Table 4.

Performance in AUC of Machine Learning and FRAX Score

	Total fracture	p a	Hip fracture	p a
Machine‐learning model (CatBoost)	0.688 (0.687–0.688)		0.656 (0.655–0.656)
FRAX (major, without BMD), %	0.638	<0.001	‐	‐
FRAX (major, with BMD), %	0.660	<0.001	‐	‐
FRAX (major, with TBS), %	0.663	<0.001	‐	‐
FRAX (hip, without BMD), %	‐	‐	0.528	<0.001
FRAX (hip, with BMD), %	‐	‐	0.545	<0.001
FRAX (hip, with TBS), %	‐	‐	0.549	<0.001

AUC = area under the curve; FRAX = fracture risk assessment tool; TBS = trabecular bone score.

^aFRAX scores compared with the machine‐learning model.