Enhancing mental well-being: An artificial intelligence model for predicting mental disorders

Abstract

Pervasive mental health conditions like depression, stress, and anxiety significantly affect individual well-being. Early identification and understanding are critical to minimize their negative effects. This study explores how AI models can be leveraged for improved mental health assessment. We have introduced some machine learning classifiers along with a deep learning model. Among the applied ML classifiers, the ”Ensemble” approach aims to outperform individual models by harnessing their collective strengths. While data acquisition for mental health research can be challenging, we utilized a publicly available dataset from the Kaggle site, acknowledging potential limitations like data imbalances and missing values. This imbalanced dataset is balanced by the Random Oversampling model. In our study, we introduced a state-of-the-art approach to predicting mental conditions such as depression, anxiety, and stress. For every condition, we utilized three machine learning models: the K-Nearest Neighbors (KNN), Random Forest (RF), and Support Vector Machine (SVM), combining these models into an ensemble model where a Voting classifier is used in the ensemble model. We also applied the Artificial Neural network (ANN) as a deep learning model for each disorder. Though the ensemble model performed an excellent outcome, the ANN model found outstanding outcomes from all the models. The ANN model achieved the highest accuracy of 99.73% for depression, 99.89% for anxiety, and 99.39% for stress.