Evaluating Drug Risk Using GAN and SMOTE Based on CFDA’s Spontaneous Reporting Data

Table 5

Evaluation metrics based on the actual ADR data.

Classifier	Label	Precision	Recall	F1	Accuracy (%)
Model 1 (RF)	0	0.89	0.99	0.94
	1	0.80	0.36	0.50	88.63
	2	0.72	0.28	0.40
	Macro-avg	0.81	0.54	0.61
	Weighted-avg	0.88	0.89	0.87
Model 2 (SMOTE + RF)	0	0.98	0.95	0.97
	1	0.80	0.90	0.85	94.65
	2	0.72	0.94	0.81
	Macro-avg	0.83	0.93	0.88
	Weighted-avg	0.95	0.95	0.95
Model 3 (FS_GAN + SMOTE + RF)	0	0.99	0.99	0.99
	1	0.92	0.93	0.93	97.90
	2	0.96	0.94	0.95
	Macro-avg	0.96	0.95	0.95
	Weighted-avg	0.98	0.98	0.98