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

Table 2

Results of the four constituting components on the MDVR-KCL dataset.

Components	Evaluation
	Precision (%)	Recall (%)	F1 score (%)	Acc (%)
TmmNet (NoFT)	86.39	56.87	68.59	80.46
TmmNet (NoConv)	97.16	69.02	80.71	86.82
TmmNet (NoTN)	71.73	61.92	66.46	75.44
TmmNet (NoAttn)	96.32	66.74	78.85	85.22
TmmNet	100.00	75.26	85.88	90.23

The bold values in Table 2 represents the highest results among the split models.