First, we calculate the original activity score: S = ∑a j ·u j , where a j is the weight coefficient of the jth question in the assessment of manic tendency, u j is the user's answer score to the jth question; then calculate the model prediction value True value, where θ0 is the baseline mania value, that is, when the original activity score S i =0, the manic tendency score predicted by the system; is the manic activity sensitivity, that is, the original activity total score S i For every 1 unit increase in the predicted manic tendency score The change in . Finally, the following is the calculation formula of the loss function:

[0162] To achieve the predicted value Score for true manic tendency M i This study constructs a supervised learning framework to optimize parameters by minimizing the regularized mean square error loss function. By the dynamic weight coefficient a j and the discretized answer score u j The linear combination of In the equation, θ0 represents the baseline manic tendency, and θ1 quantifies the marginal effect of activity on the predicted value. The true value is corrected by nonlinearity. To adapt to the marginal decreasing characteristics of manic symptoms. Loss function The gradient descent method is used to optimize the model to ensure the generalization performance of the model under limited samples.

[0163] Explanation of the specificity and universality of the calculation method of sentiment quantification indicators:

[0164] To accurately assess anxiety, depression, and mania, the core algorithm of this invention designs specific calculation methods for different emotional characteristics:

[0165] 1. Anxiety tendency score:

[0166] Use weight adjustment based on the question relevance matrix (formula: ), to reflect the multidimensional correlates of anxiety symptoms.

[0167] General part: Questionnaire weighted summation framework It shares basic logic with depression and mania.

[0168] 2. Depression tendency score

[0169] Through the Sigmoid function The raw total scores were nonlinearly normalized to accommodate the threshold effect of depressive symptoms.

[0170] General part: Normalization of physiological data (such as ) can be uniformly applied to all sentiment indicators.

[0171] 3. Manic tendency score

[0172] Introducing a square root correction factor To simulate the nonlinear growth characteristics of manic activity.

[0173] Common part: Virtual scene generation rules (such as threshold judgment, color / sound parameter design) can reuse the same framework.

[0174] Design principles:

[0175] Specificity: The core algorithm needs to remain independent to ensure accurate modeling of emotional characteristics (such as correlation weights for anxiety and sigmoid mapping for depression).

[0176] Universality: Basic modules such as questionnaire weighted summation, data normalization, and scenario generation logic can be designed as universal processes to improve system efficiency.

[0177] Virtual reality scene intervention module:

[0178] Based on the emotion quantification index EI(t) obtained by the emotion monitoring module and the information in the user feature database, a suitable virtual reality scene is generated. Assuming the immersion of the virtual reality scene is I, the scene complexity is C, and the matching degree with the user's emotional tendency is M, the scene generation function can be expressed as: VRScene(EI(t), A, S, AS, DS, MS, ···) = f(I, C, M, ···)

[0179] Taking immersion I as an example, assuming that immersion is related to the richness ER of scene elements, the coordination CC of scene colors, and the stereoscopic sense SS of sound, a mathematical model can be established: I = δ × ER + ∈ × CC + ζ × SS (where δ, ∈, and ζ are corresponding weight coefficients).

[0180] The scene element richness ER is related to factors such as the number of objects N and the types of objects K in the scene, for example, ER = λ×N+μ×K (λ and μ are weight coefficients).

[0181] The specific rules for constructing virtual scenarios based on different emotion indicators are as follows:

[0182] When EI(t) indicates that the user is in a state of high excitement or mania (e.g. MS is high, exceeding the set threshold MS th )hour:

[0183] Create a tranquil, soothing natural scene, such as a seaside night. Use dark blue and black as dominant colors, and the sounds of gentle waves and a soft breeze. Keep the scene small and relatively static, such as a few distant islands or sparse stars in the sky, to reduce visual stimulation and psychological activity.

[0184] When EI(t) shows that the user is in an anxious state (AS is high, exceeding the set threshold AS th ) : Create a simple, orderly, and enclosed spatial scene, such as a simply furnished study. Use soft, warm colors, such as light beige or pale yellow. Use soothing classical music or the sound of natural rain. Objects in the scene should be neatly arranged, such as neatly arranged books on a bookshelf or neatly arranged stationery on a desk, giving users a sense of stability and control, alleviating anxiety.

