Offcall Physician Spotlight: Meet Dr. Kameron Matthews

Offcall Team

Have a response for Dr. Kameron Matthews? Reply to this post directly — she’ll personally read every message.

Sign up for our newsletter

On/Offcall is the weekly dose of information and inspiration that every physician needs.

1. Kameron, what's one task you've fully handed off to AI that you'd never take back?

The question people ask me about AI in medicine usually focuses on diagnostics or population health analytics. My honest answer starts somewhere more practical: clinical research at the point of care.

For years, the traditional textbook approach or even UpToDate was the default. You had a question mid-encounter, you opened it, you got a reasonable answer, you moved on. It worked. But it was also slow, often behind the curve on emerging evidence, and — if I'm being honest — it trained a certain kind of lookup behavior rather than genuine synthesis.

I've shifted to OpenEvidence, and I'm not going back.

What's different isn't just the speed. It's that I'm getting answers grounded in primary literature, with citations I can actually verify (and I do), synthesized in a way that maps to the clinical question I'm actually asking — not a protocol written for the median patient. When you spend your career caring for Medicaid patients with five comorbidities, housing instability, and medication access issues, the median patient assumption breaks down constantly. I need a tool that handles nuance, not one that hands me a treatment ladder designed for someone else's population.

I still verify. I still apply judgment. The physician is still accountable. But the research task — the "let me look that up" that used to eat three to five minutes mid-visit or twenty minutes after clinic — that's handed off. And the quality of what comes back is better than what I was doing on my own in that time window. I'm comfortable having a partner for medical knowledge, while I use my time more appropriately with the patient.

2. Are you worried about deskilling — that the next generation won't develop the intuition you did because the model handled it?

This is the question I find myself returning to most, and I want to be careful not to be glib about it.

When I was a staff physician at Cook County Jail, I saw patients who were profoundly ill — often with conditions that had gone undiagnosed for years — in an environment where resources were limited, patient autonomy was non-existent, and the clinical complexity was extraordinary. That experience built something in me that I genuinely don't know how to replicate in a curriculum. It was the friction that created the formation.

The honest concern about AI is that it removes productive struggle. When a model surfaces the top differential before a resident has had to generate it independently, we may be accelerating surface efficiency while hollowing out clinical reasoning. The analogy I think about: GPS has made it easier to get places but harder to develop spatial memory. At the scale of navigation, that's mostly fine. At the scale of medicine — when the GPS fails, or encounters a case it wasn't trained on — the stakes are different.

That said, the deskilling anxiety can also become gatekeeping. The question isn't just whether the next generation will develop my intuition. The question is whether we're designing training environments intentionally enough to ensure they develop better intuition — calibrated for an AI-augmented practice, with clear protocols for when to override the model. We're not close to having that figured out.

3. Will AI solve the physician shortage or quietly make it worse?

The physician shortage framing often obscures what's actually a physician maldistribution problem. We don't have too few physicians in aggregate — we have too few willing to work in the places and with the populations that need them most. AI doesn't automatically fix that. It can amplify existing patterns.

During my time at the VA overseeing the Office of Community Care, I saw the complexity of extending care to veterans in communities across the country through a combination of direct care and community partnerships. The MISSION Act created new pathways. But technology, policy, and resource availability had to be aligned deliberately — it didn't happen organically.

At Cityblock, we cared almost exclusively for Medicaid and dually eligible patients — people with multiple chronic conditions, housing instability, and behavioral health needs. AI tools that work beautifully on a commercially insured, digitally literate, English-speaking population often require adaptation to work on our patients, with the workflows that make most sense with teams and interoperable systems focused on that population. Training data, language models, risk stratification tools — they carry the biases of the systems that generated them.

My answer: AI will likely solve the physician shortage for the people who were already close to having it solved. For communities like the ones I've spent my career in, it depends entirely on whether we treat equity as a design requirement from the beginning — not an afterthought, and not a DEI footnote at the end of a product launch.

2025 PHYSICIANS AI REPORT

Access the Complete Report

✓Complete quantitative breakdown of what physicians really think about AI

✓Strategic implications for healthcare organizations and AI companies

✓Sentiment analysis of physician attitudes about AI and the future

Download full report

4. If you could redesign residency from scratch, what's the first thing you'd change?

Residency was designed in a different era for a different physician workforce, and in many ways, we've layered reforms onto a fundamentally unchanged structure.

