Application of Machine Learning Algorithm-Assisted Intelligent Systems in College Student Mental Health Education: In View of Open Psychometrics Database

This study aims to improve the efficiency and effect of college students’ mental health management and promote the healthy development of students’ body and mind. Firstly, this study uses the data in the Open Psychometrics database, and adopts Category Boosting (CatBoost) in the machine learning algorithm to build a prediction model of college students’ mental health to identify high-risk individuals who may have psychological problems. Secondly, the Genetic Algorithm (GA) is used to optimize the model parameters. A GA-CatBoost intelligent identification system is constructed, which is compared with Random Forest (RF), K-nearest neighbors (KNN), Support Vector Machine (SVM), C4.5, Deep Learning-based Psychological Health Prediction Model (DeepPsy), Extreme Gradient Boosting, (XGBoost) and light gradient boosting machine (LightGBM), and the performance of each algorithm is evaluated under different K values. The results show that with the increase of K value, the overall performance of the algorithm presents a certain fluctuation trend. When K = 1, GA-CatBoost algorithm performs best, with an accuracy of 0.927, and its precision, recall rate, F1 value and Area Under Curve (AUC) are higher than other algorithms. With the increase of K value, the performance of GA-CatBoost is still excellent, and its accuracy gradually increases to 0.961 when K = 5. Although it drops slightly when K = 6, its accuracy is 0.929, but it is still at a high level compared with other algorithms. Among other algorithms, RF and SVM are the most stable when K = 5, and their precision and AUC value are always high, while KNN and C4.5 have great fluctuations, especially when K = 6, their precision and recall rate decrease. In addition, algorithms such as DeepPsy and XGBoost continue to improve when the K value increases, but they can never surpass the performance of GA-CatBoost, which proves the effectiveness and robustness of GA-CatBoost in predicting the mental health status of college students. This study improves the accuracy and efficiency of college students’ mental health prediction, and provides new technical means and theoretical support for colleges and universities to carry out accurate mental health education and intervention.