AI Triage Cuts Fracture Turnaround to Under 8 Hours

SimonMed's RSNA 2026 study reports fracture-positive turnaround fell 83.8% in a deployed system, while its chest X-ray report model remains at research stage.

Updated on Nov. 20, 2026 in Artificial Intelligence

AI Triage Cuts Fracture Turnaround to Under 8 Hours

SimonMed presented research at RSNA 2026 on AI-driven worklist prioritization for fracture reporting and a vision-language foundation model for automated chest X-ray report generation. The fracture work is described as an integrated system, fully deployed in 2025, while the chest X-ray report generation model is research-stage and has not been described as an available product.

Why it matters

The fracture study suggests AI can reorder radiology queues in ways that cut turnaround for urgent findings, a workflow problem central to imaging departments. The report-generation work points toward models that could draft radiology reports, a direction the company says it is exploring toward autonomous report generation.

The fracture study analyzed more than 510,000 musculoskeletal radiographs from over 200 centers, and mean turnaround for fracture-positive exams fell 83.8% after AI worklist prioritization, from about two days to under eight hours. The chest X-ray model was trained on more than 5 million image-report pairs and fine-tuned on over 150,000 SimonMed exams, but the source does not report its specificity or any external validation.

The players

SimonMed

SimonMed is a U.S. outpatient imaging provider whose research group is developing AI tools for radiograph triage and vision-language models for chest X-ray reporting.

Sean Raj

Sean Raj is SimonMed's chief medical officer and chief innovation officer, presenting the company's AI research at RSNA 2026.

The details

The worklist system reorders the queue of musculoskeletal radiographs so that studies likely to show fractures are read sooner, rather than in order of arrival. The chest X-ray work uses a vision-language foundation model, a neural network that learns to connect images with text, first trained on more than 5 million chest X-ray image-report pairs. It was then fine-tuned on SimonMed outpatient exams and tested on a consecutive cohort from 143 outpatient centers. Fine-tuning on institution-specific data improved sensitivity, the model's ability to catch true positives, at a modest cost to specificity, its ability to rule out disease.

Timeline

  1. 2022: SimonMed introduced its AI-based worklist prioritization system.

  2. 2025: The worklist prioritization system was fully integrated into SimonMed's workflow.

  3. November 16, 2026: SimonMed announced its RSNA 2026 presentations.

  4. December 2, 2026: The fracture-detection paper is scheduled for presentation at RSNA in Chicago.

The Tech Race

The work is presented at RSNA 2026 in Chicago, the annual meeting where radiology researchers compare AI tools for imaging workflow. It places SimonMed's triage system and its chest X-ray report generation research alongside other AI efforts in radiology, though the source does not name specific competing programs.

Patients with musculoskeletal injuries could see fracture results returned sooner if worklist prioritization is adopted more widely, since SimonMed reports fracture-positive turnaround falling to under eight hours. The chest X-ray report generation model is research-stage, and the source gives no timeline for availability or pricing.

The takeaway

AI triage is moving from research into deployed radiology workflow, while automated report drafting remains at the research stage. Watch the December 2, 2026 RSNA presentation of the fracture-detection paper for the full study data, including the fracture-negative turnaround results the source does not quantify.

What happens next

The fracture-detection paper is scheduled for presentation at RSNA 2026 on December 2, 2026.

Further reading

For more on the tools reshaping clinical work, see Artificial Intelligence coverage across the United States.

Source note: This article includes information reported by Imaging Technology News.