Chest X-ray AI flags high fracture risk in 53k-patient NEJM AI study
zakkohane · x · 2026-11-17
Zak Kohane shares a NEJM AI study using deep learning to opportunistically identify patients at high fracture risk from routine chest X-rays, without extra DXA scans.
- Development cohort: 53,109 patients from institution A (3,084 fracture-positive); external testing on 85,284 patients from two other institutions
- Architecture: DenseNet-161 backbone with a discrete-time logistic hazard survival head producing cumulative survival probability curves
- Kohane finds the results encouraging but notes a prospective study in an unselected population is needed for a robust PPV
More from Models
- "No Taste but Genuinely the Best": Early Verdict on GPT 6 Astra — justalexoki · 2026-11-17
- Google expands Antigravity program to 10,000 scientists with free Gemini 4 Carbon access — GoogleDeepMind · 2026-11-17
- Google DeepMind unveils Gemini 4 Carbon for physics, math and life science research — GoogleDeepMind · 2026-11-17
- Gemini 4 Carbon proves Signal crypto security in weeks, finds 40x more protein sequences — GoogleDeepMind · 2026-11-17
- ThursdAI Podcast: Claude Fable 5.2 Autonomously Wrote Metal Shaders From a Bare Intent — thursdai_pod · 2026-11-17
- Unsloth adds computer use for local models: Qwen3.8-27B builds expense reports in GIMP — danielhanchen · 2026-11-17