Intelligent classification early warning method and system for psychological health risk of college students
Through the intelligent classification early warning method based on multi-layer perceptron neural network, the problems of complex data processing, insufficient model performance, poor real-time and insufficient security in college students' mental health risk warning are solved, and accurate, real-time and safe risk warning is achieved. Improves model stability and data security.
Patent Information
- Application Number
- CN202411495390.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-25
- Publication Date
- 2024-11-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing mental health risk warning methods for college students have problems such as complex data processing, insufficient model performance optimization, poor real-time performance, and insufficient security and privacy protection, making it difficult to achieve accurate, real-time and safe risk classification warning.
Using intelligent classification early warning method based on multi-layer perceptron (MLP) neural network, the steps provided by data preprocessing, model training, online reasoning and interfaces include standardized processing, cross-validation, sample balance, L2 regularization, outlier detection , permission management and data encryption to build an efficient and secure early warning system.
It realizes accurate, real-time and safe classified warnings on mental health risks of college students, improves the generalization ability and data quality of the model, ensures the security of data transmission and user privacy protection, and provides a transparent risk assessment report.
Smart Images

Figure CN119028574A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of early warning of mental health risks, and in particular to an intelligent classification early warning method and system for college students' mental health risks. Background Art
[0002] With the development of society and the increase of competitive pressure, the mental health problems of college students are becoming increasingly serious. Studies have shown that mental health problems not only affect college students' academic performance, but may also have a significant impact on their future career development and quality of life. Therefore, timely and effective detection and early warning of college students' mental health risks are of great significance for promoting college students' mental health, improving education quality and social stability.
[0003] At present, there are mainly the following types of early warning methods for college students’ mental health risks: Psychological test questionnaire: The traditional method is to evaluate the mental health of college students by regularly conducting psychological test questionnaires. Although this method is simple and easy to implement, it has some problems, such as the test results rely on the subjective answers of students, which are easily affected by their emotional state and psychological defense mechanism, resulting in inaccurate and incomplete results.
[0004] Manual intervention: School psychological counseling centers help students with psychological problems through manual intervention and psychological counseling. However, manual intervention has problems such as limited human resources, insufficient coverage, and inability to monitor in real time, making it difficult to effectively manage and warn large groups of students.
[0005] Data-driven machine learning methods: In recent years, with the development of big data and artificial intelligence technology, data-driven machine learning methods have gradually been applied to the field of mental health risk warning. These methods analyze students' behavioral data, social data, learning data and other multi-source data to establish mental health risk prediction models, which can achieve more objective and accurate risk warnings.
[0006] However, existing data-driven methods still have some problems in practical applications: Complex data processing: There are many types of data related to college students' mental health, including psychological test data, basic information, daily behavior data, etc. The data preprocessing and feature extraction process is complex, which can easily lead to data quality problems such as outliers, missing values, and duplicate values, affecting the accuracy of the model.
[0007] Insufficient model performance optimization: Existing methods often do not make full use of cross-validation, sample balance, and regularization techniques during model training, resulting in insufficient generalization and stability of the model, making it difficult to cope with complex and changing practical application scenarios.
[0008] Poor real-time performance: After model training is completed, most existing methods lack an efficient online reasoning mechanism and cannot perform real-time analysis and early warning on newly input data, which limits their immediacy and effectiveness in practical applications.
[0009] Insufficient security and privacy protection: Mental health data is sensitive information. Existing methods often lack comprehensive security and privacy protection measures during data transmission and storage, which can easily lead to data leakage and privacy infringement. Summary of the invention
[0010] The purpose of this invention is to propose an intelligent classification and early warning method and system for college students' mental health risks based on a multi-layer perceptron (MLP) neural network, which can achieve accurate, real-time and safe classification and early warning of college students' mental health risks through data preprocessing, model training, online reasoning and interface provision.
