Intelligent emotion intervention and personalized recommendation method and system based on large model in medical industry

Through emotional intelligent intervention and personalized recommendation methods based on big models, the efficiency and accessibility of mental health services are solved, efficient personalized treatment and privacy protection are achieved, and different environments and user groups are adapted to improve the overall mental health services. Effect.

CN119007942APending Publication Date: 2024-11-22SHANDONG LANGCHAO YUNTOU INFORMATION TECH CO LTD
View PDF 0 Cites 31 Cited by

Patent Information

Application Number
CN202411002570.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-25
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The prior art is difficult to improve the efficiency and accessibility of mental health services while ensuring the security and privacy of user data.

Method used

Using a large model-based emotional intelligent intervention and personalized recommendation method, through multi-dimensional data collection, sentiment analysis, personalized model construction, intelligent recommendation, feedback and learning mechanism, privacy protection and security measures, combined with deep learning and reinforcement learning, we provide Personalized mental health services.

Benefits of technology

It realizes efficient emotional recognition, personalized treatment plans, intelligent recommendation, continuous optimization and privacy protection, improves the efficiency and accessibility of mental health services, ensures the security of user data, reduces costs, and adapts to different medical environments and User group.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119007942A_ABST
    Figure CN119007942A_ABST
Patent Text Reader

Abstract

The invention discloses an emotion intelligent intervention and personalized recommendation method and system based on a large model in the medical industry, belongs to the technical field of natural language processing, and aims to solve the technical problem of how to improve the efficiency and accessibility of psychological health services and ensure the safety and privacy of user data. According to the technical scheme, the method comprises the following steps: data acquisition and preprocessing: acquiring related data in multiple dimensions, performing data cleaning and structured processing on the acquired related data, and extracting emotion keywords and themes; sentiment analysis: performing fine tuning by using pre-training models such as BERT or GPT-3 to adapt to medical sentiment recognition; constructing a personalized model: customizing a user emotion intervention strategy through time sequence analysis, social network analysis, behavior pattern mining and user portrait construction; generating an emotion intervention strategy; intelligent recommendation; a feedback and learning mechanism; privacy protection and security are realized; designing a user interface and interaction; and integrating and deploying the architecture.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of natural language processing technology, and in particular to a method and system for emotional intelligence intervention and personalized recommendation based on a large model in the medical industry. Background Art

[0002] With the rapid development of information technology, artificial intelligence is increasingly used in the medical field, especially in mental health services. Traditional mental health services rely on face-to-face communication between professional counselors and patients, which is limited by time, location and resources.

[0003] Therefore, how to improve the efficiency and accessibility of mental health services while ensuring the security and privacy of user data is a technical problem that needs to be solved urgently. Summary of the invention

[0004] The technical task of the present invention is to provide a method and system for emotional intelligence intervention and personalized recommendation based on a large model in the medical industry to solve the problem of how to improve the efficiency and accessibility of mental health services while ensuring the security and privacy of user data.

[0005] The technical task of the present invention is achieved in the following way: a method for emotional intelligence intervention and personalized recommendation based on a large model in the medical industry, the method is as follows:

[0006] Data collection and preprocessing: Collect relevant data from multiple dimensions, clean and structure the collected data, and extract emotional keywords and themes;

[0007] Sentiment analysis: Use pre-trained models such as BERT or GPT-3 and fine-tune them to suit medical sentiment recognition;

[0008] Build personalized models: customize user emotion intervention strategies through time series analysis, social network analysis, behavior pattern mining, and user portrait construction;

[0009] Generate emotional intervention strategies: Use a multi-factor decision-making algorithm to dynamically update and recommend personalized emotional intervention strategies;

[0010] Intelligent recommendation: Combine collaborative filtering and deep learning to achieve real-time resource recommendations, and improve recommendation results through learning and optimization;

[0011] Feedback and learning mechanism: Utilize reinforcement learning, user feedback processing and continuous model updates to continuously improve model performance;

[0012] Privacy protection and security: Apply data encryption technology, anonymization processing and data security strategies;

[0013] User interface and interaction design: including UI, multi-language support, accessibility and 24 / 7 online support;

[0014] Integration and deployment architecture: design API interface design, cloud platform deployment, elastic expansion capabilities and high availability design.

[0015] As a preferred embodiment, data collection and preprocessing are specifically as follows:

[0016] Online communication records, personal diaries and logs, questionnaire survey results, physiological and environmental sensor data, and voice recordings are used to achieve data diversification and integration. Among them, online communication records are obtained by analyzing users' chat records, emails, and social media posts on the platform to capture emotional expressions and psychological state clues in daily communication; personal diaries and logs are used to encourage users to record their daily feelings and understand the changes in the user's inner world through text analysis; questionnaire survey results are obtained by designing professional mental health assessment questionnaires and regularly collecting user self-assessment reports on depression scales and anxiety scales; physiological and environmental sensor data are obtained by integrating biological indicator data (such as heart rate, sleep quality) and environmental data (such as light and sound level) provided by wearable devices. The biological indicator data and environmental data indirectly reflect the user's emotional state and living environment; voice recordings use voice recognition technology to analyze the emotional color and tone changes in the user's voice to supplement the lack of text information;

[0017] Data cleaning and structural processing: denoising, duplicate data processing, standardization and unification processing, and missing value processing; among them, denoising is to filter out irrelevant characters, links and emoticons using regular expressions and natural language processing technology; duplicate data processing is to remove duplicate records through hash comparison to maintain the uniqueness of the data set; standardization and unification processing is to unify the data formats from different sources for subsequent analysis. The data format includes date format and text encoding format; missing value processing is to use mean filling and model prediction filling interpolation technology to process missing data to ensure data integrity;

[0018] Emotional keyword and topic extraction: Natural language processing technology is used to extract emotional keywords and topics, perform topic model analysis (such as LDA), and evaluate emotional intensity. Among them, emotional keyword extraction uses emotional dictionaries and keyword extraction algorithms to identify words that are closely related to emotional expression, such as "sadness", "anxiety", "excitement", etc. Topic model analysis is to discover potential topic structures from a large amount of text to help understand the focus of users and emotional triggers; emotional intensity evaluation is to evaluate the emotional polarity and intensity of keywords in combination with the context, providing a more detailed perspective for sentiment analysis.

[0019] As a preference, sentiment analysis is specifically as follows:

[0020] Selecting medical big models: Using BERT and GPT-3 advanced medical big models, which have been pre-trained on large-scale text data, have strong language understanding and generation capabilities, and have been fine-tuned in the medical and mental health fields to adapt to specific sentiment analysis tasks;

[0021] Model fine-tuning and domain adaptation: Dataset preparation, fine-tuning, and domain adaptation; Dataset preparation is to collect labeled data in the medical and mental health fields from patient diaries, forum discussions, and medical records, and build a domain dataset specifically for sentiment analysis; Fine-tuning is to use labeled data in the medical and mental health fields to fine-tune the medical big model and optimize model parameters so that the medical big model can better understand the emotional expression in the medical context; Domain adaptation is to enable the medical big model to migrate from the source domain (such as general text) to the target domain (such as medical sentiment analysis) through contrastive learning and transfer learning strategies;

[0022] Multimodal emotion analysis: Integrate multimodal data of speech, video and physiological signals, specifically: Speech emotion recognition: Use acoustic features and speech recognition technology to analyze emotional features in speech, such as intonation, pitch and rhythm; Visual emotion analysis: Analyze facial expressions and body language, and use computer vision technology for emotion recognition; Physiological signal analysis: Combine physiological indicators such as heart rate and blood pressure to assist in judging the user's emotional state;

[0023] Emotional state recognition algorithm: hybrid model, sentiment analyzer, uncertainty quantification and dynamic emotion trajectory modeling; among them, the hybrid model combines the features of text, voice and physiological signals to build a hybrid emotion recognition model, such as a deep integrated network or a multi-task learning framework; the sentiment classifier is a deep learning model that trains convolutional neural networks, recurrent neural networks or Transformer architectures as sentiment classifiers to identify multiple emotion categories such as joy, anger, fear and sadness; uncertainty quantification is the uncertainty predicted by the Bayesian method or Monte Carlo dropout estimation model, providing a more reliable basis for decision-making; dynamic emotion trajectory modeling uses RNN or LSTM to capture the dynamic changes of user emotions over time to form an emotion trajectory model.

