Use of machine learning for simplification of University Personality Inventory (UPI)

Abstract

Background: Rapid diagnosis of mental health problems is crucial for college students. The University Personality Inventory (UPI) is a commonly used tools for assessing mental health in college students; however, it has certain limitations. This study aimed to develop a machine learning model for predicting the simplified UPI items that can rapidly and effectively screen for mental health issues. Methods: To construct the dataset, we administered the UPI to 5155 college students. We compared three machine learning models: Support Vector Machine, Random Forest, and K-Nearest Neighbors. Additionally, we analyzed individual features and feature combinations. Results: Among the three models, the Random Forest model performed the best, with an accuracy of 89.4 %. The UPI number of items in the scale was reduced by 90 % from 60 to 6. And the four items of Sensitive emotions, Physical exhausted, Feel self-abased and Uneasy without reason have high stability, the frequency of occurrence is >90 %. Conclusion: The results indicate that machine learning can successfully be applied to simplify the UPI and may be applicable for simplifying other lengthy measurement tools. This simplified measurement tool is expected to help college students assess their own mental health status more quickly and conveniently, enabling them to take timely measures to maintain their mental health.

Keywords

• Machine learning
• University personality inventory
• Mental health
• Simplification