Depression Prediction: A Clinical Questionnaire and Machine Learning Based Approach

Abstract

Depression is specified as a mental health disease that leads to continual gloomy mood and detached interest in daily tasks, which a person was able to do easily and showed much interest normally, thus crippling a person’s daily life. It’s important to identify the symptoms of mental health problems early in a person as there are many conditions for which effective treatments work best during the early stages thus preventing the aggravation of the disease. Over the past decade, Machine learning techniques have become immensely popular for examining the medically available data and identifying the root issue. In this work, we also use a machine learning based approach to propose a response-based classifier that uses just a quick questionnaire to diagnose and predict the early onset of depression in a person being screened. The questionnaire used in our approach has been designed strictly based on the Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM- 5). In particular, our approach uses the responses made by the person being screened to predict depression, using a non-linear support vector machine (SVM) classifier. In general, the classifier is straightforward, computationally lightweight, and can predict depression with 83.87% accuracy. We also test our proposed method on a Kashmir-based dataset.

Topics & Concepts