Psychological immune index AI quantitative evaluation platform and enhancement method

Abstract

The invention relates to the technical field of artificial intelligence and psychological health assessment, and discloses a psychological immune index AI quantitative assessment platform and an enhancement method. Comprising a multi-source data acquisition module, a data preprocessing and feature engineering module, a psychological immune index AI quantitative evaluation model construction module, a psychological immune enhancement strategy generation module and a user interaction and feedback module. By integrating multi-source heterogeneous data, including physiological, psychological, behavior and environmental data, and applying advanced data preprocessing, feature engineering and a deep neural network model based on a Transform architecture, the psychological immune index of an individual can be comprehensively and accurately quantified and evaluated. Compared with a traditional evaluation method, errors caused by a single data source and subjective factors are effectively avoided, and the accuracy and reliability of evaluation are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence and mental health assessment, in particular to a psychological immunity index AI quantitative evaluation platform and enhancement method. BACKGROUND

[0002] In today's fast-paced, high-pressure social environment, individuals face psychological stressors from various aspects of life, work, and study. Psychological immunity, as a key factor in coping with psychological stress and preventing mental illness, is increasingly important. Individuals with strong psychological immunity can better adapt to stressful environments and maintain psychological balance and health. Individuals with weak psychological immunity are more likely to experience anxiety, depression, and other psychological problems under stress.

[0003] Traditional mental health assessment methods have many limitations. On the one hand, common psychological scale assessment methods, such as the Symptom Checklist (SCL-90), the Self-Rating Depression Scale (SDS), and the Self-Rating Anxiety Scale (SAS), rely mainly on individual subjective self-reporting. This makes the assessment results easily influenced by individual subjective cognitive bias, emotional state, social expectations, and other factors. For example, some individuals may not accurately reflect their psychological state when answering the scale due to insufficient understanding of psychological problems or stigma. Or, due to emotional fluctuations at the time, they may not accurately judge the questions. On the other hand, these traditional scales are static assessments that cannot monitor individual psychological state changes in real time and dynamically, making it difficult to capture the dynamic fluctuations of psychological immunity in different situations and times. In the face of complex and changing real-life scenarios, traditional assessment methods cannot meet the demand for accurate, real-time, and dynamic assessment of individual psychological immunity levels. In view of this, we propose a psychological immunity index AI quantitative evaluation platform and enhancement method. SUMMARY

[0004] The purpose of the present application is to provide a psychological immunity index AI quantitative evaluation platform and enhancement method to solve the problems raised in the background art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solution: a psychological immunity index AI quantitative evaluation platform, comprising a multi-source data acquisition module, a data preprocessing and feature engineering module, a psychological immunity index AI quantitative evaluation model construction module, a psychological immunity enhancement strategy generation module, and a user interaction and feedback module; each module realizes real-time data interaction through a data interface.

[0006] The multi-source data acquisition module is configured to acquire individual physiological, psychological, behavioral, and environmental multi-dimensional data, the data preprocessing and feature engineering module is configured to clean, standardize, and extract and fuse the collected data, the psychological immunity index AI quantitative evaluation model construction module is configured to generate a psychological immunity index based on the fused features, the psychological immunity enhancement strategy generation module is configured to generate a personalized enhancement scheme according to the index and individual characteristics, and the user interaction and feedback module is configured to realize evaluation result display, strategy receiving, and feedback information collection.

[0007] Optionally, the multi-source data acquisition module comprises a physiological data acquisition unit, a psychological questionnaire data acquisition unit, a behavioral data acquisition unit, and an environmental data acquisition unit.

[0008] The physiological data acquisition unit is configured to collect heart rate variability (HRV), galvanic skin response (GSR), electroencephalogram (EEG), and cortisol level through a wearable device interface, with a sampling frequency of 100 Hz-500 Hz for HRV, GSR, and EEG, and once every 2-4 hours for cortisol, and the data is transmitted through encryption; the psychological questionnaire data acquisition unit is integrated with the Connor-Davidson Resilience Scale (CD-RISC) and the Perceived Stress Scale (PSAS), supports three modes of regular triggering, event triggering (such as major life events), and symptom triggering, and associates the data with a time stamp and a context label; the behavioral data acquisition unit is configured to obtain social media text (extracting emotional tendency), APP usage records (duration / frequency), geographic location trajectory (activity range), and voice data (intonation / speech rate) through authorization, and extract features after desensitization processing; and the environmental data acquisition unit is configured to collect light intensity, environmental noise, temperature, humidity, air quality index, and social environment data (personnel density / support network), and align the data with individual data in space and time.