[0185] When EI(t) indicates that the user has a tendency to depression (DS is high, exceeding the set threshold DS th ) : Create a vibrant and energetic scene, such as a spring garden. The scene is vibrant and colorful, with expanses of green grass and vibrant flowers. The sounds are the cheerful chirping of birds and the gurgling of a stream. The scene features numerous dynamic elements, such as fluttering butterflies and swaying flowers, evoking the user's appreciation for beauty and enhancing emotional vitality.

[0186] 7) Feedback on the effectiveness of anxiety intervention

[0187] Calculation formula and parameters:

[0188] The formula E=C×(I1-I2) is used (where E is the mood improvement score, C is the effectiveness coefficient of the intervention measure, I1 is the mood intensity before the intervention, and I2 is the mood intensity after the intervention).

[0189] The formula for calculating mood improvement is:

[0190] Before the intervention, average anxiety scores were measured using the Self-Rating Anxiety Scale (SAS). (The total score on this scale is 100, with higher scores indicating higher anxiety levels.) Anxiety scores were measured again after the 8-week CBT intervention. Based on previous research and expert evaluations, the effectiveness coefficient of cognitive behavioral therapy for anxiety intervention has been established.

[0191] Calculation process and results:

[0192] Substituting into the formula, we can get E = 0.7 × (65-40) = 17.5, the mood improvement rate This result shows that anxiety was improved to a certain extent through psychological intervention, with a mood improvement score of 17.5 and an improvement rate of 38.4%. The higher mood improvement score suggests that the intervention may be effective.

[0193] Examples and clinical validation

[0194] 1. Actual operation process of the system

[0195] Step 1: Wearing the device and collecting data

[0196] Device configuration: The user wears a VR headset with integrated multimodal sensors (such as heart rate sensor, galvanic skin response electrodes) and a controller.

[0197] Data synchronization:

[0198] Physiological data: real-time collection of heart rate HR(t), skin conductance GSR(t), brain wave EEG(t) (sampling rate 256H z ).

[0199] Behavioral data: Track the frequency and amplitude of user movements (such as the number of arm swings per unit time) through the handle.

[0200] Questionnaire filling: Users complete standardized mood questionnaires (such as YMRS mania scale,

[0201] PHQ-9 Depression Scale), answer score u j Automatic entry system.

[0202] Step 2: Algorithm Analysis and Sentiment Modeling

[0203] Data preprocessing:

[0204] Physiological signal denoising: Wavelet transform is used to filter out high-frequency noise (such as muscle movement artifacts).

[0205] Normalization: Map the original physiological data to the [0, 1] interval (Formula: ).

[0206] Sentiment indicator calculation:

[0207] Manic Tendency Score: (Example: User A's S=12→M=3.46).

[0208] Real-time prediction value: (Parameters θ0=0.8, θ1=0.3 trained with historical data).

[0209] Step 3: Dynamic generation of virtual scenes

[0210] Threshold determination rules:

[0211] like Generate low-stimulation scenes (such as meditation in a snowy field, with low-frequency white noise as the ambient sound and color saturation reduced by 60%).

[0212] like Generate neutral scenes (such as a forest trail, with dynamic leaves falling at a frequency synchronized with the user's breathing rhythm).

[0213] like Generate high-energy scenes (such as a festival square, with a 40% increase in color contrast and faster background music tempo).

[0214] Real-time interactive feedback: Users select scene elements through the handle (such as clicking a butterfly to trigger a soothing animation), and the system records the interaction frequency and adjusts subsequent scene parameters.

[0215] Step 4: Feedback on intervention effects and iterative short-term feedback: After each intervention, the system displays the emotion score change curve (e.g., manic tendency drops from M=3.8 to 2.9).

[0216] Long-term tracking: Generate weekly mood reports and compare them to baseline data (such as a 15% increase in heart rate variability (HRV)).

[0217] 2. Experimental Design and Data Case

[0218] Experiment 1: Verification of the effect of intervention on manic tendency

[0219] Sample size: N=80 (experimental group n=40, control group n=40).