[0024] As a preferred method, the personalized model is constructed as follows:

[0025] Time series analysis: Use time series analysis methods to track and predict the user's emotional state, capture the periodic changes, trends and anomalies of the user's emotions, and help identify potential psychological problems and emotional fluctuation patterns; time series analysis methods include ARIMA (autoregressive integrated moving average model), LSTM (long short-term memory network) or Prophet algorithm; among them, emotional fluctuation patterns include emotional timelines and predicted future emotional states; emotional timelines are to construct the user's emotional timeline, showing the changes in emotional states within a set time in the past, so as to identify patterns and anomalies; predicting future emotional states is to use time series prediction to warn of possible negative emotional outbreaks in advance so as to intervene in time;

[0026] Social network analysis: By analyzing the user's social network data, understand the user's emotional expression and interaction patterns in the social environment; specifically: Emotional communication analysis: study the impact of users' emotional communication in social networks, identify the patterns and key nodes of emotional contagion; Social influence: evaluate the influence of users in social networks, and consider their role in emotional fluctuations; Friends circle analysis: analyze the structure of users' friends circles and identify important people who have an impact on users' emotions;

[0027] Behavior pattern mining: using machine learning and data mining techniques such as association rules, cluster analysis and deep learning to reveal the relationship between users’ daily behaviors and emotional states; specifically: Activity pattern recognition: identifying users’ daily activity patterns, such as sleep, diet, and exercise habits, which are related to their emotional states; Trigger event detection: discovering key events that lead to emotional changes, such as work pressure, family problems, etc.

[0028] Behavior-emotion correlation: Analyze the correlation between specific user behaviors and specific emotions to provide a basis for personalized intervention;

[0029] User portrait construction: construct the user's emotional portrait, which is a comprehensive report that details the user's emotional state, behavioral patterns, social network and life events; specifically: feature extraction: extract the user's emotional features, social features and behavioral features to form a multi-dimensional user feature vector; portrait construction: map the feature vector to the user portrait to form a visual overview of the user's emotions; portrait update: as user data increases, the user portrait is updated in real time to reflect the latest changes in the user's emotional state.

[0030] As a preferred method, the emotion intervention strategy is generated as follows:

[0031] Multi-factor decision-making algorithm: Fuzzy logic, genetic algorithm or rule-based reasoning system multi-factor decision-making algorithm is used to comprehensively consider factors such as user emotional state, personal characteristics, life events, social network influence, and behavior patterns to generate emotional intervention strategies; among them, user emotional state is based on the results of the sentiment analysis model to determine the user's current and predicted emotional state; personal characteristics include age, gender, occupation, health history, etc.; life events are life stress, important events or changes within a set time range; social network influence refers to the impact of emotional state and interaction in the social network on users; behavior patterns are users' behavioral habits and activity patterns;

[0032] Dynamic treatment plan update: The emotional intervention strategy is not static, but is dynamically updated based on user feedback and real-time data. When the user's emotional state changes or the intervention effect is not ideal, real-time monitoring, feedback learning, and adaptive adjustment are carried out. Among them, real-time monitoring is to continuously monitor the user's emotional changes and adjust the plan according to new data. Feedback learning refers to incorporating user feedback on treatment effects into model learning to optimize treatment strategies. Adaptive adjustment is based on reinforcement learning, and the plan is dynamically adjusted according to user behavior and treatment response.

[0033] Personalized psychotherapy recommendations: Based on the user profile and treatment plan, personalized psychotherapy suitable for the user is recommended, including cognitive behavioral intervention, mindfulness meditation, online psychological counseling, self-help reading materials and applications; cognitive behavioral intervention is to adjust specific thinking patterns; mindfulness meditation is to teach users mindfulness skills to help relieve anxiety and stress; online psychological counseling is to recommend online counselors or intervention groups suitable for users; self-help reading materials are to provide relevant books, articles and guidance manuals; applications are to recommend mobile applications that help with emotion management;

[0034] Self-help resource library: educational resources, exercises and activities, community support, and mental health tools; among them, educational resources include psychology lectures, workshops, and online courses; exercises and activities include meditation, breathing exercises, and relaxation techniques; community support includes online forums and mutual aid groups to promote interaction and support among users; mental health tools include mood tracking tools and anxiety management tools to help users manage themselves.

[0035] As a preferred option, the intelligent recommendations are as follows:

[0036] Combination of collaborative filtering and deep learning: Integrate collaborative filtering and deep learning technologies to improve the accuracy and personalization of recommendations; specifically: User-user collaborative filtering: Analyze the behavior patterns of user groups, identify users with similar emotional needs, and recommend resources they have successfully used; Item-item collaborative filtering: Analyze the correlation between resources, and recommend similar resources based on the user's past response to any resource; Deep learning model: Use models such as convolutional neural network (CNN) or recurrent neural network (RNN) to learn users' emotional preferences and behavior patterns, and further improve the accuracy of recommendations;

[0037] Real-time recommendation algorithm: adopting real-time recommendation algorithm, such as stream-based recommendation system, can generate personalized suggestions at the moment of user interaction with the platform; specifically: real-time sentiment analysis: instant analysis of the user's current emotional state to match the corresponding resources; dynamic update: as the user's behavior and emotional state change, the recommendation list is updated in real time; event-driven: respond to user login, browsing, search for specific events, and quickly push relevant resources;

[0038] Personalized resource sorting: Sort resources individually based on the user's emotional state, historical behavior, preferences, and satisfaction; Rating prediction: Predict the user's preference for untouched resources and sort them based on the predicted scores; Diversity considerations: Avoid over-recommendation of similar resources and ensure the diversity and novelty of the recommendation list; Interest evolution: Track changes in user interests and adjust the sorting strategy to reflect the latest needs;

[0039] System learning and optimization: The specific learning and optimization strategies are as follows: Online learning: Collect user feedback during the actual recommendation process and continuously adjust model parameters; A / B testing: Compare the effects of different recommendation strategies and select the best solution; Meta-learning: By learning how to learn, quickly adapt to new users and new resources, and improve the generalization ability of the recommendation system; Reinforcement learning: Through interaction with users, the system learns the best recommendation strategy through rewards and punishments.

[0040] Preferably, the feedback and learning mechanism is as follows:

[0041] Reinforcement learning applications: including environment modeling, action selection, reward function and strategy iteration; environment modeling: considering user emotional state and platform behavior as the environment for reinforcement learning; action selection: recommending resources or intervention measures as actions for reinforcement learning; reward function: defining rewards based on indicators such as user satisfaction, resource usage and emotional improvement; strategy iteration: learning and optimizing recommendation strategies through multiple interactions to maximize long-term rewards;

[0042] User feedback processing: including satisfaction surveys, emotional feedback, complaints and suggestions; among them, satisfaction surveys: regularly conduct satisfaction surveys to understand users' views on recommended resources and intervention measures; emotional feedback: users provide real-time feedback on emotional changes to help the system understand the effects of interventions; complaints and suggestions: set up complaint and suggestion functions to collect users' opinions on improving the platform;

[0043] Continuous model update: including online update, offline evaluation and model fusion; online update: real-time update of model parameters without affecting services; offline evaluation: regular offline evaluation to compare the performance of new and old models; model fusion: combining the advantages of multiple models to build a hybrid model to improve overall performance;

[0044] Expert knowledge base integration: including expert consultation, knowledge base update and knowledge graph; expert consultation: regularly invite experts for consultation to obtain the latest research results and clinical practices; knowledge base update: integrate expert advice and the latest research into the model to improve the scientificity and effectiveness of recommendations; knowledge graph: build a knowledge graph of emotional health to facilitate the model to understand and apply professional knowledge;

[0045] Privacy protection and security are as follows:

[0046] Data encryption technology: including transport layer encryption, storage encryption, key management, i.e. end-to-end encryption; transport layer encryption: using SSL / TLS protocol to ensure data security during transmission; storage encryption: user data is encrypted and stored on the server, so that even if the data is stolen, it cannot be directly read; key management: following best practices, strictly managing and protecting encryption keys to prevent unauthorized access; end-to-end encryption: achieving end-to-end communication encryption between user devices and the platform to ensure that information is not attacked by middlemen during transmission;

[0047] Anonymization: including differential privacy, desensitization technology and aggregate statistics; differential privacy: adding random noise to the data so that the analysis results cannot be traced back to individuals; desensitization technology: replacing or deleting information that can directly or indirectly identify personal names, addresses and phone numbers; aggregate statistics: performing statistical analysis on anonymous data sets to provide overall trends without exposing individual information;

[0048] Data security strategy: including access control, audit tracking, security updates and security training; among them, access control: implement multi-layer permission management to ensure that only authorized personnel can access sensitive data; audit tracking: record all data access and modification operations so that they can be traced back when security incidents occur; security updates: regularly update systems and software to patch known security vulnerabilities; security training: provide data security awareness training to employees to enhance their awareness of prevention;

