Ensemble Classifier to Enhance Computer Aided Diagnosis of Parkinson Disease

Abstract

The aim of ensembled model is to calculate the Unified Parkinsons disease rating score(UPDRS) from various voice measures. Parkinsons disease is a neurodegenerative disorder of the central nerve system which affects movements. We collected data from 42 persons having early stage of Perkinsons disease. Total number of 5875 voice recordings are present in dataset. We use the different machine learning models which can predict the motor UPDRS score from the various voice measures. Then evaluation parameters (Correlation, R Square, RMSE, Accuracy) between the actual and the prediction are evaluated and results are compared. After comparing the results of the various models, we ensemble the top 3 models and results are evaluated which gives stronger overall prediction. K-Fold validation approach is used to measure the robustness of ensembled model.