Diagnosis of Parkinson disease using machine learning and data mining systems from voice dataset

Abstract

Parkinson’s disease is one of the most painful, dangerous and non curable diseases which occurs at older ages (mostly above 50 years) in humans. The data-set for the disease is retrieved from UCI repository. A relative study on feature relevance analysis and the accuracy using different classification methods was carried out on Parkinson data-set. Sieve multigram data and Survey graph provide the statistical analysis on the voice data so that the healthy and Parkinson patients would be correctly classified. KStar and NNge present good accuracy based classification methods. Sieve multigram shows the edges between the nodes such as Fhi, Flo, Jitter, JitterAb, RAP and PPQ. KStar and NNge have connections with Shimmer and ShimmerDB . ADTree shows 21 leaves with 31 leaves and SimpleCART shows 13 leaves and 7 leaves. Most of the clusters vary with DBScan and SimpleKMeans with 25% and 38% towards Parkinson disease.