[0009] Optionally, the data preprocessing and feature engineering module comprises a data cleaning module, a data standardization module, and a feature extraction and fusion module.

[0010] The data cleaning module removes physiological signal noise through wavelet transform and corrects abnormal values according to the 3σ criterion; invalid psychological questionnaire data (incomplete / logical errors) is removed; garbled codes and misoperation records in behavioral data are filtered; environmental parameter abnormal values are corrected; the data standardization module standardizes physiological data using Z-score, linearly transforms psychological questionnaire data according to a norm, and maps behavioral and environmental data to the [0, 1] interval using minimum-maximum standardization; the feature extraction and fusion module extracts time domain (mean / standard deviation), frequency domain (power spectral density), and nonlinear features (approximate entropy) from physiological data; uses scale scores and dimension correlations as features for psychological questionnaire data; extracts emotional scores, APP usage proportion, and activity entropy from behavioral data; extracts statistical features (mean / coefficient of variation) and trend features (daily change rate) from environmental data; and integrates the features through an attention mechanism multi-modal fusion algorithm.

[0011] Optionally, the psychological immunity index AI quantitative evaluation model construction module comprises a model architecture module, a model training module, and a model evaluation module.

[0012] The model architecture module adopts a Transformer deep neural network, comprising an input layer (feature embedding), multiple Transformer encoding layers (multi-head self-attention mechanism), an attention mechanism layer (dynamic feature weight allocation), a full connection layer (feature integration), and an output layer (regression prediction index); the model training module uses labeled data (psychological immunity levels determined by professional evaluators) as a training set, iteratively updates parameters using a cross-entropy loss function and an Adam optimizer, and prevents overfitting through L1 / L2 regularization and Dropout;

[0013] The model evaluation module evaluates accuracy through the root mean square error (RMSE), mean absolute error (MAE), and determination coefficient (R²) of the test set, and verifies the generalization ability through k-fold cross-validation and expert manual evaluation.

[0014] Optionally, the psychological immunity enhancement strategy generation module comprises a personalized strategy generation algorithm module, an intervention strategy library module, and a strategy adjustment mechanism module.

[0015] The personalized strategy generation algorithm module combines rule-based reasoning (such as preferentially recommending cognitive restructuring for low index and negative coping tendency) and reinforcement learning (optimizing strategies using index changes as reward signals); the intervention strategy library module covers cognitive behavioral therapy (cognitive restructuring / relaxation training), psychological resilience training (goal setting / social support expansion), positive psychology intervention (gratitude practice), and environmental optimization strategies (light / noise control), each strategy containing applicable features, steps, and expected effects; the strategy adjustment mechanism module monitors individual data changes in the intervention process in real time, and automatically replaces the highest correlation alternative strategy if the index does not improve (such as less than 5% improvement within 2 weeks) after strategy implementation.

[0016] Optionally, the user interaction and feedback module comprises a feedback collection unit module and a social community module.

[0017] The feedback collection unit module supports questionnaires, text, and voice feedback, extracts key information through natural language processing, and triggers an explanation mechanism for evaluation results questions; the social community module allows users to exchange experiences, invites experts to conduct online lectures, and enhances social support and a sense of belonging.

[0018] The present application also proposes a psychological immunity enhancement method for a platform, comprising the following steps:

[0019] S1, the multi-source data acquisition module continuously acquires individual physiological, psychological, behavioral, and environmental data;

[0020] S2, the data preprocessing and feature engineering module extracts fused features after data cleaning and standardization by the data cleaning module;

[0021] S3, the psychological immunity index AI quantitative evaluation model outputs the real-time psychological immunity index of the individual;

[0022] S4, the psychological immunity enhancement strategy generation module recommends a personalized intervention strategy based on the index and individual characteristics;

[0023] S5, the user receives the strategy through the interaction module and executes it, and the system monitors the data changes in real time;

[0024] S6, if the index does not improve as expected (<5% / 2 weeks) after the implementation of the strategy, repeat S4 to adjust the strategy until the index is stable at the healthy threshold (≥70 points).

