Predicting depression level based on human activities and feelings: A fuzzy logic-based analysis

Millions of individuals die by suicide each year, and many more suffer from severe depression. Furthermore, these deaths harm education, the economy, and healthcare worldwide. An individual’s persistent feelings and activities, which include sleep disorders such as insomnia, sleeping overtime, or spending the majority of time lying down; pessimistic thinking about the future; and thoughts of committing suicide, help uncover the cause of depression. Several attempts have been made to prevent these losses and deaths. Our study proposes a simple fuzzy inference model that accurately predicts depression levels based on emotions and activities—even with incomplete data and uncertainties—and enhances mental health prediction, bridging theory and practice effectively. Psychologists and professors collaborated to create a survey to collect data for this study. The experiment was conducted using a Google survey form. This method effectively captures the ambiguity and imprecision in depression evaluation by combining linguistic elements of psychological traits. Using the Pearson correlation and R-squared methods, 15 features were chosen from 30 features, followed by five membership functions (poor, mediocre, average, decent, and good) and fuzzy rules to evaluate and create accurate forecasts of depression severity. Our proposed architecture can correctly classify depression levels based on human activities and feelings with 94% accuracy using a less sophisticated rules dictionary than previous pre-trained or hybrid models. Fuzzy logic performs better here by accurately categorizing ambiguous human emotional inputs into distinct degrees.