Table 20.

Multi-class (CO (Class 0) vs. PD Stage 2 (Class 2) vs. PD Stage 2.5 (Class 2.5), and PD Stage 3 (Class 3)) and two-class classification (CO vs. PD (Class 2, Class 2.5, and Class 3)) summary using several CNN classifiers (AlexNet, ResNet-50, ResNet-101, and GoogLeNet) with 10-fold cross-validation for Ga dataset.

Dataset	vGRF Signal	CNN Classifier	Time Window	Frequency Range	Classification Task		Evaluation Parameters
							Sen (%)	Spec (%)	Acc (%)	AUC
Ga	LF	GoogLeNet	15 s	0.83–1.95 Hz	Multi-Class	Class 0	97.98	99.17	98.77	0.9857
		AlexNet	10 s	1.95–50 Hz		Class 2	95.93	98.25	97.40	0.9709
		AlexNet	10 s	1.95–50 Hz		Class 2.5	93.75	98.83	97.92	0.9629
		AlexNet	30 s	1.95–50 Hz		Class 3	98.21	97.44	97.54	0.9783
		ResNet-50	15 s	0.83–1.95 Hz	Two-Class		99.01	99.17	99.11	0.9996
	RF	ResNet-101	10 s	0.83–1.95 Hz	Multi-Class	Class 0	98.43	99	98.81	0.9872
		AlexNet	10 s	1.95–50 Hz		Class 2	97.97 *	97.78 *	97.85 *	0.9787 *
		AlexNet	15 s	0.83–1.95 Hz		Class 2.5	96.88 *	96.88 *	96.88 *	0.9688 *
		ResNet-50	10 s	1.95–50 Hz		Class 3	94.05	99.24	98.59	0.9664
		AlexNet	10 s	1.95–50 Hz	Two-Class		99.56	98.18	98.59	0.9941
	CF	AlexNet	15 s	1.95–50 Hz	Multi-Class	Class 0	98.65 *	99 *	98.89 *	0.9883 *
		AlexNet	15 s	1.95–50 Hz		Class 2	96.65	98.42	97.77	0.9753
		AlexNet	15 s	1.95–50 Hz		Class 2.5	91.88	99.19	97.88	0.9553
		AlexNet	15 s	1.95–50 Hz		Class 3	99.11 *	98.98 *	99 *	0.9904 *
		ResNet50	10 s	0.83–1.95 Hz	Two-Class		99.77 *	98.80 *	99.11 *	0.9995 *

Note: * denotes the best classification result and was selected using Youden’s index criteria.