Paper Title
EARLY DETECTION OF PARKINSON’S DISEASE USING EXTRA-TREES ENSEMBLE LEARNING

Abstract
Abstract - Early detection of Parkinson’s Disease (PD) is very important as this disease has no cure using medical procedure and can only be controlled using medical Means in early stages. But detection of PD is very difficult in early stages as there is no specific test to diagnose PD. An expert in neurology will diagnose PD based on the patient’s medical history and symptoms. Since this procedure is not systematic and is solely based on human experience so this procedure is prone to human errors as there are many other neurodegenerative diseases that have similar symptoms as that of PD but actually is not PD and hence it is very much difficult to distinguish among these diseases. To overcome all these problems this paper proposes an expert system named, Early Detection of Parkinson’s Disease using Extra-Trees (EDPDET) for early detection of PD. The proposed expert system uses K-Means SMOTE oversampling technique to deal with class imbalance problem, Boruta feature selection to perform feature selection and the classification is done using bagging ensemble Extra-Trees (ET) to overcome the variance and overfitting problem of the single-classifier based model. The accuracy and F1-score achieved by the proposed model EDPDET is compared with different single-classifier based model, ensemble models and various models present in the literature and it is observed that the proposed model outperforms these models greatly. Keywords - Machine Learning; Parkinson’s Disease; Boruta; K-Means SMOTE; Extra-Trees.