Ensemble Based Precautionary Detection of Parkinson's Disease from Voice Signal

Parkinson's Disease is a neurological progressive disorder that affects the nervous system of a human being, which effects the parts of the body. Tremors, imbalance, slowed movement, stiffness, difficulty in walking and talking etc. are observed. The reasons being, brain disorder, injuries, toxins certain drugs etc. The disease has no cure however treatment can be offered for relief from symptoms. However early detection can cure the disease in the first stages and extend the life span of the patient. This paper proposes a Ensembled learning model for early detection of Parkinsons disease through voice features. Random forest ensemble models have outperformed the Machine learning and Neural Network models in detecting disease from voice. An accuracy of 94.8% is achieved with this model with proper preprocessing techniques and training. As a future work the proposed model's software prototype can be released for mobile application for cost effective early detection of Parkinsons Disease.