“…The first approach involves hand-crafting acoustic features, including certain variants of the jitter, shimmer, and harmonic-to-noise ratio that are indicative of PD speech impairments [ 17 , 18 , 19 , 20 , 21 , 22 ] and using traditional machine learning (ML) methods, such as support vector machines (SVM), random forests (RF), k-nearest neighbors (KNN), and regression trees (RT) [ 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 ].…”