Parkinson's Disease Feature Subset Selection Based on Voice Samples

In this study, semi automation prediction of PD is investigated based on twenty two features of voice samples extracted from 147 subjects. Firstly, the original features of voice are used for recognition of PD or otherwise with MLP as classifier and Levenberg Marquardt and Scaled Conjugate Gradient as training algorithm. Next, to identify the number of significant features amongst the original attributes, Principal Component Analysis is implemented to perform this task. Upon implementation of PCA, the first four eigenvalues are identified as the significant principal components and further validated by the rule of thumb of PCA namely the Scree Test as well as Cumulative Variance rule. Based on initial findings attained, it was found that SCG as training algorithm contributed as the most suitable algorithm to be used by the classifier based on 92.9% accuracy rate with original features as inputs to classifier and 94.2% upon completion of PCA as feature subset selection.