A Triplet Multimodel Transfer Learning Network for Speech Disorder Screening of Parkinson’s Disease

Table 4

Performance of machine learning methods on the IPVS dataset.

Vowel					Vowel					Vowel
Methods	Precision (%)	Recall (%)	F1 score (%)	Acc (%)	Methods	Precision (%)	Recall (%)	F1 score (%)	Acc (%)	Methods	Precision (%)	Recall (%)	F1 score (%)	Acc (%)
DT	98.46	98.06	98.26	98.33	DT	98.17	98.81	98.49	97.95	DT	97.76	97.18	97.47	97.23
GBDT	97.92	98.47	98.19	98.26	GBDT	97.47	99.49	98.47	97.90	GBDT	97.68	98.08	97.88	97.67
KNN	97.07	97.07	97.07	97.21	KNN	97.50	98.48	97.99	97.23	KNN	95.04	97.87	96.43	95.97
LR	97.48	98.57	98.02	98.09	LR	97.92	99.11	98.51	97.97	LR	97.15	97.32	97.24	96.96
RF	99.94	99.69	99.82	99.82	RF	99.42	99.96	99.69	99.58	RF	99.75	99.40	99.58	99.54
LDA	97.07	97.07	97.07	97.21	LDA	97.50	98.48	97.99	97.23	LDA	95.04	97.87	96.43	95.97
LightGBM	99.69	100.00	99.84	99.85	LightGBM	99.89	99.73	99.81	99.74	LightGBM	99.93	99.52	99.72	99.69
Vowel					Vowel					TmmNet
Methods	Precision (%)	Recall (%)	F1 score (%)	Acc (%)	Methods	Precision (%)	Recall (%)	F1 score (%)	Acc (%)	Methods	Precision (%)	Recall (%)	F1 score (%)	Acc (%)
DT	98.71	98.46	98.58	98.34	DT	98.40	98.14	98.27	98.05	Vowel	99.89	99.89	99.89	99.90
GBDT	97.58	99.28	98.42	98.13	GBDT	98.11	99.57	98.84	98.68	Vowel	99.86	99.83	99.84	99.79
KNN	99.74	99.79	99.76	99.72	KNN	97.62	97.26	97.44	97.09	Vowel	99.72	99.61	99.66	99.63
LR	96.41	97.79	97.09	96.57	LR	97.24	99.15	98.18	97.93	Vowel	99.95	99.85	99.90	99.87
RF	99.79	99.79	99.79	99.75	RF	99.89	99.89	99.89	99.88	Vowel	99.89	99.94	99.92	99.91
LDA	96.65	97.14	96.90	96.33	LDA	97.62	97.26	97.44	97.09
LightGBM	99.89	99.59	99.74	99.69	LightGBM	99.89	99.68	99.78	99.76

The bold values in Table 4 represents the highest results among all the compared methods.