Stanford Medicine’s AI system lets clinicians chat with medical records

Doctors at Stanford Health Care just got a new digital sidekick. It's called ChatEHR, and it lets them "chat" with medical records the way you'd chat with ChatGPT. Ask a question like, "Has this patient had a colonoscopy?" — and boom, the answer pops up.

"We're thrilled to bring this to the workforce at Stanford Health Care," said Dr. Michael Pfeffer, chief information and digital officer. And who wouldn't be? This AI-powered tool is turning time-consuming, soul-draining record reviews into something faster, smoother, and surprisingly intuitive.

Developed in-house and still in pilot testing, ChatEHR's already showing signs of being the kind of tech that sticks — like stethoscopes, only smarter.

How does it work?

Think of ChatEHR as a medical record whisperer. It's tucked neatly inside the existing electronic health record system, and when a doctor logs in, they're greeted with a friendly, "Hi, 👋 I'm ChatEHR! Here to help you securely chat with the patient's medical record."

From there, it's as simple as typing a question:

• "Any allergies?"
• "What's their latest cholesterol reading?"
• "Had any surgeries?"
• "Were the results normal?"

The tool pulls real-time data directly from the patient's medical record and gives it back in plain English. Not guesses. Not generic info. The real deal.

Dr. Sneha Jain, one of the first physicians to try it, put it plainly: "ChatEHR can help them get that information up front so they can spend time on what matters — talking to patients and figuring out what's going on."

Beyond single answers, the tool can generate:

• Full patient summaries
• Timeline reviews
• Contextual follow-ups (e.g., "Why was this med discontinued?")

And it doesn't stop there. The team is building "automations" — tasks that can evaluate things like:

• Whether a patient is eligible for hospice
• If they can be transferred to a unit with more beds
• What kind of follow-up they'll need after surgery

In short, it's not just answering questions. It's doing homework.

Why does it matter?

Any doctor will tell you: electronic health records can be a bear. A 200-page chart might hold gold — but first, you've gotta find it.

"It's a ton of work to go back and find all of that information during a time-sensitive case," said Dr. Jonathan Chen, who sees patients in the hospital and also teaches at Stanford. "Speeding up that process would be a big help."

That's where ChatEHR earns its keep. It:

• Cuts chart review time
• Surfaces critical details faster
• Helps make better, quicker clinical decisions

And in situations like ER admissions or patient transfers, where every second counts, those benefits aren't just nice to have — they're life-saving.

"It's not just the chest pain they're having in that moment that matters — it's their whole story," said Chen. "Having ChatEHR boil that down into a relevant summary would make that process smoother."

The context

ChatEHR results from a deliberate, expert-led push to bring AI into health care responsibly.

Back in 2023, Dr. Nigam Shah and his team saw the rising tide of large language models and asked, "How do we make this truly useful for doctors?"

So they built something that:

• Lives inside the existing workflow
• Uses real patient data, securely
• Follows Stanford's responsible AI guidelines

"It's not helpful unless it's embedded in their workflow and the information the algorithm is using is in a medical context," said Shah. And he's not wrong — AI doesn't help if it's out of sync with the real work of caregiving.

Right now, only 33 clinicians are piloting the system. But plans are in motion to expand. Shah's team is using MedHELM, a new open-source evaluation framework, to test accuracy in the wild. They're even working on adding citations, so doctors can see exactly where an answer came from in the record.

What's next? Making ChatEHR available to all Stanford clinicians — and maybe, one day, to health systems everywhere.

As Pfeffer summed it up: "This is a unique instance of integrating LLM capabilities directly into clinicians' practice and workflow."

And you know what? It might just stick.