Multitask Interactive Attention Learning Model Based on Hand Images for Assisting Chinese Medicine in Predicting Myocardial Infarction

Table 7

Evaluation results of two different weight combinations in the IIM module.

	Task	Result (%)
		Accuracy	Sensitivity	Specificity	AUC	-score
(0, 0.2)	M	81.02	79.23	83.21	82.25	79.39
	P	80.37	78.16	81.91	82.01	79.31
(0, 0.4)	M	82.71	80.85	84.79	83.03	81.45
	P	82.26	81.11	84.63	82.98	81.27
(0, 0.6)	M	82.39	80.16	83.33	83.92	81.02
	P	81.72	81.21	82.70	83.54	81.25
(0, 0.8)	M	80.35	78.24	81.26	82.22	78.14
	P	80.54	78.93	82.38	82.83	80.03
(0, 1)	M	79.29	78.45	80.35	81.44	77.76
	P	80.06	78.82	82.96	82.07	78.31
(0.2, 0)	M	82.84	80.02	83.29	83.03	80.15
	P	83.45	80.89	84.67	84.41	81.73
(0.4, 0)	M	81.10	79.24	83.25	82.74	79.31
	P	82.26	80.23	83.33	83.67	81.20
(0.6, 0)	M	80.43	78.85	81.01	82.03	78.26
	P	81.93	79.64	80.40	82.39	79.33
(0.8, 0)	M	80.22	78.65	81.23	81.42	78.35
	P	80.47	78.38	82.17	81.89	78.94
(1, 0)	M	78.91	75.42	80.52	80.30	77.43
	P	79.09	78.51	81.33	81.35	78.41