[0025] Optionally, the generation logic of the personalized intervention strategy in S4 includes:

[0026] Primary screening: if the psychological immunity index is less than 60 points (low level), preferentially select an emergency intervention strategy (such as 2 times of breathing relaxation training per day + cognitive diary recording); if the index is 60-80 points (medium level), recommend a preventive strategy (such as 3 times of mindfulness meditation per week + social activity planning); if the index is greater than or equal to 80 points (high level), push a maintenance strategy (such as 1 time of psychological resilience consolidation exercise per month);

[0027] Secondary adaptation: optimize parameters combined with individual characteristics, such as extending the relaxation training time to 20 minutes for users with high neuroticism characteristics, and setting a ladder type social task (from online text interaction to offline short meeting) for users with social avoidance tendency;

[0028] Contraindication exclusion: avoid over-ventilation type breathing training for asthma patients, and avoid high-intensity cognitive training at night for sleep disorder users.

[0029] Compared with the prior art, the present application provides a psychological immunity index AI quantitative evaluation platform and enhancement method, which has the following beneficial effects:

[0030] 1、The psychological immunity index AI quantitative evaluation platform and enhancement method can comprehensively and accurately quantify and evaluate the psychological immunity index of individuals by integrating multi-source heterogeneous data including physiological, psychological, behavioral and environmental data, and using advanced data preprocessing, feature engineering and deep neural network models based on the Transformer architecture. Compared with traditional evaluation methods, the errors caused by single data source and subjective factors are effectively avoided, and the accuracy and reliability of the evaluation are significantly improved. The root mean square error (RMSE) of the evaluation model of the present application on the test set is reduced by 30%-40% compared with the traditional scale evaluation method, the determination coefficient (R²) is increased to 0.85 or more, and the real psychological immunity level of the individual can be more accurately reflected.

[0031] 2、Based on the accurate psychological immunity index evaluation results of individuals and the unique characteristics of individuals reflected by multi-source data, the psychological immunity enhancement strategy of the present application is tailored for individuals by using an algorithm based on rule-based reasoning and reinforcement learning. Compared with traditional general intervention strategies, the personalized strategy of the present application is more suitable for the actual needs of individuals and can more effectively improve the psychological immunity ability of individuals. In practical applications, after a period of personalized strategy intervention, the psychological immunity index of individuals is improved by an average of 15%-20%, negative emotional symptoms such as anxiety and depression are significantly alleviated, and the psychological resilience and ability to cope with stress are significantly enhanced.

[0032] 3、The psychological immunity index AI quantitative evaluation platform and enhancement method, through continuous multi-source data collection, real-time monitoring of the changes of individuals in the process of receiving psychological immunity enhancement strategy intervention, and real-time adjustment of the intervention strategy according to the monitoring results. This dynamic monitoring and adjustment mechanism ensures that the intervention strategy can always adapt to the dynamic change needs of individuals, improving the effectiveness and pertinence of the intervention. For example, when an individual experiences progress stagnation or adverse reactions during the implementation of a certain intervention strategy, the system can timely discover and adjust the strategy, making the intervention process more scientific and reasonable, and greatly improving the effect and efficiency of psychological immunity enhancement. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 The structure of the present application is shown in the figure. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present application will be described in detail below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0035] As Figure 1 indicated, the present application provides a technical solution: a psychological immunity index AI quantitative evaluation platform, comprising a multi-source data acquisition module, a data preprocessing and feature engineering module, a psychological immunity index AI quantitative evaluation model construction module, a psychological immunity enhancement strategy generation module and a user interaction and feedback module; each module realizes real-time data interaction through a data interface;

[0036] Among them, the multi-source data acquisition module is used to acquire individual physiological, psychological, behavioral and environmental multi-dimensional data, the data preprocessing and feature engineering module is used to clean, standardize and feature extraction and fusion of the collected data, the psychological immunity index AI quantitative evaluation model construction module is used to generate a psychological immunity index based on the fused features, the psychological immunity enhancement strategy generation module is used to generate a personalized enhancement scheme according to the index and individual characteristics, and the user interaction and feedback module is used to realize evaluation result display, strategy receiving and feedback information collection.

