Table 3.

Model performance validation for PD classification.

Validation dataset		LR		NB		RF
		AUC	ACC	AUC	ACC	AUC	ACC
Finger tapping
Single hand	Amplitude_mean	0.472	0.333	0.583	0.500	0.458	0.500
(unilateral)l	Amplitude_std	0.583	0.500	0.472	0.500	0.458	0.417
	Speed	1.000	1.000	0.833	0.667	0.750	0.667
	Fatigue	0.722	0.583	0.611	0.583	0.417	0.500
Two-hands	Amplitude_mean	0.777	0.750	0.750	0.500	0.556	0.417
(bilateral)	Amplitude_std	0.444	0.500	0.500	0.583	0.597	0.583
	Speed	0.208	0.500	0.125	0.250	0.458	0.417
	Fatigue	0.500	0.416	0.222	0.500	0.556	0.500
Hand movements
Single hand	Amplitude_mean	0.639	0.667	0.417	0.500	0.528	0.583
(unilateral)	Amplitude_std	0.278	0.500	0.222	0.333	0.486	0.417
	Speed	1.000	1.000	0.972	0.917	0.778	0.583
	Fatigue	0.472	0.417	0.750	0.583	0.556	0.583
Two-hands	Amplitude_mean	0.750	0.667	0.528	0.667	0.292	0.250
(bilateral)	Amplitude_std	0.472	0.500	0.528	0.417	0.708	0.667
	Speed	0.389	0.417	0.389	0.500	0.056	0.250
	Fatigue	0.333	0.417	0.500	0.417	0.569	0.583
Pronation-supination movement of the hands
Single hand	Amplitude_mean	0.917	0.667	0.778	0.750	0.667	0.417
(unilateral)	Amplitude_std	0.944	0.917	0.889	0.917	0.889	0.833
	Speed	0.750	0.750	0.417	0.250	0.611	0.667
	Fatigue	0.444	0.417	0.611	0.500	0.778	0.750
Two-hands	Amplitude_mean	0.833	0.750	0.861	0.833	0.542	0.500
(bilateral)	Amplitude_std	0.611	0.750	0.861	0.750	0.625	0.583
	Speed	0.611	0.583	0.750	0.750	0.750	0.750
	Fatigue	0.306	0.333	0.583	0.583	0.306	0.500

Validation results of the combined right and left motor features for PD classification. The validation was made using three classifiers: Logistic regression (LR), Gaussian Naïve-Bayes (NB), and Random Forest (RF). Features with cross-validation AUC > 0.6 are highlighted in bold. Units: normalized amplitude [0–1] for finger tapping and hand movements; amplitude (degrees) for pronation supination for amplitude features. Time (frames), for speed in all the tasks. AUC, cross-validation area under curve; ACC, accuracy.