Automatic Speech Analysis in Patients with Parkinson's Disease using Feature Dimension Reduction

Dysphonia is a common speech disorder in Parkinson's disease. Speech analyses have already been used in patients with Parkinson's disease and class prediction is an essential task in automatic speech treatment. Speech data contain large amounts of redundancies and ambiguities among the attributes which result in considerable noise. Modern data analysis often faces high-dimensional data using dimension reduction statistical techniques. In this work, the potential of Common Factor Analysis (CFA), Principal Component Analysis (PCA) based modeling in dimensionality reduction is taken into consideration as the data smoothening tool with multiclass target expression data. On the basis of suggested CFA and PCA-based modeling, the power class prediction of logistic regression (LR) and Decision Tree (C5.0) in numeric data to develop an advanced classification model is investigated in publicly available Parkinson's disease dataset Silverman voice treatment (LSVT). In addition, using only 9 dysphonia features, classification accuracy was (99,20%) and (99,11%) for CFA-LR and PCA-C5.0 respectively. In sum, our combined dimension reduction and data smoothening approaches have significant potential to minimize the number of features and increase the classification accuracy and then automatically classify subjects in Parkinson's disease patients from healthy speakers.