Depressionmap: um sistema para indicadores de depressão e análise de dados sociodemográficos

Abstract

Context: According to the World Health Organization, after the COVID-19 pandemic, there was an increase of more than 25% in depression, making it the most disabling mental condition in the world. A system based on depression indicators can assist mental health professionals in screening for the disease. This approach facilitates the identification of vulnerable groups, the mapping of the most affected areas and the possibility of more efficient treatment. Objective: To develop DepressionMap, a system focused on generating indicators related to depression, contributing to intelligent solutions related to the disease. Method: This research began through a Systematic Literature Mapping (SLM), by reading related works, research gaps related to the topic were identified. In this context, DepressionMap was developed, a system that generates depression indicators based on the DASS-42 scale and sociodemographic variables, in addition to providing information on the performance of Machine Learning methods for classifying disease severity levels. Furthermore, an improved structure was proposed with the participation of mental health professionals. Results: Among the algorithms tested, the Support Vector Machine and Logistic Regression stood out with the best results in classifying depression levels, reaching accuracies of 99.66% and 99.69%, respectively. Conclusion: DepressionMap is a system capable of generating depression indicators; however, limitations related to the database used and the lack of professionals specialized in depression were identified. Due to these factors, an improved structure was proposed, with greater control over data collection and the participation of a health team. Furthermore, this study presented advances in the classification of depression levels compared to related research.