UVM receives up to $38M to build AI digital twins for critically ill patients

By Splitfeed EditorialLast updated Sep 30, 2026, 21:30 UTC

On September 30, 2026, the University of Vermont was awarded up to $38 million to build artificial intelligence-based digital twins of critically ill patients. The grant, described by the university and local outlets including WCAX and VTDigger as a record for the institution, will fund the development of individualized AI models that replicate a specific patient's physiology and condition. The intention is for clinicians to use these virtual models to test and guide treatment decisions before applying them to the actual patient.

The significance of this project lies in what digital twins would represent if they work as described: a shift from population-level treatment guidelines toward fully individualized clinical decision-making in intensive care settings, where the stakes and the pace of decisions are both extremely high. The core questions in dispute are whether AI models can accurately enough capture the complexity of a critically ill individual to be clinically useful, how errors in the model would be caught before they influence care, and what regulatory and liability frameworks would govern their use. The scale of the funding signals institutional confidence, but the gap between a promising research grant and a validated clinical tool is substantial.

None of the three camps had published reactions to this story at the time of reporting, so what follows reflects what each would typically argue. Pro-AI voices would likely frame this as exactly the kind of high-value, life-saving application that justifies large public investment in AI research, pointing to the potential to reduce trial-and-error medicine in critical care. Anti-AI voices would probably raise concerns about over-reliance on models that could embed biases or fail in ways that are difficult to detect, particularly for vulnerable patients who cannot advocate for themselves. The middle ground would typically welcome the research funding while calling for rigorous clinical validation, independent oversight, and transparency about the models' limitations before any deployment at the bedside.

The clearest next markers to watch are the publication of early research results from the UVM team, any regulatory guidance from the FDA on AI-based clinical decision tools in critical care, and whether peer-reviewed trials eventually compare patient outcomes with and without digital twin guidance.