Table 6:

Performance of the best performing model (i.e., RoBERTa-MIMIC-Trial) by entity types on FRD.

Model	Strict Criterion			Relaxed Criterion
	Precision	Recall	F1	Precision	Recall	F1
Overall	0.7680	0.8027	0.7849	0.9028	0.9290	0.9157
age	0.9696	0.9733	0.9714	0.9696	0.9733	0.9714
allergy_name	0.8553	0.7010	0.7705	0.9434	0.7680	0.8467
bmi	0.9130	0.8750	0.8936	0.9565	0.9583	0.9574
cancer	0.7436	0.7835	0.7630	0.8983	0.9250	0.9114
chronic_disease	0.7297	0.7789	0.7535	0.8892	0.9304	0.9093
clinical_variable	0.7888	0.7633	0.7759	0.9151	0.8729	0.8935
contraception_consent	0.6473	0.7929	0.7128	0.8164	0.9349	0.8717
ethnicity	1.0000	0.7500	0.8571	1.0000	0.7500	0.8571
gender	0.9333	0.9385	0.9359	0.9861	0.9944	0.9902
language_fluency	0.8222	0.7872	0.8043	0.9778	0.9574	0.9675
lower_bound	0.8230	0.8623	0.8422	0.9163	0.9565	0.9360
pregnancy	0.8228	0.8698	0.8457	0.9489	0.9968	0.9723
technology_access	0.3846	0.4545	0.4167	1.0000	1.0000	1.0000
treatment	0.7178	0.7747	0.7451	0.8745	0.9223	0.8978
upper_bound	0.8208	0.8417	0.8311	0.9353	0.9555	0.9453