Diagnosis of Parkinson’sDisease (PD) in the early stages is an arduous assignment to the physicians and researchers, because the disease symptoms are subtle and poorly characterized. In the present decade, diagnosis of PD through machine learning approach may provide a better understanding of the disease. PD is characterized by a multitude motor and nonmotor symptoms. Acquisition of speech articulation difficulty symptoms of PD patients is noninvasive and economic. So, in this study speech signal is used as a biomarker for diagnosis of PD. The performance of seven classifiers:Logistic Regression, KNN, SVN, Decision tree,Random Forest, Gradient Boosting and XGBoost in identifying PD subjects is investigated with various performance matrices i.e. accuracy, sensitivity, specificity and area under the curve. The simulation results suggest that the classifiers can accurately predict PD patients and healthy subjects. Among all the classifiers, XGBoost classifier has achieved the highest performance.
Volume 12 | Issue 6
Pages: 310-315
DOI: 10.5373/JARDCS/V12I6/S20201034