Classification of Parkinson’s Disease using Speech Signal with Machine Learning and Deep Learning Approaches

Abstract

Parkinson's disease (PD) is a chronic neurological condition that is growing in prevalence and manifests both motor and non-motor symptoms. Most PD patients have trouble speaking, writing, and walking during the early stages of the disease. Analysis of speech problems has been effective in identifying Parkinson's disease. According to studies, 90% of Parkinson's disease patients experience speech problems. Even though there is no known cure for Parkinson's disease, using the right medication at an early stage can greatly reduce the symptoms. One of the key categorization issues for the diagnosis of Parkinson's disease is the correct interpretation of speech signals. The major goal of this project is to use deep learning and machine learning approaches to predict and categorize PD patients at an early stage. A trustworthy dataset from the UCI repository for Parkinson disease has been used to evaluate the method's performance. Several classification models are successfully used in this study for classification tasks, including Extreme Gradient Boosting (XGBoost), Ada Boost, Light Gradient Boosting Machine, CatBoost, Gradient Boosting, Random Forest, Ridge, Decision Tree, Logistic Regression, K Neighbors, SVM - Linear Kernel, Naive Bayes, and deep neural networks (DNN1, DNN2, DNN3). The Extreme Gradient Boosting classifier achieved the greatest classification accuracy of 92.18% (among the machine learning classifiers). By using the chosen features as input, the three layer deep neural network (DNN2) has the best accuracy of 95.41% amongst deep learning techniques. The collected results indicate that deep neural networks performed better than machine learning methods.