A Comparative Study of Early Detection of Parkinson's Disease using Machine Learning Techniques

Parkinson's Disease (PD) is considered a malison for mankind for several decades. Its detection with the help of an automated system is a subject undergoing intense study. This entails a need for incorporating a machine learning model for the early detection of PD. For discovering a full proof model, the cardinal prerequisite is to study the existing computational intelligent techniques in the field of research used for PD detection. Many existing models focus on singular modality or have a cursory analysis of multiple modalities. This encouraged us to provide a comparative literature study of four main modalities signifying major symptoms used for early detection of PD, namely, tremor at rest, bradykinesia, rigidity, and, voice impairment. State- of-the-art machine learning implementations namely Logistic Regression (LR), Support Vector Machine (SVM), Decision Tree (DT), K-nearest neighbors (KNN), Stochastic Gradient Descent (SGD) and Gaussian Naive Bayes (GNB) are executed in these modalities with their respective datasets. Furthermore, ensemble approaches such as Random Forest Classifier (RF), Adaptive Boosting (AB) and Hard Voting (HV) are implemented. Our results are compared with those obtained with their respective researches. Among all the tests, applying Random Forest (RF) on Static Spiral Test (for detecting tremor) gave us the most significant result, i.e. the highest accuracy of 99.79%. This leads to the conclusion that the multimodal approach with the help of the ensemble method should be used to get better and accurate results.