Comparing Ensemble-Based Machine Learning Classifiers Developed for Distinguishing Hypokinetic Dysarthria from Presbyphonia

It is essential to understand the voice characteristics in normal aging process accurately distinguish presbyphonia from neurological disorders. This study developed best ensemble-based machine learning classifier that could hypokinetic dysarthria using classification and regression tree (CART), random forest, gradient boosting algorithm (GBM), XGBoost compared prediction performance of models. The subjects this were 76 elderly patients diagnosed with 174 presbyopia. models for distinguishing by CART, GBM, XGBoost, forest accuracy, sensitivity, specificity development identify them. results showed had when it was tested test dataset (accuracy = 0.83, sensitivity 0.90, 0.80, area under curve (AUC) 0.85). main predictors detecting Cepstral peak prominence (CPP), jitter, shimmer, L/H ratio, ratio_SD, CPP max (dB), min CPPF0 order magnitude. Among them, most important predictor identifying dysarthria.