[0220] Control group setting:

[0221] Experimental group: Use this system for VR intervention (dynamic scene adaptation).

[0222] Control group: watched a static natural scenery video (no interaction or dynamic adjustment).

[0223] Intervention duration: 20 minutes per day for 4 weeks.

[0224] Assessment Tools:

[0225] Manic tendency: Young Mania Rating Scale (YMRS, total score 0-60).

[0226] Physiological indicators: RMSSD value of heart rate variability (HRV) (ms).

[0227] Experimental results:

[0228] Improvement in YMRS score:

[0229] Experimental group: baseline 28.5±4.2 → post-intervention 18.7±3.6 (improvement rate 34.4%, p<0.001).

[0230] Control group: baseline 27.8±4.0→post-intervention 25.1±4.1 (improvement rate 9.7%, p=0.12).

[0231] HRV improvement: RMSSD of the experimental group increased from 42.3±6.5ms to 53.8±7.2ms (p<0.01).

[0232] Experiment 2: Anxiety Intervention Case

[0233] User case: 32-year-old male, Self-Rating Anxiety Scale (SAS) baseline score of 68 (severe anxiety).

[0234] Intervention process:

[0235] Scene generation: closed study (color RGB = 255, 240, 220), background sound is rain (55dB).

[0236] Interactive task: Tidy up the virtual desk (each time a task is completed, the scene complexity is reduced by 10%).

[0237] Improvement effect:

[0238] After 2 weeks of intervention: SAS score dropped to 52 (improvement rate 23.5%).

[0239] After 4 weeks of intervention: SAS score was 44 (improvement rate 35.3%), and HRV increased from 38ms to 49ms.

[0240] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis, characterized by: It includes multimodal data acquisition module, emotion quantification analysis module, virtual scene generation module and intervention effect feedback module; The multimodal data acquisition module is used to collect the user's physiological signals and behavioral data; The physiological signals include heart rate data HR(t), skin electrical data GSR(t), and brain wave data EEG(t), where t represents time; The behavioral data includes action frequency and questionnaire responses; The emotion quantification analysis module includes physiological data normalization processing and multi-dimensional emotion index calculation; The physiological data normalization processing includes normalizing each physiological signal; The normalized heart rate data is: The normalized skin electrical data is: The normalized EEG data is: Among them, HR max , HR min are the minimum and maximum values ​​of the heart rate data, GSR max , GSR min are the minimum and maximum values ​​of skin electrical data, EEG max , EEG min are the minimum and maximum values ​​of the EEG data respectively; The emotional quantification index is obtained by weighted summation: EI(t)=α×HR n (t)+β×GSR n (t)+γ×EEG n (t) Where α, β, γ are weight coefficients, and α+β+γ=1; The multi-dimensional emotion index calculation includes calculating the anxiety tendency score AS, calculating the depression tendency score DS and calculating the mania tendency score MS; The virtual scene generation module includes dynamically constructing an immersive scene based on emotion quantification indicators; The intervention effect feedback module includes calculating the improvement rate through scale retesting and changes in physiological indicators.

2. The psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis according to claim 1 is characterized in that: The calculation of the anxiety tendency score AS includes: The anxiety tendency assessment questionnaire is used. The questionnaire contains m questions, and each question is assigned an initial weight according to its importance in judging the degree of anxiety. and Assume that the user's answer score to question i is a i , the answer options are divided into k levels; calculate the adjustment weight w for each question i , the calculation of weight adjustment takes into account the correlation between problems; let the correlation coefficient matrix of problem i and other problems be R ij (i=1, 2, ..., m), where R ij Indicates the degree of relevance between question i and question j. Its value range is [-1, 1]. The adjustment weight is calculated using the following formula: Calculate the anxiety tendency score:

