Medicare is building a payment lane for AI-assisted care before anyone has decided who answers when the AI is wrong. The default answer is the physician who signed.
Medicare is pricing the machine
This month Medicare started asking how to pay for AI. Buried in the 2027 Physician Fee Schedule final rule, released November 4, is a request for information on technology and AI in primary care. CMS wants to know how AI could support the annual wellness visit by collecting pre-visit data, assessing health risk, and turning patient questionnaires into language patients actually understand. It also wants to know how AI saves Medicare money.
This is primary care, not podiatry. Not yet. But it is the first time CMS has formally asked how tech-enabled care should be valued inside the fee schedule. Payment ideas that start in primary care rarely stay there.
The same rule shows where CMS wants accountability to land. Starting in 2029, Medicare will pay for remote patient monitoring only when the patient’s own physician renders it. Third-party monitoring vendors are out. The agency is willing to pay for technology, as long as a physician’s name is attached to it.
The window into the machine got smaller
The accountability side moved the other way. In late October, HHS finalized the HTI-5 rule. Among its changes, it removed the transparency, documentation and annual review requirements for AI-enabled decision support in certified EHRs. Those were the requirements that obligated vendors to tell us what a predictive tool was built on and how it was tested.
That same week, the American Medical Association published an issue brief on AI liability. Its position is reasonable and overdue. If physicians are going to be liable for AI recommendations, they need to be in a position to meaningfully assess the outputs. Developers, and the health systems that deploy these tools, should carry some of that liability. Without those agreements, the AMA says, physicians should be able to decline the tool.
Where that leaves the physician
Put those two developments side by side. The payment conversation is moving forward while the transparency conversation moves back, and the liability defaults to whoever signed the note. That is us.
None of this is a reason to avoid AI. I am the editor of this newsletter, and I hired AI to do my research and first drafts. I review everything before it goes out. Podiatrists already know how to supervise. We run clinics, train residents, direct teams in the OR and sign for all of it.
The standard we apply to people is the right standard for software. We sign what we can see. When a vendor cannot show us how a tool reached its answer, that tool has not earned a place in front of a patient.
The signature at the bottom of a note has always meant a physician reviewed the work. That meaning does not change because software wrote the first draft. The podiatrists who hold that line are the ones patients, payers and partners will trust with the pen.
The Brief
Everything else from the past two weeks that matters to a practicing podiatrist: the research, the rules, the technology and the trade. Each item links to its source.
Research
Fibular nailing vs plating, again
Foot & Ankle Orthopaedics published a new comparison of operative time and complication rates for rotational ankle fractures fixed with a fibular intramedullary nail versus a plate on November 4 (FAO). The paper lands on an evidence base that already leans one way. A 2025 meta-analysis found nailing carried a lower risk of wound complications, about five minutes less operative time, and fewer cases of symptomatic hardware, with comparable union, though event rates were low and the studies small (Cureus / PubMed). If you still plate every Weber B, the literature keeps building the case for a second look, particularly in older patients and those with poor soft tissue.
MIS vs open Haglund resection
Another FAO paper (November 3) compared patient-reported outcomes after minimally invasive and open Achilles debridement with Haglund resection. The study covered four fellowship-trained surgeons at one institution from 2018 to 2025 and found similar PROMIS outcomes between approaches, with one postoperative Achilles avulsion in the open group and none in the MIS group (PMC). It is retrospective, but it adds to a steady signal that MIS does not cost patients function.
The second metatarsal under a load
FAO also published work from Padhye, Willson, Meardon and colleagues on second metatarsal stress during prolonged load carriage (November 4). The same group previously showed that a vest carrying 10% of body mass raised peak second metatarsal stress about 3%, and running 10% faster raised it about 4% (PubMed). Rucking has become a mainstream fitness trend, so expect to see this pattern well beyond military recruits.
Bisphosphonates before Charcot surgery
JFAS published a retrospective cohort from Narayan and colleagues on November 7 examining whether prior bisphosphonate use relates to foot and ankle surgery in Charcot patients (JFAS). We have only the title and abstract listing so far and will report the numbers once we have read the full paper. For context, prior work found bisphosphonates raise foot bone mineral content in active Charcot, but long-term deformity and amputation rates were similar across treatments (PLOS One).
