Table 2.

AI Model Performance Metrics

	A	B	C	D	E	F
BPE Classification Method	AI Model (MIPs)	AI Model (Slabs)	AI Model (Slabs)	AI Model (Slabs)	AI Model (Slabs)	Radiology Report
Reference Standard	Reference Standard #1 Radiology Report	Reference Standard #1 Radiology Report	Reference Standard #1 Radiology Report	Reference Standard #1 Radiology Report	Reference Standard #2 Consensus Reading	Reference Standard #2 Consensus Reading
Test Set	Full Testing Set	Full Testing Set	BI-RADS 4/BI-RADS 5 Subgroup	Reader Study Subgroup	Reader Study Subgroup	Reader Study Subgroup
Accuracy	76% (74%, 79%)	79% (77%, 81%)	77% (75%, 80%)	89% (84%, 93%)	94% (91%, 98%)	88% (84%, 93%)
Sensitivity	57% (50%, 63%)	75% (71%, 79%)	76% (73%, 80%)	57% (42%, 73%)	77% (62%, 92%)	80% (65%, 94%)
Specificity	86% (83%, 89%)	81% (79%, 83%)	78% (76%, 81%)	97% (95%, 100%)	97% (95%, 100%)	90% (85%, 95%)
AUC	0.79 (0.76, 0.81)	0.84 (0.82, 0.86)	0.84 (0.81, 0.86)	0.83 (0.74, 0.91)	0.96 (0.94, 0.99)	------

Abbreviations: AI, artificial intelligence; AUC, area under the curve; BI-RADS, Breast Imaging Reporting and Data System; BPE, background parenchymal enhancement; MIP, maximum intensity projection