November 16, 2026 – SimonMed has announced that Sean Raj, MD, MBA, chief medical officer and chief innovation officer, will present two scientific studies at the Radiological Society of North America (RSNA) 2026 in Chicago.
The pair of studies draw on data from across SimonMed's nationwide outpatient network. One examines how AI-driven worklist prioritization shortens the time it takes to report fractures on musculoskeletal (MSK) X-rays. The other evaluates how fine-tuning a vision-language foundation model (VLM) on outpatient data affects the quality of automatically generated chest X-ray reports. The research highlights SimonMed's work advancing next-generation AI technology designed to interpret medical images and generate clinically relevant reports, with the potential to improve both diagnostic performance and radiologist workflows.
Fracture Research
SimonMed's fracture research evaluated the impact of an AI-based worklist prioritization system on report turnaround time for musculoskeletal (MSK) radiographs across a large-scale, nationwide outpatient imaging network. Researchers analyzed more than 510,000 MSK radiographs from over 200 centers, tracking turnaround from before the AI system was introduced in 2022 through its full integration in 2025.
Mean turnaround time fell 83.8% for exams showing a fracture, from about two days to under eight hours. The integration resulted in a drastic and sustained reduction in reporting time for fractures, and those gains were achieved without penalizing routine workflow. Turnaround for fracture-negative studies improved as well. For referring physicians, that means actionable results arrive within hours instead of days.
Automated Radiology Reporting
Dr. Raj will present the paper “AI Worklist Reprioritization: Measuring Fracture Detection Turnaround Time for MSK X-Rays in Nationwide Outpatient Imaging” on Dec. 2.
Earlier that day, Dr. Raj will present a poster evaluating the clinical performance and workflow impact of a vision-language foundation model for automated outpatient chest X-ray report generation, before and after institution-specific fine-tuning. This improvement demonstrates how adapting foundation models to real patient populations can improve their ability to detect more findings. The investigational model, trained on more than 5 million chest X-ray image-report pairs, was fine-tuned on more than 150,000 SimonMed adult outpatient exams and tested on a consecutive cohort from 143 outpatient centers.
Institution-specific fine-tuning improved sensitivity with a modest reduction in specificity. The largest gain came in pulmonary nodules where sensitivity rose from 62.6% to 80.8%. The findings support improved generalizability through local model adaptation, and suggest fine-tuned report generation models may reduce radiologist workload while maintaining clinically acceptable performance.
AI is transforming not just how quickly we identify important findings, but how radiology reports themselves are created,” said Dr. Sean Raj. “Our research demonstrates how AI can reduce fracture reporting times from days to hours. Even more exciting, we're exploring autonomous chest X-ray report generation using vision-language models fine-tuned on data from our own outpatient practice. This represents an important step toward a future where AI can help radiologists work more efficiently and deliver meaningful insights to patients and physicians faster.

November 19, 2026 