[0049] Regulatory compliance: including compliance review, user right to know, data minimization and user consent mechanism; among them, compliance review: regularly conduct privacy and security compliance reviews to ensure compliance with the latest regulatory requirements; user right to know: provide a clear privacy policy to inform users of how their data is collected, used and shared; data minimization principle: only collect data necessary for the service and avoid unnecessary data collection; user consent mechanism: obtain explicit consent from users before processing and using the corresponding personal data;

[0050] The user interface and interaction are as follows:

[0051] Friendly UI design, with the following features:

[0052] ① Simple layout: clear page layout reduces user cognitive burden;

[0053] ② Color psychology: Use the principles of color psychology to create a visual environment that helps regulate emotions;

[0054] ③Touch-friendly: Optimized touch screen operation to adapt to various devices and screen sizes;

[0055] ④ Responsive design: automatically adjust the interface to suit different devices and resolutions;

[0056] Multi-language support, as follows:

[0057] ① Language switching: users select the interface language according to their preferences;

[0058] ② Localized content: resources and prompt information are localized and translated to ensure cultural adaptability;

[0059] ③ Voice Assistant: supports voice input and output in multiple languages, which is convenient for users with different language backgrounds;

[0060] Accessibility features, as follows:

[0061] ① Screen reader compatibility: optimize HTML structure to ensure compatibility with screen readers;

[0062] ②High contrast mode: Provides high contrast options for the convenience of visually impaired users;

[0063] ③Keyboard navigation: All functions can be operated by keyboard, without mouse;

[0064] ④Font adjustment: Allow users to customize font size and type to meet different vision needs;

[0065] 24 / 7 online support, details as follows:

[0066] ① Real-time chat: Integrate real-time chat function, users can quickly ask questions and get answers;

[0067] ② Self-service: set up FAQs and knowledge base to allow users to solve problems by themselves;

[0068] ③Manual customer service: Provide telephone and email support, and provide professional answers to complex questions;

[0069] ④ Community forum: encourage users to exchange experiences and support each other in the forum;

[0070] Integration and deployment, as follows:

[0071] The API interface design is as follows:

[0072] ①Open standards: follow the RESTful API design principles and use the JSON data format;

[0073] ② Authentication: Implement OAuth or other security authentication mechanisms to protect data security;

[0074] ③ Version control: Versioning APIs to ensure that upgrades do not affect existing integrations;

[0075] ④ Documentation: Provide detailed API documentation to simplify developers’ integration work;

[0076] The cloud platform architecture is as follows:

[0077] ①Containerization: Use Docker and Kubernetes to containerize services for easy deployment and management;

[0078] ② Microservice architecture: split into independent services to improve maintainability and scalability;

[0079] ③ Load balancing: Configure a load balancer to disperse traffic and prevent single point failure;

[0080] ④Distributed storage: Use Cassandra or MongoDB distributed database system to ensure data consistency;

[0081] Elastic expansion capabilities, as follows:

[0082] ① Dynamic expansion and contraction: Automatically expand or reduce computing resources by monitoring resource utilization;

[0083] ②Horizontal expansion: Increase the number of instances when needed to handle more concurrent requests;

[0084] ③ Resource scheduling: Intelligently schedule system resources to ensure that key services are given priority in resource allocation;

[0085] High availability design, as follows:

[0086] ① Redundancy design: set up backup services and data to prevent the main service from failing;

[0087] ② Failure switching: When the main service fails, it can quickly switch to the backup service;

[0088] ③Regular backup and recovery: Back up data regularly to ensure data security and support rapid recovery;

[0089] ④Health Check: Implement service health check to promptly identify and resolve potential problems.

[0090] An emotional intelligence intervention and personalized recommendation system based on a large model, the system is used to implement the emotional intelligence intervention and personalized recommendation method based on a large model in the medical industry as described above; the system includes:

[0091] The data collection and preprocessing module is used to collect relevant data in multiple dimensions, and extract emotional keywords and themes after data cleaning and structure processing of the collected relevant data;

[0092] Sentiment analysis module, which is used to fine-tune pre-trained models such as BERT or GPT-3 to adapt to medical emotion recognition;

[0093] Personalized model building module, used to customize user emotion intervention strategies through time series analysis, social network analysis, behavior pattern mining and user portrait construction;

[0094] The emotion intervention strategy generation module is used to dynamically update and recommend personalized emotion intervention strategies using a multi-factor decision-making algorithm;

[0095] Intelligent recommendation module, which combines collaborative filtering and deep learning to achieve real-time resource recommendations and improve recommendation results through learning and optimization;

[0096] Feedback and learning mechanism module, which is used to continuously improve model performance by using reinforcement learning, user feedback processing and continuous model updating;

[0097] Privacy protection and security module, used to apply data encryption technology, anonymization processing and data security strategies;

[0098] User Interface and Interaction Design module to include UI, multi-language support, accessibility features, and 24 / 7 online support;

[0099] The architecture integration and deployment module is used to design API interface design, cloud platform deployment, elastic expansion capabilities and high availability design.

[0100] An electronic device comprising: a memory and at least one processor;

[0101] Wherein, the memory stores computer-executable instructions;

[0102] The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the emotional intelligence intervention and personalized recommendation method based on a big model in the medical industry as described above.

[0103] A computer-readable storage medium stores computer-executable instructions. When a processor executes the computer-executable instructions, the above-mentioned emotional intelligence intervention and personalized recommendation method based on a large model in the medical industry is implemented.

[0104] The medical industry's emotional intelligence intervention and personalized recommendation method and system based on a large model of the present invention has the following advantages:

[0105] (i) The present invention combines medical health information technology, big data analysis, psychotherapy and consultation, deep learning, personalized recommendation, and network security and privacy protection technology, focusing on the development of intelligent medical solutions, applied to online mental health services, and providing personalized emotional intervention and treatment strategies by analyzing and understanding the user's emotional state, while ensuring the security and privacy of user data; in addition, it improves the efficiency and accessibility of mental health services while ensuring the security and privacy of user data;

[0106] (II) The present invention aims to bring revolutionary changes to the field of mental health, providing smarter, more intimate and more extensive treatment services to meet the growing needs of mental health, as follows:

[0107] ① Efficient emotion recognition: Through the medical big model, the user's speech and behavior data are deeply analyzed to quickly and accurately identify and understand the user's emotional state so as to intervene in time;

[0108] ② Personalized treatment plan: Generate a customized treatment plan based on each user’s emotional characteristics and mental health needs, including but not limited to psychotherapy, self-regulation techniques and educational resources to improve the effectiveness and satisfaction of treatment;

[0109] ③ Intelligent recommendation: Using machine learning algorithms, dynamically recommend psychological counseling, online courses, application tools, etc. that are most suitable for the user's current situation to ensure maximum resource utilization;

[0110] ④Continuous optimization: Through user feedback and data analysis of treatment effects, we continuously optimize model parameters, improve intervention strategies, and enhance overall treatment effects;

[0111] ⑤ Privacy protection: Design and implement strict privacy protection measures to ensure the security of user data, comply with relevant laws and regulations, and protect the privacy rights and interests of users;

[0112] ⑥Scalability and universality: It has good scalability and adaptability, can be applied to different medical environments and user groups, and supports large-scale online mental health services;

[0113] ⑦ Resource conservation and efficiency improvement: reduce the pressure on human resources in traditional psychotherapy, improve service efficiency, and enable more people to obtain psychological support in a timely and convenient manner;

[0114] (III) The present invention creates an efficient, intelligent and safe environment, which not only improves the mental health level of individuals, but also has a positive impact on the entire medical system, as follows:

[0115] ① Instant intervention capability: By analyzing users’ emotional expressions in real time, the platform can quickly identify potential mental health problems and provide corresponding intervention measures immediately, reducing the risk of delayed treatment;

[0116] ②Personalized service: Provide customized treatment plans based on the unique situation of each user, improve the pertinence and effectiveness of treatment, and enhance user experience;

[0117] ③ Resource optimization: The automated process reduces the reliance on professional psychological consultants, allowing limited medical resources to be allocated and used more efficiently;

[0118] ④ Popularity and accessibility: Users can access the platform anytime and anywhere through the Internet without being restricted by geographical location, which expands the coverage of mental health services, especially for remote areas and busy people;

[0119] ⑤Continuous treatment: It can continuously monitor the user's emotional changes, dynamically adjust the treatment plan, and ensure the continuity and coherence of the treatment;

[0120] ⑥Data-driven optimization: By collecting and analyzing user feedback, we continuously iterate and optimize the model to improve the overall performance and treatment effect of the platform;

[0121] ⑦ Privacy protection: Use advanced data encryption and anonymization technologies to ensure the privacy and security of user data and enhance user trust in the platform;

[0122] ⑧ Reduce costs: Compared with traditional one-on-one consultation, it can reduce the overall cost of mental health services, making psychotherapy affordable for more people;

[0123] ⑨Social impact: It helps to reduce the burden of social and psychological health, improve the overall level of mental health, and promote social harmony and personal happiness;

[0124] ⑩Research value: It provides a large amount of valuable data for research in the field of mental health, which will help further explore the prevention and treatment strategies of mental illness. BRIEF DESCRIPTION OF THE DRAWINGS

[0125] The present invention is further described below in conjunction with the accompanying drawings.

