Daily Briefing for October 4, 2026

Show Notes

FDA plans final AI device, surgical robot guidance; five systems hold off on Epic AI tools; Senate passes hospital cyber bill; clinicians favor AI discharge notes

The voices on this episode are generated, and so is the script, from the day's post at Physician in the Loop. Every fact in the post links to its source.

Read the post and its sources

The Food and Drug Administration plans to finalize guidance on AI-enabled devices and surgical robots in fiscal 2027 and to draft evidence recommendations for generative AI conversational devices for mental disorders, its priority list shows.

The device center's list for fiscal 2027, which MedTech Dive reported Oct. 2, puts seven final guidance documents and one draft on its top tier. Separately, five health systems told Becker's Hospital Review they are holding off on some of Epic's AI tools, and Children's Healthcare of Atlanta said one of them produced too many inaccuracies about diagnoses and patient locations.

The Senate passed a hospital cybersecurity bill with provisions for rural providers by unanimous consent Oct. 1, sending it to the House. In a Stony Brook study, clinicians preferred discharge summaries drafted by a large language model over human-written ones in 95% of 60 longer hospitalizations.

In this episode

• FDA puts final AI device and surgical robot guidance on its fiscal 2027 A-list, with a draft on generative AI for mental disorders
• Five health systems hold off on some Epic AI tools; Children's Healthcare of Atlanta cites inaccurate diagnoses and patient locations
• Clinicians prefer AI-drafted discharge summaries in 95% of 60 hospital stays of seven to 21 days in Stony Brook study
• Breast ultrasound AI reaches 0.952 AUROC for biopsy triage on outside cohort; four radiologists miss fewer cancers with it
• Senate passes Health Care Cybersecurity and Resilience Act by unanimous consent, with provisions for rural providers
• Nearly 12 million people in 41 states live in broadband deserts; 650,000 rural residents also lack nearby ambulance and health facilities
• Google says its Gemini 4 Argon model found a critical flaw in health care software used by hospitals worldwide

What is Physician in the Loop?

How technological innovation is changing medicine for patients and the people who practice on the front lines.

Voice 1: This is Physician in the Loop, the daily briefing on AI in medicine. It's Sunday, October 4th, 2026.

Voice 1: The FDA's device center just published its priority list for fiscal 2027, and artificial intelligence sits at the top of it: seven final guidance documents, led by AI-enabled devices and surgical robots.

Voice 2: Seven final documents in one year is a lot. Which ones are actually locked in, and which are still being argued over?

Voice 1: Also today: five health systems holding off on Epic's AI tools, a Stony Brook study on AI discharge notes, and the Senate's hospital cybersecurity bill. The top tier includes final guidance titled “Marketing Submission and Lifecycle Management Recommendations for Artificial Intelligence-Enabled Device Software Functions,” plus premarket submission recommendations for robotically assisted surgical devices.

Voice 2: And the one item that isn't final yet?

Voice 1: Evidentiary recommendations for generative AI conversational devices for mental disorders. That's the only draft on the top tier, everything else there is final guidance.

Voice 2: So the AI chatbot question for mental health is the unsettled one. What about the surgical robot guidance, is that open for comment still?

Voice 1: Yes. That draft went up in September and comments close November 24th. There's also a second tier: a draft on device software policy, a draft on postmarket cybersecurity management, and final guidance on pulse oximeter testing and labeling.

Voice 2: Busy office either way. Meanwhile five health systems just told Becker's Hospital Review they're holding off on some of Epic's own AI tools.

Voice 1: Holding off, not walking away entirely. What's the actual complaint?

Voice 2: Children's Healthcare of Atlanta paused Epic's inpatient insights feature. Its CIO, Jeremy Meller, said it produced too many inaccuracies about diagnoses and patient locations, plus narratives that ran on too long. It's on a cycle for reevaluation.

Voice 1: Getting the patient's location wrong is an odd failure mode for software.

Voice 2: The post doesn't explain why, just that it happened. St. Luke's Health System in Boise is sticking with Ambience Healthcare instead of Epic's Art. Penn Medicine is holding off on Epic's Emmie too, leaning on Ambience and its own tools.

Voice 1: And the rest?

Voice 2: Atlantic Health's CIO, Sunil Dadlani, said they're taking what he called “a measured approach to tools that make autonomous clinical decisions or engage patients without appropriate human oversight.” Baptist Health in Jacksonville says it's weighing maturity and return on investment.

Voice 1: Five systems, five flavors of caution, all pointed at the same vendor.

Voice 2: A Stony Brook study put AI-drafted discharge summaries up against human-written ones for 60 patients hospitalized seven to 21 days.

Voice 1: Which one did clinicians prefer?

Voice 2: The AI version, in 95% of cases. Reviewers rated it higher for quality, readability, factuality and completeness, and marked fewer factual errors or missed details in it than in the human-written ones.

Voice 1: Any catch on harm, or does that number hold up?

Voice 2: No significant difference in estimated potential for harm between the two. Though primary care physicians flagged more omissions and rated harm likelihood higher than hospitalists did reading the same summaries.

Voice 1: What about incidental findings, the things radiology turns up that nobody was specifically looking for?

Voice 2: Of 31 incidental radiology findings in the AI summaries, about 93% were stated correctly and 87% were judged appropriate to report. It was a retrospective evaluation, so nothing here was tested on live patients in real time.

Voice 1: Still, 95% preference is a striking number, retrospective or not.

Voice 2: The Senate passed the Health Care Cybersecurity and Resilience Act by unanimous consent on October 1st.

Voice 1: Unanimous consent is rare. What does the bill actually require?

Voice 2: It's meant to strengthen coordination among federal agencies on health care cybersecurity and improve security practices at rural providers specifically. The release doesn't state an authorized dollar amount for any of it.

Voice 1: No number attached at all?

Voice 2: None given. Senator Maggie Hassan, one of the co-sponsors, said it would help facilities with fewer resources “strengthen cybersecurity and respond faster to attacks.” Senator Mark Warner urged the House to act next.

Voice 1: Who sponsored it originally?

Voice 2: Senator Bill Cassidy of Louisiana, who's also a physician. Co-sponsors include Hassan, Cornyn, Warner, Hyde-Smith and King.

Voice 1: So it just needs the House to move now.

Voice 2: That's where it sits.

Voice 1: Before we go, something to watch. The committee vote on Heidi Overton's nomination for FDA commissioner has been put off until after the midterm elections.

Voice 2: Right while that generative AI mental health draft is still open. Comments on it are due October 19th.

Voice 1: So the leadership question and the policy question are both parked at the same time.

Voice 2: Second thing worth flagging: Epic's Ergo Visit is already live at Ochsner, and the company's Agent Factory arrives more widely in 2027.

Voice 1: Arriving more widely right as five systems are pumping the brakes on the tools already out there.

Voice 2: Hard to call that a vote of confidence from the field it's rolling out to next.

Voice 1: Sources and corrections are at physicianintheloop.org. This briefing was read by AI voices.