Add a staple to the ankle fusion
Sasaki and colleagues report in JFAS (November 7) that adding a compression staple was associated with greater early CT-based bone healing after open ankle arthrodesis for end-stage osteoarthritis (JFAS). Early CT healing is a surrogate. It is not the same as fewer nonunions, but it is a cheap construct change worth watching.
ChatGPT vs FootCareMD
Casciato, Mateen, Cooperman, Pesavento and Brandão compared patient information on common foot and ankle conditions from ChatGPT with the AOFAS patient site FootCareMD, published in JFAS (November–December 2024) (JFAS). ChatGPT’s answers read at a college level, with a Flesch-Kincaid grade of 13.79 versus 9.60 for FootCareMD, and were harder to read on every condition but pilon fracture (PubMed). ChatGPT scored 0 on the JAMA benchmark, which checks for authorship, sources, disclosure and currency, against 1.95 for FootCareMD, and 52.5 versus 66.9 on the DISCERN reliability scale, all p < .001 (PubMed).
The study used ChatGPT 3.5, and the models have moved since. The same lead author later found that blinded fellowship-trained surgeons preferred ChatGPT-4.5 summaries of foot and ankle literature over professionally written ones in 73% of comparisons (ScienceDirect). The finding that holds is the JAMA score. An answer that names no source gives the physician nothing to check, which is the argument of this week’s lead.
JAPMA
The two-week window was light: a qualitative study of foot pain in early rheumatoid arthritis (November 2), a subungual fibrolipoma case (November 3), and a lateralplasty series for ingrown nails with a 14.8% recurrence rate (JAPMA). The journal’s most-read paper remains Rogers, Alper and DeHeer’s critical appraisal of plantar fascia embolization, which finds the evidence limited (JAPMA). It is the paper to have read when a patient asks about the procedure.
GLP-1s and bone: what the review actually says
The GLP-1 bone story made the rounds in podiatry media this week, so we traced it to the source. The review is a systematic review of 16 human studies from 2011 to 2025, published this spring in the Journal of the American Osteopathic Academy of Orthopedics by a team of osteopathic medical students and an orthopaedic resident (JAOAO).
The findings are mixed by drug. Liraglutide preserved bone density in some studies, and exenatide was associated with less bone resorption. Semaglutide was linked to increased resorption and lower bone density, and the authors found no consistent reduction in fracture risk across the class (JAOAO). The semaglutide concern in that review rests largely on a single phase 2 trial.
Larger analyses point the other way on fractures. A 2025 meta-analysis of 25 studies in Acta Diabetologica found GLP-1 receptor agonists were not associated with increased fracture risk (RR 0.80, 95% CI 0.47 to 1.36) (Acta Diabetologica).
For the surgeon, the honest answer is that the population fracture question looks reassuring, while the question that matters in the OR is still open: how rapid weight loss affects bone quality and healing in the patient in front of you, scheduled for a fusion. Until better data arrive, document GLP-1 status and rate of weight loss at the pre-op visit. Weigh a bone health workup before elective arthrodesis in patients losing weight quickly, alongside the anesthesia team’s holding protocol.
Washington
The 2027 fee schedule is final
CMS released the rule on November 4. For clinicians outside an advanced APM, the conversion factor falls 2.31%, from $33.40 to $32.63, which is steeper than the July proposal because a one-time 2.5% bump for 2026 expires (Fierce Healthcare). Dermatology, otolaryngology and podiatry still take the largest hits from the policy changes, though smaller than proposed (Medical Economics).
Same-day visits
CMS softened its same-day E/M cut from 50% to 20%, and it applies only to procedures with a 0-day global period, not 10-day or 90-day procedures (MSMS). APMA denounced the cut as harmful to practices and patient access (APMA). MGMA called on Congress to reverse the conversion factor cut and the same-day change and to pass the Patients First Act (via PM Pulse).
Remote monitoring changes hands
Starting in 2029, Medicare will pay for remote patient monitoring only when the patient’s own physician renders it, which cuts out third-party monitoring vendors (Second Opinion). Remote therapeutic monitoring will be limited to established patients (MSMS). If your diabetic foot monitoring program runs through an outside vendor, you have two years to decide whether to bring it in-house.
Skin substitutes
The final rule prices non-sheet skin substitutes per square centimeter, matching the sheet-form rate (HMP Wounds).
MIPS
The maximum positive adjustment for 2027, based on 2025 performance, is about 1.38% (via PM Pulse). Traditional MIPS reporting ends after the 2030 performance period, with a move to MIPS Value Pathways (MSMS).
