Artificial Neural Network to Prescient the Severity of Parkinson’s Disease

Parkinson’s disease is a central nervous system disorder in which dopamine generating cells present in the substantia nigra (a part of our brain) gets damaged or perished. Due to a deficiency of dopamine, which acts as a neurotransmitter in our body, symptoms of Parkinson’s disease are caused. The change in vocal speech is one of the major symptoms shown and can be used to detect this disease. In this paper, there is an implementation and proprietary comparison of various machine learning models. The proposed model comparison shows that ANN is the comparatively better ML model to detect and predict Parkinson’s disease which has accuracy for total UPDRS is 83.56 % and for motor, UPDRS is 85.135 % which are the scales for measuring the severity elaborated further. We have utilized "Sklearn", "TensorFlow" and "Keras" python AI libraries to actualize all the ML models. It sums up that the accuracy generated by the proposed model is 84.45 % which is substantially higher than the research work done so far.