[0126] Attached Figure 1 A flowchart of the big model-based emotional intelligence intervention and personalized recommendation method for the medical industry;

[0127] Attached Figure 2 This is the structural block diagram of the emotional intelligence intervention and personalized recommendation system based on the big model. DETAILED DESCRIPTION

[0128] The emotional intelligence intervention and personalized recommendation method and system for the medical industry based on a large model of the present invention are described in detail below with reference to the drawings and specific embodiments of the specification.

[0129] Embodiment 1:

[0130] As attached Figure 1 As shown, this embodiment provides an emotional intelligence intervention and personalized recommendation method based on a large model in the medical industry, and the method is specifically as follows:

[0131] S1. Data collection and preprocessing: Collect relevant data from multiple dimensions, clean and structure the collected relevant data, and extract emotional keywords and themes;

[0132] S2. Sentiment analysis: Use pre-trained models such as BERT or GPT-3 and fine-tune them to adapt to medical emotion recognition;

[0133] S3. Build personalized models: customize user emotion intervention strategies through time series analysis, social network analysis, behavior pattern mining and user portrait construction;

[0134] S4, Generate emotional intervention strategies: Use a multi-factor decision-making algorithm to dynamically update and recommend personalized emotional intervention strategies;

[0135] S5. Intelligent recommendation: Combine collaborative filtering and deep learning to achieve real-time resource recommendation, and improve recommendation results through learning and optimization;

[0136] S6. Feedback and learning mechanism: Utilize reinforcement learning, user feedback processing and continuous model updating to continuously improve model performance;

[0137] S7. Privacy protection and security: Apply data encryption technology, anonymization processing and data security strategies;

[0138] S8, User Interface and Interaction Design: including UI, multi-language support, accessibility and 24 / 7 online support;

[0139] S9. Integration and deployment architecture: design API interface design, cloud platform deployment, elastic expansion capabilities and high availability design.

[0140] The data collection and preprocessing in step S1 of this embodiment are specifically as follows:

[0141] S102, online communication records, personal diaries and logs, questionnaire survey results, physiological and environmental sensor data, and voice recordings are used to achieve data diversification and integration. Among them, online communication records are obtained by analyzing users' chat records, emails, and social media posts on the platform to capture emotional expressions and psychological state clues in daily communication; personal diaries and logs are used to encourage users to record their daily feelings and understand the changes in the user's inner world through text analysis; questionnaire survey results are obtained by designing professional mental health assessment questionnaires and regularly collecting user self-assessment reports on depression scales and anxiety scales; physiological and environmental sensor data are obtained by integrating biological indicator data (such as heart rate, sleep quality) and environmental data (such as light and sound level) provided by wearable devices. The biological indicator data and environmental data indirectly reflect the user's emotional state and living environment; voice recordings are analyzed through voice recognition technology to analyze the emotional color and tone changes in the user's voice to supplement the lack of text information;

[0142] S102, data cleaning and structural processing: denoising, duplicate data processing, standardization and unification processing, and missing value processing; among them, denoising is to use regular expressions and natural language processing technology to filter out irrelevant characters, links and emoticons; duplicate data processing is to remove duplicate records through hash comparison to maintain the uniqueness of the data set; standardization and unification processing is to unify the data formats from different sources for subsequent analysis, and the data formats include date format and text encoding format; missing value processing is to use mean filling and model prediction filling interpolation technology to process missing data to ensure data integrity;

[0143] S103. Emotional keyword and topic extraction: Natural language processing technology is used to extract emotional keywords and topics, perform topic model analysis (such as LDA), and evaluate emotional intensity. Among them, emotional keyword extraction uses emotional dictionaries and keyword extraction algorithms to identify words that are closely related to emotional expression, such as "sadness", "anxiety", "excitement", etc. Topic model analysis is to discover potential topic structures from a large amount of text to help understand the focus of user attention and emotional triggers; emotional intensity evaluation is to evaluate the emotional polarity and intensity of keywords in combination with the context, providing a more detailed perspective for sentiment analysis.

[0144] The sentiment analysis in step S2 of this embodiment is specifically as follows:

[0145] S201. Selecting a medical big model: Use BERT and GPT-3 advanced medical big models. Medical big models have been pre-trained on large-scale text data, have strong language understanding and generation capabilities, and have been fine-tuned in the medical and mental health fields to adapt to specific sentiment analysis tasks;

[0146] S202, model fine-tuning and domain adaptation: data set preparation, fine-tuning and domain adaptation; data set preparation is to collect labeled data in the medical and mental health fields of patient diaries, forum discussions, and medical records, and build a domain data set specifically for sentiment analysis; fine-tuning is to use the labeled data in the medical and mental health fields to fine-tune the medical big model, optimize the model parameters, and enable the medical big model to better understand the emotional expression in the medical context; domain adaptation is to enable the medical big model to migrate from the source domain (such as general text) to the target domain (such as medical sentiment analysis) through contrastive learning and transfer learning strategies;

[0147] S203, Multimodal emotion analysis: Integrate multimodal data of voice, video and physiological signals, specifically: Voice emotion recognition: Use acoustic features and voice recognition technology to analyze emotional features in voice, such as intonation, pitch and rhythm; Visual emotion analysis: Analyze facial expressions and body language, and use computer vision technology for emotion recognition; Physiological signal analysis: Combine physiological indicators such as heart rate and blood pressure to assist in judging the user's emotional state;

[0148] S204. Emotional state recognition algorithm: hybrid model, sentiment analyzer, uncertainty quantification and dynamic emotion trajectory modeling; among them, the hybrid model combines the features of text, voice and physiological signals to build a hybrid emotion recognition model, such as a deep integrated network or a multi-task learning framework; the emotion classifier is a deep learning model that trains a convolutional neural network, a recurrent neural network or a Transformer architecture as an emotion classifier to identify multiple emotion categories such as joy, anger, fear and sadness; uncertainty quantification is the uncertainty predicted by the Bayesian method or the Monte Carlo dropout estimation model, providing a more reliable basis for decision-making; dynamic emotion trajectory modeling uses RNN or LSTM to capture the dynamic changes of user emotions over time to form an emotion trajectory model.

[0149] The specific details of building a personalized model in step S3 of this embodiment are as follows:

[0150] S301, time series analysis: using time series analysis methods to track and predict the user's emotional state, capture the periodic changes, trends and anomalies of the user's emotions, and help identify potential psychological problems and emotional fluctuation patterns; time series analysis methods include ARIMA (autoregressive integrated moving average model), LSTM (long short-term memory network) or Prophet algorithm; where the emotional fluctuation pattern includes the emotional timeline and the prediction of future emotional state; the emotional timeline is to construct the user's emotional timeline, showing the changes in the emotional state within the past set time, so as to identify patterns and anomalies; predicting the future emotional state is to use time series prediction to warn of possible negative emotional outbreaks in advance so as to intervene in time;

[0151] S302, Social Network Analysis: By analyzing the user's social network data, understand the user's emotional expression and interaction patterns in the social environment; specifically: Emotional Communication Analysis: Study the impact of the user's emotional communication in the social network, identify the pattern and key nodes of emotional contagion; Social Influence: Evaluate the user's influence in the social network, consider its role in emotional fluctuations; Friend Circle Analysis: Analyze the user's friend circle structure, and identify important people who have an impact on the user's emotions;

[0152] S303, Behavior pattern mining: Use machine learning and data mining technologies such as association rules, cluster analysis and deep learning to reveal the relationship between users' daily behaviors and emotional states; specifically: Activity pattern recognition: Identify users' daily activity patterns, such as sleep, diet, and exercise habits, which are related to their emotional states; Trigger event detection: Discover key events that cause emotional changes, such as work pressure, family problems, etc.;

[0153] S304, Behavior-emotion association: Analyze the association between specific user behaviors and specific emotions to provide a basis for personalized intervention;

[0154] S305. User portrait construction: construct the user's emotional portrait, which is a comprehensive report that details the user's emotional state, behavioral patterns, social network and life events; specifically: feature extraction: extract the user's emotional features, social features and behavioral features to form a multi-dimensional user feature vector; portrait construction: map the feature vector to the user portrait to form a visual overview of the user's emotions; portrait update: as user data increases, the user portrait is updated in real time to reflect the latest changes in the user's emotional state.

