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

Journal of Big Data

Table 10 RoBERTa Evaluation Metrics for each risk factor indicator based on time attribute Identification

Risk factor	Time attribute	Precision	Recall	F-Score	Support
CAD	Before_DCT	0.78	0.89	0.83	436
	During_DCT	0.67	0.44	0.53	201
	After_DCT	0.00	0.00	0.00	34
Diabetes	During_DCT	0.00	0.00	0.00	107
	Before_DCT	0.67	0.58	0.62	288
	After_DCT	0.65	0.64	0.65	265
Hypertension	Before_DCT	0.76	0.82	0.79	159
	During_DCT	0.91	0.95	0.93	521
	After_DCT	0.00	0.00	0.00	55
Obese	Before_DCT	0.77	0.41	0.54	41
	During_DCT	0.80	0.11	0.20	36
	After_DCT	0.00	0.00	0.00	44
Medication	Before_DCT	0.68	0.51	0.58	995
	During_DCT	0.58	0.43	0.50	722
	After_DCT	0.62	0.45	0.52	727
Hyperlipidemia	Before_DCT	1.00	0.11	0.20	88
	During_DCT	0.81	0.57	0.67	77
	After_DCT	0.82	0.73	0.77	104
Weighted average		0.9390	0.9427	0.9394	42946
Macro average		0.4195	0.4037	0.3915	42946
Micro average				0.9427	42946
Accuracy				0.9427	42946