Artificial intelligence for pediatric fracture detection: impact on diagnostic revisions and patient recall rates in a tertiary emergency setting.
AI fracture detection works well technically but doesn't meaningfully reduce diagnostic revisions or patient outcomes in a hospital setting.
A prospective evaluation of commercial AI fracture detection in 667 pediatric ED patients showed non-significant reduction in diagnostic revisions and patient recall (8.6% to 5.7%) with high AI accuracy (95.1%), but minimal therapeutic impact and unchanged length of stay, suggesting limited incremental benefit in an academic tertiary setting. The authors conclude a confirmatory powered trial may be difficult to justify given the small expected effect size.
What the study was
- Study design
- Prospective quasi-experimental alternating-day comparison (AI vs. no-AI)
- Population
- Pediatric patients 2-18 years with appendicular skeletal radiography during out-of-hours ED care (n=667)
- Sample size
- 667
- Category
- Diagnostics
- Maturity
- Exploratory
- Journal
- BMC Emergency Medicine
Why it surfaced
Prospective real-world AI evaluation for pediatric fracture detection; provides actionable evidence on expected benefit magnitude for AI deployment decisions; BMC open access.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.