Particle.news
Download on the App Store

Predictive AI Becomes Standard in U.S. Hospitals as Digital Divide Persists

An ASTP/ONC analysis ties the rise to EHR-integrated tools, warning of gaps for smaller hospitals.

Overview

  • In 2024, 71% of non-federal acute care hospitals reported using predictive AI integrated into EHRs, up from 66% in 2023.
  • The divide was pronounced by ownership and location, with 86% of system-affiliated hospitals using predictive AI versus 37% of independents, and usage at 81% in urban settings compared with 56% in rural and about 50% in critical access facilities.
  • Hospitals largely relied on EHR vendors for models, with 80% sourcing from their developer; adoption reached roughly 90% among hospitals on the leading EHR compared with about 50% on other systems.
  • Operational use cases grew fastest year over year, with billing up 25 percentage points and scheduling up 16, while higher-risk clinical deployments like treatment recommendations remained less common.
  • Most hospitals reported formal oversight, including accuracy checks (82%), bias assessments (74%), and post-implementation monitoring (79%), often shared across committees and division leaders.