[0155] The specific details of generating the emotion intervention strategy in step S4 of this embodiment are as follows:

[0156] S401. Multi-factor decision-making algorithm: A multi-factor decision-making algorithm using fuzzy logic, genetic algorithm or rule-based reasoning system is used to comprehensively consider factors such as user emotional state, personal characteristics, life events, social network influence, and behavior patterns to generate emotional intervention strategies; wherein user emotional state is based on the results of the emotional analysis model to determine the user's current and predicted emotional state; personal characteristics include age, gender, occupation, health history, etc.; life events are life stresses, important events or changes within a set time range; social network influence refers to the impact of emotional states and interactions in social networks on users; and behavior patterns are user behavior habits and activity patterns;

[0157] S402, Dynamic Treatment Plan Update: The emotional intervention strategy is not static, but is dynamically updated based on user feedback and real-time data. When the user's emotional state changes or the intervention effect is not ideal, real-time monitoring, feedback learning and adaptive adjustment are carried out. Among them, real-time monitoring is to continuously monitor the user's emotional changes and adjust the plan according to new data; feedback learning refers to incorporating user feedback on treatment effects into model learning to optimize treatment strategies; adaptive adjustment is based on reinforcement learning, and dynamically adjusts the plan according to user behavior and treatment response;

[0158] S403, Personalized Psychological Therapy Recommendation: Based on the user profile and treatment plan, recommend personalized psychological therapies suitable for the user, including cognitive behavioral intervention, mindfulness meditation, online psychological counseling, self-help reading materials and applications; cognitive behavioral intervention is to adjust specific thinking patterns; mindfulness meditation is to teach users mindfulness skills to help relieve anxiety and stress; online psychological counseling is to recommend online counselors or intervention groups suitable for users; self-help reading materials are to provide relevant books, articles and guidance manuals; applications are to recommend mobile applications that help with emotion management;

[0159] S404, Self-help resource library: educational resources, exercises and activities, community support and mental health tools; among them, educational resources include psychology lectures, workshops and online courses; exercises and activities include meditation, breathing exercises, relaxation techniques; community support includes online forums and mutual aid groups to promote interaction and support among users; mental health tools include mood tracking tools and anxiety management tools to help users manage themselves.

[0160] The intelligent recommendation in step S5 of this embodiment is specifically as follows:

[0161] S501. Combination of collaborative filtering and deep learning: Integrate collaborative filtering and deep learning technologies to improve the accuracy and personalization of recommendations; specifically: User-user collaborative filtering: Analyze the behavior patterns of user groups, identify users with similar emotional needs, and recommend resources they have successfully used; Item-item collaborative filtering: Analyze the correlation between resources, and recommend similar resources based on the user's past response to any resource; Deep learning model: Use models such as convolutional neural network (CNN) or recurrent neural network (RNN) to learn users' emotional preferences and behavior patterns, and further improve the accuracy of recommendations;

[0162] S502, real-time recommendation algorithm: adopt real-time recommendation algorithm, such as stream-based recommendation system, which can generate personalized suggestions at the moment of user interaction with the platform; specifically: real-time sentiment analysis: instant analysis of the user's current emotional state to match the corresponding resources; dynamic update: as the user's behavior and emotional state change, the recommendation list is updated in real time; event-driven: respond to user login, browsing, search for specific events, and quickly push relevant resources;

[0163] S503, Personalized resource sorting: Personalized sorting of resources based on the user's emotional state, historical behavior, preferences and satisfaction; Rating prediction: Predict the user's preference for untouched resources and sort them according to the predicted score; Diversity consideration: Avoid over-recommendation of similar resources and ensure the diversity and novelty of the recommendation list; Interest evolution: Track changes in user interests and adjust the sorting strategy to reflect the latest needs;

[0164] S504, system learning and optimization: The specific learning and optimization strategies are as follows: Online learning: Collect user feedback during the actual recommendation process and continuously adjust model parameters; A / B testing: Compare the effects of different recommendation strategies and select the best solution; Meta-learning: By learning how to learn, quickly adapt to new users and new resources, and improve the generalization ability of the recommendation system; Reinforcement learning: Through interaction with users, the system learns the best recommendation strategy through rewards and punishments.

[0165] The feedback and learning mechanism in step S6 of this embodiment is specifically as follows:

[0166] S601, reinforcement learning application: including environment modeling, action selection, reward function and strategy iteration; environment modeling: considering user emotional state and platform behavior as the environment of reinforcement learning; action selection: recommending resources or intervention measures as actions of reinforcement learning; reward function: defining rewards based on indicators such as user satisfaction, resource usage and emotional improvement; strategy iteration: learning and optimizing recommendation strategies through multiple interactions to maximize long-term rewards;

[0167] S602, user feedback processing: including satisfaction survey, emotional feedback, complaints and suggestions; among them, satisfaction survey: regularly conduct satisfaction surveys to understand users' views on recommended resources and intervention measures; emotional feedback: users provide real-time feedback on emotional changes to help the system understand the intervention effect; complaints and suggestions: set up complaint and suggestion functions to collect users' opinions on platform improvements;

[0168] S603, continuous model update: including online update, offline evaluation and model fusion; online update: real-time update of model parameters without affecting services; offline evaluation: regular offline evaluation to compare the performance of new and old models; model fusion: combining the advantages of multiple models to build a hybrid model to improve overall performance;

[0169] S604, Expert knowledge base integration: including expert consultation, knowledge base update and knowledge graph; expert consultation: regularly invite experts for consultation to obtain the latest research results and clinical practices; knowledge base update: integrate expert advice and the latest research into the model to improve the scientificity and effectiveness of recommendations; knowledge graph: build a knowledge graph of emotional health to facilitate the model to understand and apply professional knowledge;

[0170] The privacy protection and security in step S7 of this embodiment are specifically as follows:

[0171] S701. Data encryption technology: including transport layer encryption, storage encryption, and key management, i.e., end-to-end encryption. Transport layer encryption: using SSL / TLS protocol to ensure data security during transmission; storage encryption: user data is encrypted and stored on the server, so that even if the data is stolen, it cannot be directly read; key management: following best practices, strictly managing and protecting encryption keys to prevent unauthorized access; end-to-end encryption: implementing end-to-end communication encryption between user devices and the platform to ensure that information is not attacked by middlemen during transmission;

[0172] S702. Anonymization: including differential privacy, desensitization technology and aggregate statistics; differential privacy: adding random noise to the data so that the analysis results cannot be traced back to individuals; desensitization technology: replacing or deleting information that can directly or indirectly identify an individual’s name, address and phone number; aggregate statistics: performing statistical analysis on anonymous data sets to provide overall trends without exposing individual information;

[0173] S703, Data security strategy: including access control, audit tracking, security updates and security training; among them, access control: implement multi-layer permission management to ensure that only authorized personnel have access to sensitive data; audit tracking: record all data access and modification operations so that they can be traced back when security incidents occur; security updates: regularly update systems and software to patch known security vulnerabilities; security training: provide data security awareness training to employees to enhance their awareness of prevention;

[0174] S704, Regulatory Compliance: including compliance review, user right to know, data minimization and user consent mechanism; among them, compliance review: regularly conduct privacy and security compliance review to ensure compliance with the latest regulatory requirements; user right to know: provide a clear privacy policy to inform users of how their data is collected, used and shared; data minimization principle: only collect data necessary for the service and avoid unnecessary data collection; user consent mechanism: obtain explicit consent from users before processing and using the corresponding personal data;

[0175] The user interface and interaction in step S8 of this embodiment are as follows:

[0176] S801, friendly UI design, has the following features:

[0177] ① Simple layout: clear page layout reduces user cognitive burden;

[0178] ② Color psychology: Use the principles of color psychology to create a visual environment that helps regulate emotions;

[0179] ③Touch-friendly: Optimized touch screen operation to adapt to various devices and screen sizes;

