Machine Learning based approaches for Identification and Prediction of diverse Mental Health Conditions

Today, looking after our mental health has become as essential as maintaining our physical health and well-being. It is found that a lot of people belonging to different age groups, gender and a wide range of professional backgrounds suffer from different mental disorders globally that lead to sizable losses in health and functioning of an individual. Mental health issues such as depression, anxiety and stress can originate due to various biological, psychological, and environmental factors which give rise to suicidal thoughts. These challenges have encouraged the advancement of numerous machine learning based approaches for providing an impactful solution to predict and identify the severity of various mental health issues on time, and therefore suggest potential treatment outcomes. It has been observed that Machine learning is one of the most efficient approaches that can be used to analyze enormous amounts of data for making accurate predictions in the healthcare system. This paper explores various sources for data collection (Clinical questionnaires, interviews, social media, medical history and speech alterations), automated methods and ML-based depression detection algorithms of different classes including classification, regression algorithms, deep learning, and ensemble method used by researchers for analyzing varied mental health conditions. Few studies made use of Synthetic Minority Oversampling Technique (SMOTE) to reduce the class imbalance of the training data in order to achieve better accuracy. A comparison of different parameters such as objectives, results and limitations between the referred research papers presented in the domain of depression detection has also been included in this paper.