Evolutionary Jaya Algorithm for Parkinson’s Disease Diagnosis using Multi-objective Feature Selection in Classification

Parkinson's disease is world's leading disorder in people aged above 60. It gradually affects the entire body and the brain losses control over the organs. Early detection of Parkinson's disease can reduce the severity of the disease. Medical diagnosis decision support system (MDDSS) assists medical experts in accurate diagnosis of Parkinson's disease and can also be useful in telemedicine. Data mining classification algorithms can diagnose Parkinson's disease with the best possible accuracy when optimal numbers of features are selected. This work proposes a novel algorithm BEJA-V: an evolutionary version of Jaya Algorithm for feature selection in classification using scalarization method of solving multi-objective optimization problems. The continuous Jaya Algorithm is converted into binary encoded Jaya algorithm using V-shaped transformation function. The proposed BEJA-V algorithm improves the classification accuracy and ensures optimal size of features subset.