Stanford Health Care has launched a pilot program combining ambient AI and real-world data to give physicians clinical evidence within minutes during patient visits.
Nigam Shah, MD, PhD, chief data scientist at Stanford Health Care, said the pilot merges ambient AI from Microsoft’s Nuance DAX with technology from Atropos Health, a company spun out of Stanford research. The goal is to generate relevant evidence-based reports in under five minutes—without interrupting the clinical encounter.
“Real-world evidence has been part of my work almost since 2011,” Dr. Shah said in an interview with the Becker’s Health IT podcast. “Most of the time, real-world evidence is an observational study that takes anywhere from six to nine months or maybe one year. It gets published in a journal… and maybe a physician reads it or doesn’t read it. The chance that it affects patient care is very low.”
Stanford’s work to close that gap began with its “green button” project, which reduced the time to generate evidence from nine months to two days. Atropos Health brought it down to just a few hours.
“At the bedside, you have a question. You ask something—what happened to similar patients—and you get a written report before the end of the day,” Dr. Shah said.
The new pilot aims to shorten that timeline further by combining automatic clinical note transcription with real-time data queries to deliver answers in under five minutes.
That development marks a significant step in a broader industry shift toward using AI to streamline workflows, reduce administrative burdens and enhance decision-making at the point of care.
The service will begin with Stanford’s community-based providers, known as University Medical Partners.
“These are physicians who are not at the university, they’re in the community practices all over the Bay Area,” Dr. Shah said. “They’re the ones who are super excited that finally technology is at a stage, at ease of use, that they can incorporate into their care.”
Dr. Shah said the system could be live within 60 to 90 days, with broader rollout continuing through 2025.
Physicians using the technology will no longer need to break focus to take notes or manually search databases. Instead, the AI captures the full conversation, generates a transcript within minutes and—when clinical questions arise—triggers instant literature or data reviews.
“If there’s a question, what does a physician do today? They go to UpToDate or some other summary. Or, if it’s something literature can’t answer, they trigger a real-world evidence report from Atropos. But that’s a manual step,” Dr. Shah said.
The pilot aims to automate that process.
Dr. Shah noted that the potential extends beyond clinical decisions.
“Documentation for billing could be completely automated,” he said. “We might be able to produce summaries that are directed to the patient and outline the three things the patient needs to watch out for.”
Still, challenges remain. Dr. Shah said technical issues like latency and data privacy are just one piece of the puzzle. A larger concern is how to evaluate new AI tools.
“How do you evaluate? How do you know it’s adding value?” he said. “Those are the things I think in the next year everybody is going to be thinking about.”
He encouraged health system leaders to think beyond what technology is already doing and instead imagine what it could make possible.
“What are the actions that we should be taking but are not yet taking that can be done cheaply and fast with AI?” he said.
