COIOS
Emergent technology Item confirms peer-reviewed

AI models predict death after hip fracture but are seldom externally validated

A systematic review of 40 studies published between 2010 and 2026 found that machine learning models for outcomes after hip fracture, most often death, achieved moderate to strong discrimination but were limited by inconsistent external validation and sparse calibration.

The review searched four databases for machine learning models predicting clinical outcomes after osteoporotic hip fracture, identifying 40 eligible studies published between 2010 and 2026. Death was the most frequently modelled outcome, with delirium, complications, resource use, rehabilitation and surgical failure examined less consistently. Models generally showed moderate to strong discrimination, but external validation was inconsistent, calibration was rarely reported and uncertainty estimates were incomplete.

Why it is interesting: Hip fracture mortality is among the most-studied prediction tasks in medicine, and this finds the literature still short of the checks that would make a model usable.

Source
Current Osteoporosis Reports, 22 September 2026
DOI
10.1007/s11914-026-00986-x
Type
Peer-reviewed article
Design
PRISMA systematic review and meta-analysis of prediction model performance, 40 studies, 2010-2026
Verdict
Confirms prior evidence
Driver
AI in medicine