Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Health Ai topic
No spam. Unsubscribe anytime.
Hospitals describe AI benefits and urge state-supported pilots, but stress governance
Summary
Hospital leaders described concrete uses for AI — from missed‑follow‑up detection to ambient documentation — and urged state pilot funding and strong governance. They emphasized human oversight, staged testing and monitoring for model drift.
Hospital system leaders told a joint House Health and Communications & Technology Committee hearing that artificial intelligence is already embedded in many clinical workflows and can both surface missed care and reduce documentation burden, but that deployments require rigorous governance and funding support.
Sridhu Smali of Penn Medicine said AI is best understood as “augmented intelligence” that should elevate clinicians’ judgment, not replace it. He described use cases including record‑scanning to detect overdue imaging follow‑ups and an EHR‑embedded assistant that summarizes lengthy records for bedside action. Smali said his system uses a New Technology Review Committee and stage‑gated testing — in some cases up to randomized trials — before broad deployment.
Dr. Robert Krukletis of the Guthrie Clinic described the health system’s Pulse Center, a 24/7 clinical support model that leverages continuous AI monitoring to detect early signs of sepsis and alert clinicians. “Every single clinical decision at Guthrie is ultimately made by a human,” Krukletis said, and he reported that the system has coincided with “a considerable decrease in the mortality of our patients.” He urged the legislature to establish and fund pilot programs to help hospitals adopt similar AI care models because many rural and smaller hospitals lack resources to build or validate such systems on their own.
Committee members pressed witnesses on procurement, vendor concentration and liability. Representative Ben Waxman expressed concern that a small number of foundation model companies control much of the underlying technology; Smali and other witnesses said many clinical products are built by third‑party vendors or EHR vendors that package models into clinician‑facing tools. On liability, witnesses described contractual indemnification and internal privacy protections but said the complexity of layered vendors makes accountability challenging.
The hearing underscored two recurring themes: hospital leaders see measurable clinical benefits from narrowly scoped, well‑governed AI tools, but expanded use will require state assistance, robust governance, and clearer legal frameworks to address vendor liability and monitoring requirements. The committees said they expect further hearings to explore those issues in detail.