[0037] Optionally, the multi-source data acquisition module comprises a physiological data acquisition unit, a psychological questionnaire data acquisition unit, a behavior data acquisition unit and an environmental data acquisition unit.

[0038] The physiological data acquisition unit is configured with a wearable device interface to collect heart rate variability (HRV), galvanic skin response (GSR), electroencephalogram (EEG) and cortisol level, with a sampling frequency of 100-500 Hz for HRV, GSR and EEG, and cortisol collected every 2-4 hours, and the data is transmitted after encryption; the psychological questionnaire data acquisition unit integrates the psychological resilience scale (CD-RISC), the psychological stress assessment scale (PSAS) and the like, supports three modes of regular triggering, event triggering (such as major life events) and symptom triggering, and associates the data with a time stamp and a context label; the behavior data acquisition unit acquires social media text (extracts emotional tendency), APP usage record (duration / frequency), geographic location trajectory (activity range) and voice data (intonation / speech rate) through authorization, and extracts features after desensitization processing; the environmental data acquisition unit acquires light intensity, environmental noise, temperature, humidity, air quality index and social environment data (personnel density / support network), and aligns the individual data with space-time.

[0039] Optionally, the data preprocessing and feature engineering module comprises a data cleaning module, a data standardization module, a feature extraction and fusion module.

[0040] The data cleaning module removes physiological signal noise using wavelet transform and corrects abnormal values using the 3σ criterion; invalid psychological questionnaire data (incomplete / logical errors) are excluded; garbled codes and error operation records in behavior data are filtered; environmental parameter abnormal values are corrected; the data standardization module uses Z-score standardization for physiological data, linear transformation according to the norm for psychological questionnaire data, and minimum-maximum standardization to map behavior and environmental data to the [0, 1] interval; the feature extraction and fusion module extracts time domain (mean / standard deviation), frequency domain (power spectral density), and nonlinear features (approximate entropy) from physiological data; psychological questionnaire data are characterized by scale scores and dimension correlations; behavior data extract emotional scores, APP usage proportions, activity entropy, etc.; environmental data extract statistical features (mean / coefficient of variation) and trend features (daily variation rate); and features are integrated through an attention mechanism multi-modal fusion algorithm.

[0041] Optionally, the psychological immunity index AI quantitative evaluation model construction module includes a model architecture module, a model training module, and a model evaluation module

[0042] The model architecture module uses a Transformer deep neural network, including an input layer (feature embedding), multiple Transformer encoding layers (multi-head self-attention mechanism), an attention mechanism layer (dynamic feature weight allocation), a fully connected layer (feature integration), and an output layer (regression prediction index); the model training module uses labeled data (psychological immunity levels determined by professional evaluators) as the training set, uses a cross-entropy loss function and an Adam optimizer to iteratively update parameters, and uses L1 / L2 regularization and Dropout to prevent overfitting;

[0043] The model evaluation module evaluates accuracy through the root mean square error (RMSE), mean absolute error (MAE), and determination coefficient (R²) of the test set, and verifies the generalization ability through k-fold cross-validation and expert manual evaluation.

[0044] Optionally, the psychological immunity enhancement strategy generation module includes a personalized strategy generation algorithm module, an intervention strategy library module, and a strategy adjustment mechanism module.

[0045] The personalized strategy generation algorithm module combines rule-based reasoning (such as preferentially recommending cognitive restructuring for low index and negative coping tendencies) and reinforcement learning (optimizing strategies using index changes as reward signals); the intervention strategy library module covers cognitive behavioral therapy (cognitive restructuring / relaxation training), psychological resilience training (goal setting / social support expansion), positive psychology intervention (gratitude practice), and environmental optimization strategies (light / noise control), each strategy includes applicable features, steps, and expected effects; the strategy adjustment mechanism module monitors individual data changes during intervention in real time, and automatically replaces the highest-relevance alternative strategy if the index does not improve (e.g., less than 5% improvement in 2 weeks) after strategy implementation.

