Particle.news

Download on the App Store

Northwestern’s Generative AI Cuts Radiology Report Time by Up to 40% in Clinical Deployment

Unpublished follow-up studies suggest efficiency improvements as high as 80% with plans to extend the technology to CT scans.

Image
Image
Image

Overview

  • Published June 5 in JAMA Network Open, the AI has been embedded across Northwestern Medicine’s 12-hospital network to generate 95% complete radiology reports and triage critical cases.
  • In a study of nearly 24,000 radiographs, radiologists saw an average 15.5% boost in report completion efficiency and individual gains up to 40% without compromising accuracy.
  • The model demonstrated high sensitivity and specificity in flagging life-threatening conditions such as unexpected pneumothorax in emergency settings.
  • Unpublished follow-up data indicate the tool could drive efficiency gains up to 80% and support automated CT scan analysis.
  • Developed entirely in-house using Northwestern Medicine’s clinical data, the generative AI holds two approved patents and is entering early commercial deployment.