Preventing Dropouts and Promoting Student Success: The Role of Predictive Analytics

In today's society, people face increasing stress, leading to an elevated risk of mental disorders, particularly among college-level students. Detecting these disorders is crucial for their academic success, and we have conducted a study using a set of online PHQ-9 questionnaire data from a Mexican medical school. By utilizing machine learning techniques, we have demonstrated the potential of this approach to detect these hardly visible mental disorders, achieving a strong predictivity with AUC = 0.69. Furthermore, we utilized regression analysis to determine that item 6 of the PHQ-9 questionnaire is the most significant predictor for students' ideation of dropping out. This study provides a promising approach to detect mental disorders in college-level students, which could help educators and psychiatry specialists target treatment to ensure students' academic success.