[0046] Optionally, the user interaction and feedback module comprises a feedback collection unit module and a social community module.

[0047] The feedback collection unit module supports questionnaires, text, and voice feedback, extracts key information through natural language processing, and triggers an explanation mechanism for evaluation results doubts; the social community module allows users to exchange experiences, invites experts to conduct online lectures, and enhances social support and a sense of belonging.

[0048] The embodiment also proposes a psychological immunity enhancement method, comprising the following steps:

[0049] S1: The multi-source data acquisition module continuously acquires individual physiological, psychological, behavioral, and environmental data;

[0050] S2: The data preprocessing and feature engineering module extracts and fuses features after data cleaning and standardization;

[0051] S3: The psychological immunity index AI quantitative evaluation model outputs the individual real-time psychological immunity index;

[0052] S4: The psychological immunity enhancement strategy generation module recommends personalized intervention strategies based on the index and individual characteristics; the generation logic of the personalized intervention strategy includes:

[0053] First-level screening: if the psychological immunity index is <60 points (low level), preferentially select emergency intervention strategies (such as 2 times of daily breathing relaxation training + cognitive diary recording); if the index is 60-80 points (medium level), recommend preventive strategies (such as 3 times of weekly mindfulness meditation + social activity planning); if the index is ≥80 points (high level), push maintenance strategies (such as 1 time of monthly psychological resilience consolidation exercise);

[0054] Second-level adaptation: optimize parameters combined with individual characteristics, such as extending the relaxation training duration to 20 minutes for high neuroticism trait users, and setting a ladder-type social task (from online text interaction to offline short meeting) for users with social avoidance tendency;

[0055] Taboo exclusion: avoid over-ventilation type breathing training for asthma patients, and avoid high-intensity cognitive training at night for sleep disorder users.

[0056] S5: The user receives the strategy through the interaction module and executes it, and the system monitors the data changes in real time;

[0057] S6: If the index improvement after strategy implementation does not meet the expectation (<5% / 2 weeks), repeat S4 to adjust the strategy until the index is stable at the health threshold (≥70 points).

[0058] As an application of the embodiment:

[0059] The embodiment can comprehensively and accurately quantify and evaluate the psychological immunity index of an individual by integrating multi-source heterogeneous data including physiological, psychological, behavioral and environmental data, and using advanced data preprocessing, feature engineering and deep neural network model based on the Transformer architecture. Compared with traditional evaluation methods, the errors caused by single data source and subjective factors are effectively avoided, and the accuracy and reliability of the evaluation are significantly improved. The root mean square error (RMSE) of the evaluation model of the application on the test set is reduced by 30%-40% compared with the traditional scale evaluation method, the determination coefficient (R²) is increased to more than 0.85, and the real psychological immunity level of the individual can be more accurately reflected.

[0060] The above is a detailed description of the application in general, but some modifications or improvements can be made on the basis of the application, which is obvious to those skilled in the art. Therefore, the modifications or improvements without departing from the spirit of the application are within the protection scope of the application.

Claims

1. A psychometric immune index (AI) quantitative assessment platform, characterized in that, The system comprises a multi-source data acquisition module, a data preprocessing and feature engineering module, a psychological immunity index AI quantitative evaluation model construction module, a psychological immunity enhancement strategy generation module, and a user interaction and feedback module. The modules realize real-time data interaction through a data interface. The multi-source data acquisition module is configured to acquire individual physiological, psychological, behavioral, and environmental multi-dimensional data.

2. The psychoimmune index AI quantification evaluation platform according to claim 1, characterized in that, The data preprocessing and feature engineering module is configured to clean, standardize, and extract and fuse features of the collected data. The psychological immunity index AI quantitative evaluation model construction module is configured to generate a psychological immunity index based on the fused features. The psychological immunity enhancement strategy generation module is configured to generate a personalized enhancement plan based on the index and individual characteristics. The user interaction and feedback module is configured to display evaluation results, receive strategies, and collect feedback information. The multi-source data acquisition module comprises:

3. The psycho-immuno index AI quantitative evaluation platform according to claim 1, characterized in that, A physiological data acquisition unit configured to configure a wearable device interface and collect heart rate variability, skin conductance response, and cortisol level. A psychological questionnaire data acquisition unit configured to integrate a psychological resilience scale and a psychological stress assessment scale, support three modes of regular triggering, event triggering, and symptom triggering, and associate data with a time stamp and a context label. A behavioral data acquisition unit configured to obtain social media text, APP usage records, and voice data through authorization and extract features after desensitization processing. An environmental data acquisition unit configured to collect light intensity, environmental noise, temperature, humidity, air quality index, and social environment data, and align individual data with space-time.

