Parkinson’s Disease Classification Using Deep Convolutional Neural Network Based On Acoustic Signals

Abstract

 Parkinson's disease is one of the neurodegenerative diseases that is actually seen in the elderly compared to the younger ones. Since a patient may have the same behavior compared to a healthy person at the very early stage of the disease, the automatic identification of Parkinson's disease is one of the main medicine-related challenges. Comparison of classifier efficiency in both machine learning and deep learning for the diagnosis of Parkinson's disease through Support Vector Machine ( SVM), Random Forest classifiers, and Deep Convolutionary Neural Network (DCNN) is made in stages in this work. The 196 voice samples dataset in.csv format was obtained from the Kaggle database, and the Google Colaboratory framework was used to implement it, and Python is the language we used to implement it. The results reveal the potential of using Random Forest (RF) or Support Vector Machine ( SVM) techniques and Deep CNN. Once fit, these algorithms provide a reliable computational method to estimate the presence of PD with a really high accuracy. Through this we measure the classifiers performance.