Study: AI can aid physician decisions during virtual urgent care

Artificial intelligence just got graded better than actual doctors — at least when it comes to virtual urgent care. A study led by Cedars-Sinai, presented at the American College of Physicians Internal Medicine Meeting and published in the Annals of Internal Medicine, threw AI into the digital exam room alongside human physicians. The results? Well, let's just say the machines held their own. In some cases, they even outperformed the docs.

"We found that initial AI recommendations for common complaints in an urgent care setting were rated higher than final physician recommendations," said Dr. Joshua Pevnick, co-senior author and co-director of the Cedars-Sinai Division of Informatics. Turns out, the bots were especially sharp at catching red flags like antibiotic-resistant UTIs — and even recommended cultures before scripts.

Let's break down what this all means and why it could change your next virtual doctor's appointment.

How does it work?

Picture this: you're feeling lousy, and you tap open an app and start answering questions about your symptoms. No waiting room. No clipboard. Just 25 smart questions in five minutes.

• Patients using Cedars-Sinai Connect begin visits via mobile, describing symptoms and basic info
• An AI model, built by K Health, interviews them dynamically and pulls in electronic health record data
• It then offers potential diagnoses and treatment suggestions
• Patients can move on to a video call with a physician — who may or may not peek at the AI's suggestions

Here's the kicker: during the study, docs had to scroll down to see the AI's recommendations. So, not all of them did.

"The major uncertainty of this study is whether physicians scrolled down to view the prescribing, ordering, referral or other management suggestions made by AI," noted Dr. Caroline Goldzweig, Cedars-Sinai Medical Network's chief medical officer.

Why does it matter?

Well, for starters, time is money — and both are in short supply in healthcare. If AI can frontload patient data collection and decision support, doctors can do what they do best: connect, clarify, and make tough calls.

• AI flagged critical issues faster and more consistently
• It helped avoid premature prescribing, like giving antibiotics before knowing the bug
• Its suggestions were, on average, rated more appropriate than the human ones

Still, humans aren't out of the picture. Doctors brought the nuance. They could dig deeper into personal histories and adjust recommendations on the fly. As Dr. Pevnick put it, "Physicians were better at eliciting a more complete history from patients."

So no, it's not robot vs. doctor. It's more like tag-team medicine.

The context

Cedars-Sinai Connect launched in 2023 to serve patients across California through virtual urgent and primary care. It's a joint venture with K Health, whose AI systems are built to lessen the paperwork burden and get doctors back to doctoring.

This study analyzed 461 visits from June to mid-July 2024, covering symptoms from respiratory issues to dental pain. Researchers from Tel Aviv University, including first author Dr. Dan Zeltzer, were also in on the project.

Ran Shaul, K Health's co-founder, put it plainly: "We put AI to the test in real-world conditions, not contrived scenarios... You're dealing with complex human beings, and any given AI has to deal with incomplete data and a very diverse set of patients."

By feeding the system a steady diet of clinical notes and live decision-making feedback, they made the AI smarter. Not perfect. But smart enough to play ball with the best.