[0011] In order to solve the above technical problems, the present invention provides an intelligent classification and early warning method for college students' mental health risks. The method is implemented based on a multi-layer perceptron (MLP) neural network model and includes the following steps: Data collection and preprocessing steps: Collect psychological test data, basic information and daily behavior data of college students, and construct comprehensive feature vectors as model input through standardization, coding and weighted preprocessing operations; Model training steps: Based on the multi-layer perceptron (MLP) neural network architecture, the model is optimized through cross-validation, sample balancing and L2 regularization technology. The model is trained using PyTorch built-in modules and custom modules. The model can also be retrained through the provided training interface to improve the classification warning performance. Online reasoning step: Perform real-time reasoning and classification on the newly input student data, and output the student's mental health risk level based on the model, which includes four levels from 0 to 3; Interface provision steps: Provide RESTful API interface through FastAPI framework, support single sample detection, batch sample evaluation, model training and model download; Among them, the standardization processing includes mean normalization and variance normalization of the input data, the encoding includes one-hot encoding of the category data, the weighted processing includes adjusting the importance of different features according to specific weight coefficients, and the construction of the feature vector includes integrating the processed data into a unified vector form.
[0012] As a preferred implementation scheme of the intelligent classification and early warning method for college students' mental health risks of the present invention, the data collection and preprocessing steps also include outlier detection and processing to remove or correct outliers in the data, thereby improving data quality.
[0013] As a preferred implementation scheme of the intelligent classification and early warning method for college students' mental health risks of the present invention, the cross-validation technology used in the model training step includes K-fold cross-validation, where the K value is 10, and the stability and generalization ability of the model are ensured through multiple training and validation.
[0014] As a preferred implementation scheme of the intelligent classification and early warning method for college students' mental health risks of the present invention, when classifying new input data in the online reasoning step, a soft voting method is used to combine multiple prediction results to improve the accuracy and robustness of the classification.
[0015] As a preferred implementation scheme of the intelligent classification and early warning method for college students' mental health risks of the present invention, the interface providing step also includes authority management and data encryption to ensure the security of data transmission and the protection of user privacy.
[0016] The present invention provides an intelligent classification and early warning system for college students' mental health risks. The system is based on machine learning technology and uses a multi-layer perceptron (MLP) neural network model for classification and early warning, including: a) Data preprocessing module: used to process raw data, including standardization, encoding and weighted processing preprocessing operations, and construct a comprehensive feature vector as model input; b) Model training module: Based on the multi-layer perceptron (MLP) neural network architecture, the model is optimized through cross-validation, sample balancing and L2 regularization technology, and the model can be retrained through the provided training interface to improve the classification warning performance; c) Online reasoning module: used to perform real-time reasoning and classification of newly input student data and output the student's mental health risk level (four levels from 0 to 3); d) Interface module: provides RESTful API interface, supports single sample detection, batch sample evaluation, model training and model download; Among them, the data preprocessing module includes an outlier detection and processing submodule, which is used to remove or correct outliers in the data; the cross-validation technology used in the model training module includes K-fold cross-validation, where the K value is 10; the online reasoning module includes a soft voting method submodule, which is used to combine multiple prediction results to improve classification accuracy; the interface module includes permission management and data encryption submodules to ensure data transmission security and user privacy protection.
[0017] As a preferred implementation scheme of the intelligent classification and early warning system for college students' mental health risks of the present invention, the data preprocessing module also includes a data cleaning submodule for automatically identifying and processing missing values and duplicate values in the data set, thereby improving the integrity and consistency of the data.
[0018] As a preferred implementation scheme of the intelligent classification and early warning system for college students' mental health risks of the present invention, the model training module also includes a parameter tuning submodule, which automatically tunes the hyperparameters of the model based on grid search and random search techniques to find the optimal parameter combination.
[0019] As a preferred implementation scheme of the intelligent classification and early warning system for college students' mental health risks of the present invention, the online reasoning module also includes a result explanation submodule, which explains the classification results based on SHAP value (Shapley Additive Explanations) or LIME (Local Interpretable Model-agnostic Explanations) technology to provide a more transparent risk assessment report.
[0020] As a preferred implementation scheme of the intelligent classification and early warning system for college students' mental health risks of the present invention, the interface module also includes a log recording submodule for recording API call logs, error logs and performance logs to facilitate system monitoring and troubleshooting.
