AI Won’t Replace the Physician You’re Referring To

By Traci A. Kimball, MD, MBA — Founder, The WISH Clinic® and Ekagra Health AI™

The patient you refer arrives carrying two wounds

When you send me a patient with a non-healing ulcer, you are sending a clinical problem you have already characterized. Vascular status, glycemic control, offloading, adherence — you have usually worked the differential further than the referral form has room for.

Something else arrives with them, and it is rarely in the chart.

I wrote about this recently in the AIMEDENT Journal. Every patient who walks into my wound care clinic arrives with two injuries. The first is the one you referred. The second is a deep, persistent loss of faith that the healthcare system is capable of healing them at all.

If you treat chronic wounds, you have met this patient. They have been to three clinics. They were told the wound would close in six weeks, twice. They answer questions briefly because they expect the appointment to be short. By the time they reach me, the second wound is often doing more to limit the outcome than the first.

That is the context in which our field is now being handed artificial intelligence, and the reason I think the conversation about it has started in the wrong place.

How we got here: the documentation engine

Medicine began as a humanist discipline. Then the fee-for-service era arrived, and data replaced discernment. Physicians became documentation engines, with clinical time consumed by EHR fields.

I do not need to explain this to you. You live it. The referral you sent me took longer to document than to decide.

What matters for wound care specifically is that the thing the documentation burden crowds out is precisely what chronic wounds need most. Wound healing is not a single decisive intervention. It is months of adherence — offloading worn as prescribed, dressings changed correctly, glucose managed, appointments kept. Adherence runs on trust, and trust is built in minutes of attention that the schedule has spent two decades squeezing out.

So when the promise of AI in medicine is pitched primarily as throughput — see more patients, close notes faster — it is solving for the wrong variable. Do not assume efficiency and healing are the same thing.

The co-pilot, not the replacement

The distinction I draw is between AI that removes friction and AI that removes judgment. The physician is not a problem to be automated away. The physician is the point.

Applied to a wound care practice, that produces a fairly clear line.

Appropriate for AI: prior authorization packets, denial management, documentation drafting, coding support, appointment logistics, tracking which measurements are due. Administrative mass that has no clinical content and consumes clinical hours.

Not appropriate for AI: the decision to debride, and how aggressively. Whether a limb is salvageable. Whether an ulcer’s failure to progress is arterial, infectious, nutritional — or the fact that the patient cannot afford the offloading device and has not said so. That last one is worth sitting with. It is the kind of finding that surfaces only when someone has the time and the standing to ask a second question, and no model currently in clinical use will extract it from the chart.

The line is not permanent, and I do not pretend it is. Image-based tissue assessment and healing-trajectory prediction are advancing quickly and will earn a role. But the burden of proof sits with the tool, and the sequence matters: AI must learn bedside manner before it learns billing codes.

What this means for the referral you just sent

A philosophy of technology is only worth publishing if it changes what a referring provider can expect. Mine comes down to three commitments.

You get a real note, and you get it back promptly. The measure of whether administrative automation is working is not how fast we close our charts. It is whether you hear what we found, in time to act on it, in a form that tells you something. A note that arrives in four days and says what I actually think is worth more than one that arrives in four hours and reads like a template.

Your patient gets time, not a faster line. Any capacity we recover from administrative work goes back into the visit — into the second question, the reason the last device went unworn, the conversation about what limb preservation will require over the next year. I am not interested in using it to shorten appointments.

A person owns the decision, and you can reach them. Whatever assists our documentation, the clinical reasoning about your patient is mine. If my assessment diverges from yours, that is a conversation between two physicians, not a report you receive.

The patient you referred has, in all likelihood, already been let down by a system that promised a timeline it did not meet. My obligation is to be the referral that does not repeat that.

Do no hubris

In that essay I proposed an addition to the oath: do no hubris. The warning is against assuming that a model performing well on a benchmark is a model ready for a patient, and against mistaking efficiency for healing.

For a wound care practice the stakes are concrete. A prediction that an ulcer is on a healing trajectory, accepted without examination, is a delayed debridement or a missed infection. A triage tool trained on populations unlike ours is a patient deprioritized. The failure mode is not dramatic. It is a limb lost slowly, to a series of individually reasonable deferrals.

The other line I would underline for any colleague evaluating these tools is this: change does not succeed because technology exists. It succeeds because people are ready.

Referring a patient to WISH

The WISH Clinic® provides advanced wound care and limb preservation in Arvada, Colorado. We accept referrals for chronic and non-healing wounds, diabetic foot ulcers, venous and arterial ulcers, pressure injuries, and post-surgical wounds that have stalled.

To refer a patient, or to talk through a case before you do, call 303-940-1611.

About the author

Traci A. Kimball, MD, MBA is a physician executive specializing in limb preservation, chronic wound management, and value-based care. She is the founder of The WISH Clinic® in Arvada, Colorado, and of Ekagra Health AI™, a wound-care data platform built for practices working in value-based and ACO networks. She advises healthcare organizations on innovation and transformation, and was named a 2026 Ambassador by the American Board of Wound Management.

Disclosure: Dr. Kimball is the founder of Ekagra Health AI™, a healthcare data and intelligence company serving wound care practices. The views in this article are her own.

Source

Kimball, Traci A., MD, MBA. “The Wound and the System: Why AI in Medicine Must Begin with Humanity.” AIMEDENT Journal, Vol 1:4. Published June 3, 2026 by World’s Top Doctors. Read the full essay →

This article is an original piece by The WISH Clinic® drawing on Dr. Kimball’s essay. Quoted passages appear by attribution; her full argument is in the original.