AI-Assisted Failure Prediction on Postoperative X-Rays: A Head-to-Head Comparison with Consultant Trauma Surgeons
Source check: Source checked. Not independent scientific confirmation.
Finding: A prospectively assembled 466-patient application cohort (104 fixation failures) was classified from postoperative radiographs by a ResNet-50 trained on a disjoint 200-patient cohort. AI sensitivity was 96.15% and specificity 98.90%, exceeding the three surgeons numerically; the strongest surgeon difference was not significant. This supports an adjunct diagnostic result, pending external prospective validation. This supports an adjunct diagnostic result, not external prospective validation.
Technology: ResNet-50 image classifier
Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- ResNet-50 reads postoperative X-rays and classifies fixation failure against clinical outcome; AI supplies the diagnostic prediction.
- Test and result
- PDF Results pp.6–8: 466 patients, 104 failures; AI classified 458/466 (98.28% accuracy); paired surgeon analysis and comparison with strongest surgeon reported.
- Limitation / still unknown
- Same-cohort paired human comparison and a separately trained cohort support diagnostic performance; same centers and authors’ stated need for external prospective validation limit transportability.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://www.researchsquare.com/article/rs-11408305/v1
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 3/5 · Validation 2/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.