Optimization-Based Ensemble Feature Selection Algorithm and Deep Learning Classifier for Parkinson’s Disease

Table 2

Results of classifiers with triple feature (KPCA + OBEFS).

Feature combination	F-measure	Accuracy	Error	MCC
SVM classifier (%)
TQWT + MFCC + Wavelet	82.9150	85.1035	14.8965	56.2000
TQWT + MFCC + Concat	80.7960	83.6640	16.3360	54.6000
TQWT + Wavelet + Concat	86.3590	87.4662	12.5338	58.8000
MFCC + Wavelet + Concat	87.4510	88.1294	11.8706	59.7000
CNN classifier (%)
TQWT + MFCC + Wavelet	85.8697	87.5696	12.4304	57.2007
TQWT + MFCC + Concat	90.2315	91.9315	8.0684	61.4600
TQWT + Wavelet + Concat	86.3695	88.0694	11.9306	63.3994
MFCC + Wavelet + Concat	92.4752	94.1752	5.8248	64.5384
FCLSTM-CNN classifier (%)
TQWT + MFCC + Wavelet	94.2258	93.0470	6.9530	67.6669
TQWT + MFCC + Concat	91.5250	93.0854	6.9146	67.7060
TQWT + Wavelet + Concat	93.4200	93.1261	6.8739	65.4457
MFCC + Wavelet + Concat	91.6921	95.1557	4.8443	67.2960
FCBi-LSTM classifier (%)
TQWT + MFCC + Wavelet	98.3100	96.6381	3.3619	74.300
TQWT + MFCC + Concat	96.5900	98.0244	1.9756	72.300
TQWT + Wavelet + Concat	97.5200	97.3457	2.6543	70.300
MFCC + Wavelet + Concat	98.5010	98.7720	1.2280	71.400