4. The psychometric immunological index AI quantitative evaluation platform according to claim 1, characterized in that, The data preprocessing and feature engineering module comprises: A data cleaning module configured to remove physiological signal noise using wavelet transform and correct abnormal values using a 3σ criterion; eliminate invalid psychological questionnaire data, filter garbled data and misoperation records in behavioral data, and correct environmental parameter abnormal values. A data standardization module configured to standardize physiological data using Z-score, linearly transform psychological questionnaire data according to a norm, and map behavioral and environmental data to the [0, 1] interval using minimum-maximum standardization. A feature extraction and fusion module configured to extract time domain, frequency domain, and nonlinear features from physiological data; use scale scores and dimension correlations as features for psychological questionnaire data; extract emotional scores, APP usage proportion, and activity entropy from behavioral data; extract statistical and trend features from environmental data; and integrate features through an attention mechanism multi-modal fusion algorithm.

5. The psycho-immuno index AI quantitative evaluation platform according to claim 1, characterized in that, The psychological immunity index AI quantitative evaluation model construction module comprises: A model architecture module configured to use a Transformer deep neural network comprising an input layer, multiple Transformer encoding layers, an attention mechanism layer, a fully connected layer, and an output layer. A model training module configured to use labeled data as a training set, update parameters iteratively using a cross-entropy loss function and an Adam optimizer, and prevent overfitting through L1 / L2 regularization and Dropout. A model evaluation module configured to evaluate accuracy through root mean square error, mean absolute error, and determination coefficient of a test set. The psychological immunity enhancement strategy generation module comprises: A personalized strategy generation algorithm module configured to combine rule-based reasoning and reinforcement learning. Intervention strategy library module, covering cognitive behavioral therapy, psychological resilience training, positive psychology intervention, environmental optimization strategy, each strategy containing applicable characteristics, steps, expected effect; Strategy adjustment mechanism module, real-time monitoring of individual data changes during intervention, if the index does not improve after the implementation of the strategy, automatically replace the highest correlation alternative strategy.

6. The psychometric immunological index AI quantitative evaluation platform according to claim 1, characterized by, The user interaction and feedback module comprises: Feedback collection unit module, supporting questionnaire, text, voice feedback, extracting key information through natural language processing, triggering explanation mechanism for evaluation result questions; Social community module, allowing users to exchange experiences, inviting experts to carry out online lectures, enhancing social support and sense of belonging.

7. A method of psychological immunity enhancement based on the platform of any one of claims 1-6, characterized in that, Comprising the following steps: S1, the multi-source data acquisition module continuously acquires individual physiological, psychological, behavioral and environmental data; S2, the data preprocessing and feature engineering module extracts and fuses features after data cleaning and standardization; S3, the psychological immunity index AI quantitative evaluation model outputs the individual real-time psychological immunity index; S4, the psychological immunity enhancement strategy generation module recommends personalized intervention strategies based on the index and individual characteristics; S5, the user receives the strategy through the interaction module and executes, and the system monitors the data changes in real time; S6, if the index improves after the implementation of the strategy, repeat S4 to adjust the strategy until the index is stable at the health threshold.

8. The method of claim 7, wherein the psychological immune enhancement method is characterized by, The generation logic of the personalized intervention strategy in S4 comprises: Primary screening, if the psychological immunity index is less than 60 points, preferentially select emergency intervention strategy; the index is 60-80 points, recommend preventive strategy; the index is greater than or equal to 80 points, push maintenance strategy; Secondary adaptation, optimize parameters combined with individual characteristics, such as extending the relaxation training time of high neuroticism characteristic users to 20 minutes, setting ladder type social tasks for social avoidance tendency users; Taboo exclusion, avoid high-intensity cognitive training at night for sleep disorder users.

Images