Heterogeneous Bootstrapped Ensemble Model for an Early Assessment of MDD

Abstract

Background: Major depressive disorder (MDD) is a psychiatric illness that has far-reaching consequences for mental and physical health, eating and sleeping habits, professional employment activities, and lifestyle. Individualized diagnosis based on the DSM-5 has long been a strategic imperative and is entirely dependent on the clinical psychologist's and psychiatrist's subjective behavioural interpretation.

Method: A Machine Learning (ML)-based Heterogeneous bootstrapped ensemble method for an automatic detection approach for detecting MDD based on an individual's behavioural characteristics. In this study, the behavioural data of 503 undergraduates were collected.

Result: In this study, we found that the probability of occurrence of MDD in male participants pursuing technical education is higher than in females of the same group. This study found that the Heterogeneous bootstrapped ensemble method performed well with an accuracy of 87.38%, followed by SVM and Bayes Net with accuracy levels of 83.34 and 81.72, respectively.

Conclusion: This research revealed that students of technical streams are more likely to develop MDD. Another fact that came to light was that undergraduates from urban areas are more prone to MDD than rural areas. The early detection of MDD among undergraduates is essential for mental health practitioners, and the Heterogeneous bootstrapped ensemble method approach has demonstrated exemplary performance in this regard.