[0180] ④ Responsive design: automatically adjust the interface to suit different devices and resolutions;

[0181] S802, multi-language support, details are as follows:

[0182] ① Language switching: users select the interface language according to their preferences;

[0183] ② Localized content: resources and prompt information are localized and translated to ensure cultural adaptability;

[0184] ③ Voice Assistant: supports voice input and output in multiple languages, which is convenient for users with different language backgrounds;

[0185] S803, barrier-free functions, as follows:

[0186] ① Screen reader compatibility: optimize HTML structure to ensure compatibility with screen readers;

[0187] ②High contrast mode: Provides high contrast options for the convenience of visually impaired users;

[0188] ③Keyboard navigation: All functions can be operated by keyboard, without mouse;

[0189] ④Font adjustment: Allow users to customize font size and type to meet different vision needs;

[0190] S804, 24 / 7 online support, details are as follows:

[0191] ① Real-time chat: Integrate real-time chat function, users can quickly ask questions and get answers;

[0192] ② Self-service: set up FAQs and knowledge base to allow users to solve problems by themselves;

[0193] ③Manual customer service: Provide telephone and email support, and provide professional answers to complex questions;

[0194] ④ Community forum: encourage users to exchange experiences and support each other in the forum;

[0195] The integration and deployment in step S9 of this embodiment are specifically as follows:

[0196] S901, API interface design, as follows:

[0197] ①Open standards: follow the RESTful API design principles and use the JSON data format;

[0198] ② Authentication: Implement OAuth or other security authentication mechanisms to protect data security;

[0199] ③ Version control: Versioning APIs to ensure that upgrades do not affect existing integrations;

[0200] ④ Documentation: Provide detailed API documentation to simplify developers’ integration work;

[0201] S902, cloud platform architecture, as follows:

[0202] ①Containerization: Use Docker and Kubernetes to containerize services for easy deployment and management;

[0203] ② Microservice architecture: split into independent services to improve maintainability and scalability;

[0204] ③ Load balancing: Configure a load balancer to disperse traffic and prevent single point failure;

[0205] ④Distributed storage: Use Cassandra or MongoDB distributed database system to ensure data consistency;

[0206] S903, elastic expansion capability, as follows:

[0207] ① Dynamic expansion and contraction: Automatically expand or reduce computing resources by monitoring resource utilization;

[0208] ②Horizontal expansion: Increase the number of instances when needed to handle more concurrent requests;

[0209] ③ Resource scheduling: Intelligently schedule system resources to ensure that key services are given priority in resource allocation;

[0210] S904, high availability design, as follows:

[0211] ① Redundancy design: set up backup services and data to prevent the main service from failing;

[0212] ② Failure switching: When the main service fails, it can quickly switch to the backup service;

[0213] ③Regular backup and recovery: Back up data regularly to ensure data security and support rapid recovery;

[0214] ④Health Check: Implement service health check to promptly identify and resolve potential problems.

[0215] Embodiment 2:

[0216] As attached Figure 2 As shown, this embodiment provides an emotional intelligence intervention and personalized recommendation system based on a large model, which is used to implement the emotional intelligence intervention and personalized recommendation method based on a large model in the medical industry as in Example 1; the system includes:

[0217] The data collection and preprocessing module is used to collect relevant data in multiple dimensions, and extract emotional keywords and themes after data cleaning and structure processing of the collected relevant data;

[0218] Sentiment analysis module, which is used to fine-tune pre-trained models such as BERT or GPT-3 to adapt to medical emotion recognition;

[0219] Personalized model building module, used to customize user emotion intervention strategies through time series analysis, social network analysis, behavior pattern mining and user portrait construction;

[0220] The emotion intervention strategy generation module is used to dynamically update and recommend personalized emotion intervention strategies using a multi-factor decision-making algorithm;

[0221] Intelligent recommendation module, which combines collaborative filtering and deep learning to achieve real-time resource recommendations and improve recommendation results through learning and optimization;

[0222] Feedback and learning mechanism module, which is used to continuously improve model performance by using reinforcement learning, user feedback processing and continuous model updating;

[0223] Privacy protection and security module, used to apply data encryption technology, anonymization processing and data security strategies;

[0224] User Interface and Interaction Design module to include UI, multi-language support, accessibility features, and 24 / 7 online support;

[0225] The architecture integration and deployment module is used to design API interface design, cloud platform deployment, elastic expansion capabilities and high availability design.

[0226] Embodiment 3:

[0227] This embodiment also provides an electronic device, including: a memory and at least one processor;

[0228] Wherein, the memory stores computer-executable instructions;

[0229] The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the emotional intelligence intervention and personalized recommendation method based on a big model in the medical industry as described in any one of the present inventions.

[0230] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor may be a microprocessor or any conventional processor, etc.

[0231] The memory can be used to store computer programs and / or modules. The processor realizes various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function, etc.; the data storage area can store data created according to the use of the terminal, etc. In addition, the memory can also include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SMC), a secure digital (SD) card, a flash memory card, at least one disk storage period, a flash memory device, or other volatile solid-state storage devices.

[0232] Embodiment 4:

[0233] This embodiment also provides a computer-readable storage medium, which stores a plurality of instructions, which are loaded by a processor to enable the processor to execute the emotional intelligence intervention and personalized recommendation method based on a large model in the medical industry in any embodiment of the present invention. Specifically, a system or device equipped with a storage medium can be provided, on which a software program code that implements the functions of any of the above embodiments is stored, and a computer (or CPU or MPU) of the system or device reads and executes the program code stored in the storage medium.

[0234] In this case, the program code itself read from the storage medium can realize the function of any one of the above-mentioned embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present invention.

[0235] The storage medium embodiments for providing the program code include a floppy disk, a hard disk, a magneto-optical disk, an optical disk (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), a magnetic tape, a non-volatile memory card, and a ROM. Alternatively, the program code can be downloaded from a server computer by a communication network.

[0236] In addition, it should be clear that the functions of any of the above embodiments can be implemented not only by executing the program code read by the computer, but also by enabling an operating system operating on the computer to complete part or all of the actual operations based on instructions from the program code.

[0237] In addition, it can be understood that the program code read from the storage medium is written to a memory provided in an expansion board inserted into the computer or written to a memory provided in an expansion unit connected to the computer, and then based on the instructions of the program code, a CPU installed on the expansion board or the expansion unit is enabled to perform part or all of the actual operations, thereby realizing the functions of any of the above-mentioned embodiments.

[0238] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention 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 replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method of emotional intelligence intervention and personalized recommendation based on a large model in the medical industry, characterized in that: The method is as follows: Data collection and preprocessing: Collect relevant data from multiple dimensions, clean and structure the collected data, and extract emotional keywords and themes; Sentiment analysis: Use pre-trained models such as BERT or GPT-3 and fine-tune them to suit medical sentiment recognition; Build personalized models: customize user emotion intervention strategies through time series analysis, social network analysis, behavior pattern mining, and user portrait construction; Generate emotional intervention strategies: Use a multi-factor decision-making algorithm to dynamically update and recommend personalized emotional intervention strategies; Intelligent recommendation: Combine collaborative filtering and deep learning to achieve real-time resource recommendations, and improve recommendation results through learning and optimization; Feedback and learning mechanism: Utilize reinforcement learning, user feedback processing and continuous model updates to continuously improve model performance; Privacy protection and security: Apply data encryption technology, anonymization processing and data security strategies; User interface and interaction design: including UI, multi-language support, accessibility and 24 / 7 online support; Integration and deployment architecture: design API interface design, cloud platform deployment, elastic expansion capabilities and high availability design.

