Adapting transformer-based language models for heart disease detection and risk factors extraction

Journal of Big Data

Table 12 XLNET Evaluation Metrics for each risk factor indicator based on time attribute Identification

Risk factor	Time attribute	Precision	Recall	F-Score	Support
CAD	Before_DCT	0.80	0.83	0.81	406
	During_DCT	0.78	0.61	0.69	222
	After_DCT	0.66	0.63	0.65	71
Diabetes	During_DCT	0.67	0.49	0.57	272
	Before_DCT	0.72	0.57	0.64	209
	After_DCT	0.49	0.39	0.44	164
Hypertension	Before_DCT	0.70	0.64	0.67	116
	During_DCT	0.89	0.95	0.92	553
	After_DCT	0.10	0.08	0.09	51
Obese	Before_DCT	0.46	0.32	0.38	41
	During_DCT	0.77	0.52	0.62	69
	After_DCT	0.00	0.00	0.00	11
Medication	Before_DCT	0.67	0.51	0.58	863
	During_DCT	0.54	0.36	0.43	695
	After_DCT	0.67	0.61	0.64	894
Hyperlipidemia	Before_DCT	1.00	0.02	0.05	82
	During_DCT	0.75	0.72	0.73	65
	After_DCT	0.88	0.81	0.84	109
Weighted average		0.9361	0.9397	0.9371	42946
Macro average		0.3988	0.3848	0.3772	42946
Micro average				0.9397	42946
Accuracy				0.9397	42946