A Survey on Bipolar Disorder Classification Methodologies using Machine Learning

This paper focuses on a scenario based Bipolar disorder (BD) classification methodology. BD is a psychological brain disorder that affects 45 million people worldwide. It is also called manic depression and causes serious changes in energy level, behavior and shift in moods. Many researchers reported volumetric changes in brain neural networks with BD compared to a healthy person (HP). The methodology currently used for the identification of disorder using functional magnetic resonance imaging (FMRI) data analysis requires specialists in the field and the process is generally time consuming. The present breakthroughs on the prediction using retinal vasculature, where the brain and retinal share the same biomarker structure, machine learning and deep learning, which remains as the most used approach for classification. This paper reviews feature selection and various methodologies used on BP classification and identifies the best methodology that can help the future research work.