2. The method for emotional intelligence intervention and personalized recommendation based on a large model in the medical industry according to claim 1 is characterized in that: The data collection and preprocessing are as follows: Online communication records, personal diaries and logs, questionnaire survey results, physiological and environmental sensor data, and voice recordings are used to achieve data diversification and integration. Among them, online communication records are obtained by analyzing users' chat records, emails, and social media posts on the platform to capture emotional expressions and psychological state clues in daily communication; personal diaries and logs are used to encourage users to record their daily feelings and understand the changes in the user's inner world through text analysis; questionnaire survey results are obtained by designing professional mental health assessment questionnaires and regularly collecting user self-assessment reports on depression scales and anxiety scales; physiological and environmental sensor data are obtained by integrating biological indicator data and environmental data provided by wearable devices, which indirectly reflect the user's emotional state and living environment; voice recordings are obtained by analyzing the emotional color and tone changes in the user's voice through voice recognition technology to supplement the lack of text information; Data cleaning and structural processing: denoising, duplicate data processing, standardization and unification processing, and missing value processing; among them, denoising is to filter out irrelevant characters, links and emoticons using regular expressions and natural language processing technology; duplicate data processing is to remove duplicate records through hash comparison to maintain the uniqueness of the data set; standardization and unification processing is to unify the data formats from different sources for subsequent analysis. The data format includes date format and text encoding format; missing value processing is to use mean filling and model prediction filling interpolation technology to process missing data to ensure data integrity; Sentiment keyword and topic extraction: Natural language processing technology is used to extract sentiment keywords and topics, perform topic model analysis, and evaluate sentiment intensity. Among them, sentiment keyword extraction uses sentiment dictionaries and keyword extraction algorithms to identify words that are closely related to sentiment expression. Topic model analysis discovers potential topic structures from a large amount of text to help understand the focus of user attention and emotional triggers. Sentiment intensity evaluation evaluates the sentiment polarity and intensity of keywords in combination with context, providing a more detailed perspective for sentiment analysis.

3. The method for emotional intelligence intervention and personalized recommendation based on a large model in the medical industry according to claim 1 is characterized in that: The sentiment analysis is as follows: Selecting medical big models: Using BERT and GPT-3 advanced medical big models, which have been pre-trained on large-scale text data, have strong language understanding and generation capabilities, and have been fine-tuned in the medical and mental health fields to adapt to specific sentiment analysis tasks; Model fine-tuning and domain adaptation: data set preparation, fine-tuning and domain adaptation; data set preparation is to collect labeled data in the medical and mental health fields from patient diaries, forum discussions, and medical records, and build a domain data set specifically for sentiment analysis; fine-tuning is to use labeled data in the medical and mental health fields to fine-tune the medical big model, optimize model parameters, and enable the medical big model to better understand emotional expressions in the medical context; domain adaptation is to enable the medical big model to migrate from the source domain to the target domain through comparative learning and transfer learning strategies; Multimodal emotion analysis: Integrate multimodal data of speech, video and physiological signals, specifically: Speech emotion recognition: Use acoustic features and speech recognition technology to analyze emotional features in speech, such as intonation, pitch and rhythm; Visual emotion analysis: Analyze facial expressions and body language, and use computer vision technology for emotion recognition; Physiological signal analysis: Combine physiological indicators such as heart rate and blood pressure to assist in judging the user's emotional state; Emotional state recognition algorithm: hybrid model, sentiment analyzer, uncertainty quantification and dynamic emotion trajectory modeling; among them, the hybrid model combines the features of text, voice and physiological signals to build a hybrid emotion recognition model, such as a deep integrated network or a multi-task learning framework; the sentiment classifier is a deep learning model that trains convolutional neural networks, recurrent neural networks or Transformer architectures as sentiment classifiers to identify multiple emotion categories such as joy, anger, fear and sadness; uncertainty quantification is the uncertainty predicted by the Bayesian method or Monte Carlo dropout estimation model, providing a more reliable basis for decision-making; dynamic emotion trajectory modeling uses RNN or LSTM to capture the dynamic changes of user emotions over time to form an emotion trajectory model.

4. The method for emotional intelligence intervention and personalized recommendation based on a large model in the medical industry according to claim 1 is characterized in that: The details of building a personalized model are as follows: Time series analysis: Use time series analysis methods to track and predict the user's emotional state, capture the periodic changes, trends and anomalies of the user's emotions, and help identify potential psychological problems and emotional fluctuation patterns; time series analysis methods include ARIMA, LSTM or Prophet algorithms; among them, emotional fluctuation patterns include emotional timelines and predicted future emotional states; emotional timelines are to construct the user's emotional timeline, showing the changes in emotional states within a set time in the past, so as to identify patterns and anomalies; predicting future emotional states is to use time series predictions to warn of possible negative emotional outbreaks in advance so as to intervene in time; Social network analysis: By analyzing the user's social network data, understand the user's emotional expression and interaction patterns in the social environment; specifically: Emotional communication analysis: study the impact of users' emotional communication in social networks, identify the patterns and key nodes of emotional contagion; Social influence: evaluate the influence of users in social networks, and consider their role in emotional fluctuations; Friends circle analysis: analyze the structure of users' friends circles and identify important people who have an impact on users' emotions; Behavior pattern mining: using machine learning and data mining techniques such as association rules, cluster analysis and deep learning to reveal the relationship between users’ daily behaviors and emotional states; specifically: activity pattern recognition: identifying users’ daily activity patterns, which are related to their emotional states; trigger event detection: discovering key events that lead to emotional changes; Behavior-emotion correlation: Analyze the correlation between specific user behaviors and specific emotions to provide a basis for personalized intervention; User portrait construction: construct the user's emotional portrait, which is a comprehensive report that details the user's emotional state, behavioral patterns, social network and life events; specifically: feature extraction: extract the user's emotional features, social features and behavioral features to form a multi-dimensional user feature vector; portrait construction: map the feature vector to the user portrait to form a visual overview of the user's emotions; portrait update: as user data increases, the user portrait is updated in real time to reflect the latest changes in the user's emotional state.

5. The method of emotional intelligence intervention and personalized recommendation based on a large model in the medical industry according to claim 1, characterized in that: Generate emotional intervention strategies as follows: Multi-factor decision-making algorithm: Fuzzy logic, genetic algorithm or rule-based reasoning system multi-factor decision-making algorithm is used to comprehensively consider factors such as user emotional state, personal characteristics, life events, social network influence, and behavior patterns to generate emotional intervention strategies; among them, user emotional state is based on the results of the sentiment analysis model to determine the user's current and predicted emotional state; personal characteristics include age, gender, occupation, health history, etc.; life events are life stress, important events or changes within a set time range; social network influence refers to the impact of emotional state and interaction in the social network on users; behavior patterns are users' behavioral habits and activity patterns; Dynamic treatment plan update: Dynamic updates based on user feedback and real-time data; when the user's emotional state changes or the intervention effect is not ideal, real-time monitoring, feedback learning and adaptive adjustment are carried out; real-time monitoring is to continuously monitor the user's emotional changes and adjust the plan according to new data; feedback learning refers to incorporating user feedback on treatment effects into model learning to optimize treatment strategies; adaptive adjustment is based on reinforcement learning, dynamically adjusting the plan according to user behavior and treatment response; Personalized psychotherapy recommendations: Based on the user profile and treatment plan, personalized psychotherapy suitable for the user is recommended, including cognitive behavioral intervention, mindfulness meditation, online psychological counseling, self-help reading materials and applications; cognitive behavioral intervention is to adjust specific thinking patterns; mindfulness meditation is to teach users mindfulness skills to help relieve anxiety and stress; online psychological counseling is to recommend online counselors or intervention groups suitable for users; self-help reading materials are to provide relevant books, articles and guidance manuals; applications are to recommend mobile applications that help with emotion management; Self-help resource library: educational resources, exercises and activities, community support, and mental health tools; among them, educational resources include psychology lectures, workshops, and online courses; exercises and activities include meditation, breathing exercises, and relaxation techniques; community support includes online forums and mutual aid groups to promote interaction and support among users; mental health tools include mood tracking tools and anxiety management tools to help users manage themselves.

6. The method of emotional intelligence intervention and personalized recommendation based on a large model in the medical industry according to claim 1, characterized in that: Smart recommendations are as follows: Combination of collaborative filtering and deep learning: Integrate collaborative filtering and deep learning technologies to improve the accuracy and personalization of recommendations; specifically: User-user collaborative filtering: Analyze the behavior patterns of user groups, identify users with similar emotional needs, and recommend resources they have successfully used; Item-item collaborative filtering: Analyze the correlation between resources, and recommend similar resources based on the user's past response to any resource; Deep learning model: Use models such as convolutional neural networks or recurrent neural networks to learn users' emotional preferences and behavior patterns, and further improve the accuracy of recommendations; Real-time recommendation algorithm: The real-time recommendation algorithm can generate personalized suggestions at the moment when users interact with the platform; Specifically: Real-time sentiment analysis: Instantly analyze the user's current emotional state to match the corresponding resources; Dynamic update: As the user's behavior and emotional state change, the recommendation list is updated in real time; Event-driven: Respond to user login, browsing, and search for specific events to quickly push relevant resources; Personalized resource sorting: Sort resources individually based on the user's emotional state, historical behavior, preferences, and satisfaction; Rating prediction: Predict the user's preference for untouched resources and sort them based on the predicted scores; Diversity considerations: Avoid over-recommendation of similar resources and ensure the diversity and novelty of the recommendation list; Interest evolution: Track changes in user interests and adjust the sorting strategy to reflect the latest needs; System learning and optimization: The specific learning and optimization strategies are as follows: Online learning: Collect user feedback during the actual recommendation process and continuously adjust model parameters; A / B testing: Compare the effects of different recommendation strategies and select the best solution; Meta-learning: By learning how to learn, it can quickly adapt to new users and new resources and improve the generalization ability of the recommendation system; Reinforcement learning: Through interaction with users, the system learns the best recommendation strategy through rewards and penalties.