[0021] Beneficial Effects Through the above technical scheme, the present invention can effectively solve the problems in the prior art such as complex data processing, insufficient model performance optimization, poor real-time performance, and insufficient security and privacy protection, and provides an efficient, accurate and safe solution for intelligent classification and early warning of college students' mental health risks. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 is a schematic diagram of a flow chart of Embodiment 1 of the present invention; Figure 2 It is a structural diagram of embodiment 2 of the present invention. DETAILED DESCRIPTION
[0023] Example 1 Invent a method for intelligent classification and early warning of college students' mental health risks, see Figure 1 ,This method is implemented based on a multi-layer perceptron (MLP) neural network model and includes the following steps: Data collection and preprocessing steps: Collect psychological test data, basic information and daily behavior data of college students, and construct comprehensive feature vectors as model input through standardization, coding and weighted preprocessing operations; Model training steps: Based on the multi-layer perceptron (MLP) neural network architecture, the model is optimized through cross-validation, sample balancing and L2 regularization technology. The model is trained using PyTorch built-in modules and custom modules. The model can also be retrained through the provided training interface to improve the classification warning performance. Online reasoning step: Perform real-time reasoning and classification on the newly input student data, and output the student's mental health risk level based on the model, which includes four levels from 0 to 3; Interface provision steps: Provide RESTful API interface through FastAPI framework, support single sample detection, batch sample evaluation, model training and model download; Among them, the standardization processing includes mean normalization and variance normalization of the input data, the encoding includes one-hot encoding of the category data, the weighted processing includes adjusting the importance of different features according to specific weight coefficients, and the construction of the feature vector includes integrating the processed data into a unified vector form.
[0024] Furthermore, the data collection and preprocessing steps also include outlier detection and processing to remove or correct outliers in the data, thereby improving data quality.
[0025] Furthermore, the cross-validation technique used in the model training step includes K-fold cross-validation, where the K value is 10, and the stability and generalization ability of the model are ensured through multiple training and validation.
[0026] Furthermore, when classifying new input data in the online reasoning step, a soft voting method is used to combine multiple prediction results to improve the accuracy and robustness of the classification.
[0027] Furthermore, the interface providing step also includes rights management and data encryption to ensure the security of data transmission and the protection of user privacy.
[0028] Example 2 The present invention provides an intelligent classification early warning system for college students' mental health risks, see Figure 2 The system is based on machine learning technology and uses a multi-layer perceptron (MLP) neural network model for classification warning, including: a) Data preprocessing module: used to process raw data, including standardization, encoding and weighted processing preprocessing operations, and construct a comprehensive feature vector as model input; b) Model training module: Based on the multi-layer perceptron (MLP) neural network architecture, the model is optimized through cross-validation, sample balancing and L2 regularization technology, and the model can be retrained through the provided training interface to improve the classification warning performance; c) Online reasoning module: used to perform real-time reasoning and classification of newly input student data and output the student's mental health risk level (four levels from 0 to 3); d) Interface module: provides RESTful API interface, supports single sample detection, batch sample evaluation, model training and model download; Among them, the data preprocessing module includes an outlier detection and processing submodule, which is used to remove or correct outliers in the data; the cross-validation technology used in the model training module includes K-fold cross-validation, where the K value is 10; the online reasoning module includes a soft voting method submodule, which is used to combine multiple prediction results to improve classification accuracy; the interface module includes permission management and data encryption submodules to ensure data transmission security and user privacy protection.
[0029] Furthermore, the data preprocessing module also includes a data cleaning submodule for automatically identifying and processing missing values and duplicate values in the data set, thereby improving the integrity and consistency of the data.
[0030] Furthermore, the model training module also includes a parameter tuning submodule, which automatically tunes the hyperparameters of the model based on grid search and random search techniques to find the optimal parameter combination.
[0031] Furthermore, the online reasoning module also includes a result explanation submodule, which explains the classification results based on SHAP value (Shapley Additive Explanations) or LIME (Local Interpretable Model-agnostic Explanations) technology to provide a more transparent risk assessment report.
