Analysis of Parkinson's disease diagnosis using a combination of Genetic Algorithm and Recursive Feature Elimination

Parkinson's disease is a neurological disorder which affects different body parts. Voice distortion is one of the early symptoms to appear. It is present in most of the Parkinson patients and has a lot of potential in early diagnosis of the disease. In this paper, potential of feature selection techniques is explored in diagnosis of Parkinson's disease using speech features. In this, two-stage feature selection is applied which combines the benefits of Genetic Algorithm and Support Vector Machine-Recursive Feature Elimination techniques into one. Classification is done using SVM and results are compared with SVM without feature selection and individual selection techniques. The two-stage feature selection technique not only reduced the dimensionality of the dataset but also improved the performance of the system.