An early assessment of Persistent Depression Disorder using machine learning algorithm

Although various algorithms and strategies have been proposed for predicting depression and anxiety, none of the work is still suggested for an automated system for an early assessment of Dysthymia. This study aimed to enhance the accuracy of early diagnosis for Persistent Depression Disorder (PDD) through an improved machine learning technique utilizing the stacking SVM ensemble approach. To expedite the initial screening of dysthymia in students, a quantitative analysis of behavioral data based on machine learning was employed. The research collected behavioral data from 137 college students, and the gathered data was used for model development and experimentation. The findings revealed that PDD was predominantly prevalent among middle-class undergraduates majoring in non-technical fields. Notably, PDD rates were higher among rural undergraduates from both high- and low-income backgrounds. The proposed stacked SVM model demonstrated superior performance, achieving an accuracy of 89.4%. Detecting PDD early among undergraduates is crucial for mental health professionals, and the stacked SVM method proved effective in this aspect.