Enhanced Parkinson’s Disease Tremor Severity Classification by Combining Signal Processing with Resampling Techniques

Abstract Tremor is an indicative symptom of Parkinson’s disease (PD). Healthcare professionals have clinically evaluated the tremor as part Unified rating scale (UPDRS) which inaccurate, subjective and unreliable. In this study, a novel approach to enhance severity classification proposed. The proposed combination signal processing resampling techniques; over-sampling, under-sampling hybrid combination. Resampling techniques are integrated with well-known classifiers, such artificial neural network based on multi-layer perceptron (ANN-MLP) random forest (RF). Advanced metrics calculated evaluate approaches area under curve (AUC), geometric mean (Gmean) index balanced accuracy (IBA). results show that over-sampling performed better than other techniques, also techniques. improved significantly best classify ANN-MLP Borderline SMOTE has obtained 93.81% overall accuracy, 96% Gmean, 91% IBA 99% AUC. Besides, it found different differently classifiers.