[0032] Furthermore, the interface module also includes a log recording submodule for recording API call logs, error logs and performance logs to facilitate system monitoring and troubleshooting.
[0033] The preferred embodiments of the present invention described above are only the preferred embodiments of the present invention and are not intended to limit the present invention. The protection scope of the present invention is defined by the attached claims. For ordinary technicians in this field, other embodiments can be obtained according to the drawings without creative work, and any changes based on the claims of the present invention are within the protection scope of the present invention.
Claims
1. An intelligent classification and early warning method for college students' mental health risks, which is implemented based on a multi-layer perceptron neural network model and is characterized in that: The following steps are involved: Data collection and preprocessing steps: Collect psychological test data, basic information and daily behavior data of college students, and construct comprehensive feature vectors as model input through standardization, coding and weighted preprocessing operations; Model training steps: Based on the multi-layer perceptron neural network architecture, the model is optimized through cross-validation, sample balancing and L2 regularization technology. The model is trained using PyTorch built-in modules and custom modules. The model can also be retrained through the provided training interface to improve the classification warning performance. Online reasoning step: Perform real-time reasoning and classification on the newly input student data, and output the student's mental health risk level based on the model, which includes four levels from 0 to 3; Interface provision steps: Provide RESTful API interface through FastAPI framework, supporting single sample detection, batch sample evaluation, model training and model download.
2. The method according to claim 1, characterized in that: The data collection and preprocessing steps also include outlier detection and processing to remove or correct outliers in the data, thereby improving data quality.
3. The method according to claim 1, characterized in that: The cross-validation technique used in the model training step includes K-fold cross-validation, where the K value is 10, and the stability and generalization ability of the model are ensured through multiple training and validation.
4. The method according to claim 1, characterized in that: When classifying new input data in the online reasoning step, a soft voting method is used to combine multiple prediction results to improve the accuracy and robustness of the classification.
5. The method according to claim 1, characterized in that The interface providing step also includes rights management and data encryption to ensure the security of data transmission and the protection of user privacy.
6. An intelligent classification and early warning system for college students' mental health risks, based on machine learning technology, using a multi-layer perceptron neural network model for classification and early warning, characterized in that: include: a) Data preprocessing module: used to process raw data, including standardization, encoding and weighted processing preprocessing operations, and construct a comprehensive feature vector as model input; b) Model training module: Based on the multi-layer perceptron neural network architecture, the model is optimized through cross-validation, sample balancing and L2 regularization technology, and the model can be retrained through the provided training interface to improve the classification warning performance; c) Online reasoning module: used to perform real-time reasoning and classification on newly input student data and output the student’s mental health risk level; d) Interface module: provides RESTful API interface, supports single sample detection, batch sample evaluation, model training and model download; Among them, the data preprocessing module includes an outlier detection and processing submodule, which is used to remove or correct outliers in the data; the cross-validation technology used in the model training module includes K-fold cross-validation, where the K value is 10; the online reasoning module includes a soft voting method submodule, which is used to combine multiple prediction results to improve classification accuracy; the interface module includes permission management and data encryption submodules to ensure data transmission security and user privacy protection.
7. The system according to claim 6, characterized in that The data preprocessing module also includes a data cleaning submodule for automatically identifying and processing missing values and duplicate values in the data set, thereby improving the integrity and consistency of the data.
8. The system according to claim 6, characterized in that The model training module also includes a parameter tuning submodule, which automatically tunes the hyperparameters of the model based on grid search and random search techniques to find the optimal parameter combination.
9. The system according to claim 6, characterized in that The online reasoning module also includes a result explanation submodule, which explains the classification results based on SHAP value or LIME technology to provide a more transparent risk assessment report.
10. The system according to claim 6, characterized in that The interface module also includes a log recording submodule for recording API call logs, error logs and performance logs to facilitate system monitoring and troubleshooting.
Citation Information
Patent Citations
Neural network-based college students' psychological behavior abnormal change monitoring and early warning method
CN109256192A
Cited By
Risk behavior identification method, system and device based on decision tree and medium
CN120048511A
