基于机器学习的大学生心理问题早期发现研究
Research on Early Detection of College Students’ Psychological Problems Based on Machine Learning

本研究旨在通过机器学习算法实现大学生心理问题的早期发现。研究采用自编问卷和心理教师访谈的方式，对天津某高校1000名学生的心理状况进行评估，将其分为健康和亚健康两类。运用BP (Back Propagation)神经网络和极限学习机(ELM)算法分析问卷数据，构建预测模型，进而实现早期预警。结果显示，基于BP神经网络的模型在测试集200名学生中，准确识别出186名，总准确率达92.7%，其中亚健康识别率达到91.01%，显著优于极限学习机(ELM)模型。研究表明，BP神经网络在识别大学生心理亚健康方面具有较高的准确性和可靠性，为高校心理健康筛查提供了创新高效的新方法。
The study aims to achieve early detection of psychological issues among university students through machine learning algorithms. Using self-designed questionnaires and interviews with psychological counselors, the psychological status of 1,000 students at a university in Tianjin was assessed and classified into two categories: healthy and sub-healthy. Back Propagation (BP) neural network and Extreme Learning Machine (ELM) algorithms were employed to analyze the questionnaire data and construct predictive models, thereby enabling early warning. The results showed that the model based on the BP neural network accurately identified 186 out of 200 students in the test set, achieving an overall accuracy of 92.7%, with a sub-healthy identification rate of 91.01%. This significantly outperformed the Extreme Learning Machine (ELM) model. The study demonstrates that the BP neural network has high accuracy and reliability in identifying sub-healthy psychological states among university students, providing an innovative and efficient new method for mental health screening in universities.