Deep transfer learning architecture for depression detection using PHQ-9 analysis in Bengaluru urban population

Abstract

In the metropolitan cities in India, there is a scarcity of population-based statistics on depression prevalence. This study utilized the patient health questionnaire-9 (PHQ-9) instrument to look at the prevalence and risk factors for depression in major metropolitan cities. This study's population-based analysis identifies major demographic and socioeconomic characteristics associated with depression in an Indian urban context. This quantitative inquiry conducted a population-centric analysis of 987 people from a metropolitan city. Depression was measured and validated using the PHQ-9 questionnaire to calculate the prevalence rates and sociodemographic characteristics. A deep transfer learning framework integrating visual and textual features was proposed for automated classification of depression, to assess mental health patterns in urban populations. Depression symptoms were common, with many respondents reporting persistent symptoms like a persistent lack of interest, pervasive sorrow, sleep difficulties, and suicidal ideation. Bivariate studies found relationships between these symptoms and other demographic and socioeconomic characteristics, whereas multivariate analysis revealed significant predictors. In addition, the classification of people’s stress in metropolitan regions is analyzed through various pre-trained models to show better outcomes. The substantial link between sociodemographic characteristics and depression levels shows the importance of mental health interventions for this urban Indian population.