Enhancing Mental Illness Prediction using Tree based Machine Learning Approach

Abstract

A major challenge faced by the world today is to identify and rehabilitate various types of mental disorders. World Health Organization (WHO) has identified in its survey, around twenty percent of the world’s teenagers and children are suffering from mental health issues. It is also analyzed that there is a reduction of 10 to 25 years of life expectancy who suffer from a serious mental disorder. /p)(p)Machine learning approach is used in our study to predict mental illness on the data collected using DASS42 questionnaire. Severities between normal to extremely severe for stress, anxiety, and depression are classified using a Tree- based machine learning algorithm i.e. Decision Tree and its ensemble XGBoost. After choosing the right configuration for the algorithm for DASS42 dataset by tuning the hyperparameters, it is observed that the tree based machine learning algorithm gives better accuracy of 98.46% for Anxiety, 98.55% for Depression, and 98.44% for Stress, compared to other ML models.