A Supervised Machine Learning Approach To Classify Parkinson Disease

Abstract

Parkinson's disease is a degenerative brain disease. In most cases, over the age of 50 people are affected by this disease, but nowadays, young people are also affected by this disease due to genetic reasons. Due to reduced brain function in patients suffering from this disease, the patient suffers from imbalance along with difficulty in walking and speaking. Our present research work was done to assess classification algorithms in machine learning language. Here we identify Parkinson's disease from the data set that this disease is Parkinson's disease YES or No. This research work also focused on the medical field for the confirmation of clinical trials and identifying Parkinson's disease. In our study, we used six types of machine learning classification analysis algorithms (LR, KNM, DT, RF, XGB, SVM) to identify PD datasets. From the Logistic Regression classifiers algorithm, we gain 76% accuracy from the accuracy calculation. We archive 77% accuracy from KNM, from SVM we achieve 77% accuracy, from DT we get 73% accuracy, from XGB we found 78 % accuracy. But we archive the highest accuracy 81% from Random Forest (RF) in our research study where the best accuracy was found to identify Parkinson's disease. In the future, using this model in the medical field will help to classify Parkinson's disease more easily. Keywords—Parkinson’s Disease, Machine Learning, Identify, Classifications, Analysis, Logistic Regression, K Neighbors, Decision Tree, Random Forest, XGB, SVM, Impact.