3. The psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis according to claim 2 is characterized in that: The calculation of the anxiety tendency score includes predicting the user's future development status in order to ensure a more accurate judgment effect, thereby achieving positive intervention; For a given training data set (x1, y1), (x2, y2), (x3, y3), ..., (x n ,y n ); where x i is the input feature, y i is the corresponding target value; θ0 is the baseline anxiety value; θ1 is the anxiety sensitivity; by continuously adjusting these two parameters, the model can better fit the relationship between input and output, thereby providing support for the prediction or evaluation of anxiety symptoms; Establish the fitting straight line equation of the regression model: h θ (x) = θ0 + θ1x, find the appropriate parameters θ0 and θ1 so that the predicted value h θ (x) = θ0 + θ1 The error between x and the true value y is the smallest; The optimal value J(θ) of θ0 and θ1 is obtained by the following calculation formula: The optimal θ value is solved by taking the partial derivative of the loss function and setting it to 0. The partial derivative calculation formula is as follows:

4. The psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis according to claim 1 is characterized in that: The calculation of the depression tendency score DS includes: The depression tendency assessment questionnaire is used. The questionnaire has n questions. Let the severity weight of the jth question be u j , whose value range is from 0 to 1, and The score of the user's answer to question j is d j , the answer options are divided into l levels; calculate the weighted answer score s for each question j =d j ×u j The weighted scores of the questions are summed to get the original total score representing the depressive tendency Introducing nonlinear adjustment factors The original total score is standardized; the final depression tendency score is obtained: DS = f(x); x is substituted into the nonlinear adjustment factor f(x) to perform nonlinear adjustment on the comprehensive score.

5. The psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis according to claim 1 is characterized in that: The calculation of the mania tendency score MS includes: The manic tendency assessment questionnaire is used. The questionnaire has p questions, set p = 20; set the activity weight of the qth question to be v q , with a value range of 0 to 1, and The score of the user's answer to question q is e q , the answer options are divided into r levels; Calculate the weighted activity score t for each question q =e q ×v q ; Calculating comprehensive activity indicators Manic mood tendency score MS = g(y), introducing correction factor 6. The psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis according to claim 5 is characterized in that: In order to make the predicted value close to the true sentiment score, it is calculated by minimizing the loss function; Calculate the total original activity score s = ∑a j ·u j ; Among them, a j is the weight coefficient of the jth question in the assessment of manic tendency, u j is the score of the user’s answer to the jth question; Calculate model predictions where θ0 is the baseline mania value.

7. The psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis according to claim 5 is characterized in that: To achieve the predicted value Score for true manic tendency M i A supervised learning framework is constructed by gradually approximating the regularized mean square error loss function to complete parameter optimization. Original activity score By the dynamic weight coefficient a j and the discretized answer score u j The linear combination of Prediction Model In the equation, θ0 represents the baseline manic tendency, and θ1 quantifies the marginal effect of activity on the predicted value; The true value is corrected nonlinearly To accommodate the diminishing nature of manic symptoms; Loss Function The gradient descent method is used to optimize the model to ensure the generalization performance of the model under limited samples.

8. The psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis according to claim 1 is characterized in that: The virtual scene generation module includes generating a suitable virtual reality scene based on the emotion quantification index EI(t) obtained by the emotion monitoring module and the information in the user feature database; Assuming the immersion of the virtual reality scene is I, the scene complexity is C, and the matching degree with the user's emotional tendency is M, the scene generation function can be expressed as: VRScene(EI(t),A,S,AS,DS,MS,…)=f(I,C,M,…); The mathematical model of immersion I is: I = δ × ER + ∈ × CC + ζ × SS; Among them, the richness of scene elements is ER, the coordination of scene colors is CC, the three-dimensional sense of sound is SS, and δ,∈,ζ are the corresponding weight coefficients; The mathematical model of scene element richness ER is: ER = λ × N + μ × K; Among them, the number of objects in the scene is N, the type of objects is K, and λ and μ are weight coefficients.

9. The psychological intervention system based on virtual reality cognitive behavior emotion quantitative analysis according to claim 1 is characterized in that: The intervention effect feedback module includes: Calculation formula and parameters: Use the formula E = C × (I1-I2); Where E is the emotion improvement score, C is the effectiveness coefficient of the intervention, I1 is the emotion intensity before the intervention, and I2 is the emotion intensity after the intervention; The formula for calculating mood improvement is:

Images