7. The method for emotional intelligence intervention and personalized recommendation based on a large model in the medical industry according to any one of claims 1 to 6, characterized in that: The feedback and learning mechanism is as follows: Reinforcement learning applications: including environment modeling, action selection, reward function and strategy iteration; environment modeling: considering user emotional state and platform behavior as the environment for reinforcement learning; action selection: recommending resources or intervention measures as actions for reinforcement learning; reward function: defining rewards based on indicators such as user satisfaction, resource usage and emotional improvement; strategy iteration: learning and optimizing recommendation strategies through multiple interactions to maximize long-term rewards; User feedback processing: including satisfaction surveys, emotional feedback, complaints and suggestions; among them, satisfaction surveys: regularly conduct satisfaction surveys to understand users' views on recommended resources and intervention measures; emotional feedback: users provide real-time feedback on emotional changes to help the system understand the effects of interventions; complaints and suggestions: set up complaint and suggestion functions to collect users' opinions on improving the platform; Continuous model update: including online update, offline evaluation and model fusion; online update: real-time update of model parameters without affecting services; offline evaluation: regular offline evaluation to compare the performance of new and old models; model fusion: combining the advantages of multiple models to build a hybrid model to improve overall performance; Expert knowledge base integration: including expert consultation, knowledge base update and knowledge graph; expert consultation: regularly invite experts for consultation to obtain the latest research results and clinical practices; knowledge base update: integrate expert advice and the latest research into the model to improve the scientificity and effectiveness of recommendations; knowledge graph: build a knowledge graph of emotional health to facilitate the model to understand and apply professional knowledge; Privacy protection and security are as follows: Data encryption technology: including transport layer encryption, storage encryption, key management, i.e. end-to-end encryption; transport layer encryption: using SSL / TLS protocol to ensure data security during transmission; storage encryption: user data is encrypted and stored on the server, so that even if the data is stolen, it cannot be directly read; key management: following best practices, strictly managing and protecting encryption keys to prevent unauthorized access; end-to-end encryption: achieving end-to-end communication encryption between user devices and the platform to ensure that information is not attacked by middlemen during transmission; Anonymization: including differential privacy, desensitization technology and aggregate statistics; differential privacy: adding random noise to the data so that the analysis results cannot be traced back to individuals; desensitization technology: replacing or deleting information that can directly or indirectly identify personal names, addresses and phone numbers; aggregate statistics: performing statistical analysis on anonymous data sets to provide overall trends without exposing individual information; Data security strategy: including access control, audit tracking, security updates and security training; among them, access control: implement multi-layer permission management to ensure that only authorized personnel can access sensitive data; audit tracking: record all data access and modification operations so that they can be traced back when security incidents occur; security updates: regularly update systems and software to patch known security vulnerabilities; security training: provide data security awareness training to employees to enhance their awareness of prevention; Regulatory compliance: including compliance review, user right to know, data minimization and user consent mechanism; among them, compliance review: regularly conduct privacy and security compliance reviews to ensure compliance with the latest regulatory requirements; user right to know: provide a clear privacy policy to inform users of how their data is collected, used and shared; data minimization principle: only collect data necessary for the service and avoid unnecessary data collection; user consent mechanism: obtain explicit consent from users before processing and using the corresponding personal data; The user interface and interaction are as follows: Friendly UI design, with the following features: ① Simple layout: clear page layout reduces user cognitive burden; ② Color psychology: Use the principles of color psychology to create a visual environment that helps regulate emotions; ③Touch-friendly: Optimized touch screen operation to adapt to various devices and screen sizes; ④Responsive design: automatically adjust the interface to suit different devices and resolutions; Multi-language support, as follows: ① Language switching: users select the interface language according to their preferences; ② Localized content: resources and prompt information are localized and translated to ensure cultural adaptability; ③ Voice Assistant: supports voice input and output in multiple languages, which is convenient for users with different language backgrounds; Accessibility features, as follows: ① Screen reader compatibility: optimize HTML structure to ensure compatibility with screen readers; ②High contrast mode: Provides high contrast options for the convenience of visually impaired users; ③Keyboard navigation: All functions can be operated by keyboard, without mouse; ④Font adjustment: Allow users to customize font size and type to meet different vision needs; 24 / 7 online support, details as follows: ① Real-time chat: Integrate real-time chat function, users can quickly ask questions and get answers; ② Self-service: set up FAQs and knowledge base to allow users to solve problems by themselves; ③Manual customer service: Provide telephone and email support, and provide professional answers to complex questions; ④ Community forum: encourage users to exchange experiences and support each other in the forum; Integration and deployment, as follows: The API interface design is as follows: ①Open standards: follow the RESTful API design principles and use the JSON data format; ② Authentication: Implement OAuth or other security authentication mechanisms to protect data security; ③ Version control: Versioning APIs to ensure that upgrades do not affect existing integrations; ④ Documentation: Provide detailed API documentation to simplify developers’ integration work; The cloud platform architecture is as follows: ①Containerization: Use Docker and Kubernetes to containerize services for easy deployment and management; ② Microservice architecture: split into independent services to improve maintainability and scalability; ③ Load balancing: Configure a load balancer to disperse traffic and prevent single point failure; ④Distributed storage: Use Cassandra or MongoDB distributed database system to ensure data consistency; Elastic expansion capabilities, as follows: ① Dynamic expansion and contraction: Automatically expand or reduce computing resources by monitoring resource utilization; ②Horizontal expansion: Increase the number of instances when needed to handle more concurrent requests; ③ Resource scheduling: Intelligently schedule system resources to ensure that key services are given priority in resource allocation; High availability design, as follows: ① Redundancy design: set up backup services and data to prevent the main service from failing; ② Failure switching: When the main service fails, it can quickly switch to the backup service; ③Regular backup and recovery: Back up data regularly to ensure data security and support rapid recovery; ④Health Check: Implement service health check to promptly identify and resolve potential problems.

8. An emotional intelligence intervention and personalized recommendation system based on a large model in the medical industry, characterized by: The system is used to implement the emotional intelligence intervention and personalized recommendation method based on a large model in the medical industry as described in any one of claims 1 to 7; The system includes: The data collection and preprocessing module is used to collect relevant data in multiple dimensions, and extract emotional keywords and themes after data cleaning and structure processing of the collected relevant data; Sentiment analysis module, which uses pre-trained models such as BERT or GPT-3 and fine-tunes them to adapt to medical emotion recognition; Personalized model building module, used to customize user emotion intervention strategies through time series analysis, social network analysis, behavior pattern mining and user portrait construction; The emotion intervention strategy generation module is used to dynamically update and recommend personalized emotion intervention strategies using a multi-factor decision-making algorithm; Intelligent recommendation module, which combines collaborative filtering and deep learning to achieve real-time resource recommendations and improve recommendation results through learning and optimization; Feedback and learning mechanism module, which is used to continuously improve model performance by using reinforcement learning, user feedback processing and continuous model updating; Privacy protection and security module, used to apply data encryption technology, anonymization processing and data security strategies; User Interface and Interaction Design module to include UI, multi-language support, accessibility features, and 24 / 7 online support; The architecture integration and deployment module is used to design API interface design, cloud platform deployment, elastic expansion capabilities and high availability design.

9. An electronic device, characterized in that: include: memory and at least one processor; Wherein, the memory stores computer-executable instructions; The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the big model-based emotional intelligence intervention and personalized recommendation method for the medical industry as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the processor executes the computer-executable instructions, the emotional intelligence intervention and personalized recommendation method based on a large model in the medical industry as described in any one of claims 1 to 7 is implemented.

Citation Information

Cited By

  • Intelligent question-answering robot system based on online knowledge community

    CN119179768A

  • Intelligent learning behavior analysis method and system

    CN119295275A

  • Psychological crisis early warning intervention system and method based on behavior analysis

    CN119517316A

  • Personalized recommendation system based on semantic analysis

    CN119579287A

  • Personalized recommendation method based on tensor decomposition and neural network fusion

    CN119719510A