Artificial intelligence in the evaluation and treatment of osteoporotic and fragility fractures: A systematic review.
Xiaoli Liu, Elvis Chun Sing Chui, Xinbao Wu et al.
✦ AI-curated · Sources linked
In 30 seconds
This PRISMA-compliant systematic review examined original studies of artificial intelligence, machine learning, and deep learning for evaluating and managing osteoporotic and fragility fractures in adults. Searches identified 21 studies from 1286 records, covering fracture-risk prediction (10 studies), bone mineral density estimation and osteoporosis screening (6), and prognosis or postoperative outcomes (5). The review reports that AI often outperformed traditional methods such as FRAX and showed strong performance in bone-density estimation and fracture detection. However, methodological heterogeneity, risk of bias, and limited prospective or multicentre validation constrain clinical translation.
Key findings
- Of 1286 records, 21 studies were included: 10 on fracture-risk prediction, 6 on bone-density estimation or screening, and 5 on prognosis or postoperative outcomes.
- AI-based fracture-risk prediction often outperformed traditional tools such as FRAX.
- The review found robust AI performance in bone mineral density estimation and fracture detection.
Why it matters
AI methods may support fracture-risk assessment, imaging-based detection, osteoporosis screening, and postoperative monitoring. The review suggests potential advantages over traditional approaches, but the evidence base remains heterogeneous and incompletely validated. Clinicians and health systems should therefore distinguish promising model performance from demonstrated benefit in routine care.
What to keep in mind
The included evidence was methodologically heterogeneous, carried risks of bias, and had limited prospective and multicentre validation.