HTI-5
ONC’s deregulatory health IT rule cut 14 of the certification program’s 60 criteria, far fewer than the 34 it proposed (Healthcare Dive). It removed AI transparency requirements from HTI-1, left information blocking rules untouched, and an HTI-6 proposal could arrive before year-end (Holland & Knight). The rule takes effect December 28 (McDermott).
AI and technology
Scribes meet state law
States are writing rules for the microphone in the exam room. Louisiana requires a spoken disclosure before recording, Rhode Island requires patient notice and physician review of the note, and California requires a disclaimer on AI-generated messages no clinician has read (PYMNTS). Physician use of ambient scribes rose from 20% to 29% between two AMA survey waves, and three-quarters of physicians using AI said it reduced their administrative work (PYMNTS). Minnesota has no such statute yet. A written consent line in your intake paperwork costs nothing and gets ahead of it.
Epic and athena, together
athenahealth confirmed it is working with Epic to make it easier for ambulatory practices to exchange clinical information across the two platforms, with no timeline given (Becker’s via Rama on Healthcare). For independent podiatry practices on athena referring into Epic hospitals, this is the interoperability problem you live with every day.
Who is liable when AI is wrong
The AMA released an issue brief arguing that physicians should not carry all the liability for AI tools they cannot meaningfully assess, and that developers and deploying health systems should share it (Second Opinion; AMA). It is the backbone of this week’s lead essay.
The upload that was not an AI problem
A Johns Hopkins Bayview staff member uploaded a spreadsheet of patient information to a public AI tool the hospital had not approved, and notification letters went out October 22 (Physicians Practice). Lawrence Kosova, DPM, argued in a letter to PM Pulse that this was a staff compliance failure, not an AI failure (PM Pulse). He is right on the cause. The lesson for a small practice is still the same: an approved-tools list and a five-minute staff talk prevent the letter.
The trade press
Podiatry Management has a new editor-in-chief
Robert J. Snyder, DPM, MSc, MBA, took the role effective immediately on November 3 (PM Pulse). He follows Barry H. Block, DPM, JD, whose name led the masthead as recently as the June/July issue (Podiatry Management).
Snyder’s résumé is academic and wound care through and through. He is the former dean and director of clinical research at Barry University School of Podiatric Medicine and a professor at Cardiff University School of Medicine (PM Pulse). He has published more than 165 papers on wound care (HMP Global), is a past president of the Association for the Advancement of Wound Care and the American Board of Wound Management (HMP Global), and serves as an associate editor of JAPMA (HMP Global). His stated goal is to reach more podiatrists, international colleagues and aligned specialties such as vascular surgery, infectious disease and plastic surgery (PM Pulse).
Our read: Podiatry Management is heading toward academic and wound care territory, and Snyder is exceptionally qualified to take it there. That is a real service to the profession. It is also a different job from ours.
The Footnote covers the whole practice. That means the OR, the clinic, the wound center, the ledger, the software in the exam room and the policy that decides what all of it pays. We read the journals so you can practice, and we write for the podiatrist who has to make Monday’s decisions with this week’s information.
Second Opinion Summit
Christina Farr’s Second Opinion held its first in-person summit, SOS, October 19 to 21 in New York, at the Wythe Hotel in Brooklyn (Second Opinion). The invite-only event sold out, and an interest list is already open for 2027 (Second Opinion). We were among the 175 people in the room. The panel on state AI sandboxes reached a telling conclusion: health tech mostly wants to watch Utah, not join it (Second Opinion, Oct 24).
That caution is earned. Utah’s sandbox let an AI company refill certain prescriptions under clinician oversight, and the Utah Medical Licensing Board called for the pilot’s immediate suspension in April (Route Fifty). The two sides now meet weekly, but the state is still deciding how to split legal liability between the AI and the clinicians who oversee it when a patient is harmed (Route Fifty). The industry’s own conference landed on the same unanswered question as this week’s lead.
Since the summit, Second Opinion has hosted a live Q&A with Doximity CEO Jeff Tangney on where clinicians are finding value with AI (Second Opinion). It also ran a post-election panel with Aneesh Chopra, Christopher Kerns and Mike Desjadon on what a Democratic Congress will try to do on health care (Second Opinion).
On the calendar
The next three weeks, from PodiatryMeetings.com and Medical Economics.