I completed my residency in family medicine at UIC and served as chief resident my final year. Family medicine residencies tend to be somewhat more humane than some subspecialties — but even there, the financial stress of training, the expectation of endurance as a marker of competence, and the thin integration between clinical skill and systems knowledge was striking.

The first thing I'd change is the accountability structure. Currently, the burden of managing trainee wellbeing sits largely with the trainee. Programs report wellness metrics, but there's insufficient connection between those metrics and meaningful program consequences. I'd redesign accreditation to require outcomes data — actual longitudinal data on trainee mental health, financial stability, and career satisfaction — tied to continued program approval.

The second thing, closely related: I'd integrate quality, social medicine, health policy, and structural competency into the clinical curriculum. We are training physicians for a healthcare system, not a hospital simulation. The patients I've cared for — at Cook County Jail, at FQHCs, at Cityblock, at IMPaCT Care — exist within systems and zipcodes that directly determine their health. A physician who can diagnose heart failure but cannot navigate a housing instability referral or understand why their patient can't afford the medication they just prescribed is not fully trained for the practice they're about to enter.

5. Independent practice in 2026 — endangered species or quiet comeback?

The independent practice conversation gets framed as survival versus extinction, and I think that's the wrong frame entirely. The more interesting question from my perspective is: what should primary care be — and who's responsible for making sure it exists everywhere?

A piece published in JAMA in May 2026 by Song, Altman, Crichlow, and Grumbach makes an argument I've been waiting for someone to put in print: that primary care should be treated as a public utility, financed through a common fund that pools spending across public and private payers and pays practices directly. The core problem they name is one I've lived in every setting I've worked in — state-level investments in primary care consistently leave out large swaths of the population because insurance fragmentation means no single lever reaches everyone. Medicaid, Medicare, self-insured commercial lives — they all sit in different regulatory buckets, which means even well-intentioned state policy lands unevenly on the ground.

Here's the honest take: I spent years in community health settings where the independence of the practice was treated as a point of pride. And I understand why. FQHCs, in particular, have a mission accountability and community governance structure that distinguishes them from corporate acquisition. That independence matters. But independence without adequate, stable financing is just a slower path to closure.

What I've come to believe — from running clinical operations at Cityblock, from my time at the VA overseeing the largest integrated care network in the country, and now at IMPaCT Care — is that the question isn't integration versus independence. It's whether the financial architecture underneath primary care is designed to sustain it as a common good or to extract value from it as a commodity.

If a common fund model creates stable, adequate payment for primary care practices serving Medicaid patients, uninsured patients, and complex populations in underserved communities — I don't much care whether the practice is independently owned or vertically integrated, as long as mission accountability is real and the community has meaningful voice. What I care about is whether the doors stay open, whether the panel reflects the actual community, and whether the payment is sufficient to do the work right.

The independent practice debate is really a financing debate in disguise. Song and colleagues are right to name it that way.

6. What's something physicians are quietly worried about that the public doesn't know about yet?

The thing physicians are most reluctant to say out loud is that we may not be the right people to speak on behalf of our patients. And the system has been built, for a very long time, as if we are.

I don't mean that cynically. I mean it structurally.

When I was at the VA, overseeing care for millions of veterans, the clinical voice in policy conversations was almost always a physician voice. When I was at Cityblock, building care models for Medicaid patients with complex needs, the people making decisions about what those patients needed were — again, largely clinicians, largely physicians. Well-meaning. Mission-driven. But not the patients themselves.

The quiet worry I hear from the most honest physicians I know is a version of this: we have spent decades building advocacy infrastructure — professional societies, political action committees, scope of practice battles, payment reform coalitions — that is nominally about patients but is functionally about physicians. And we have told ourselves those are the same thing. They are not always the same thing.

The patients I've cared for at Cook County Jail, at Erie Family Health Center, at community health centers across Chicago — they had extraordinarily clear ideas about what they needed. Housing. Safety. Childcare. Time off work to actually attend appointments. Freedom from the particular terror of a medical bill arriving after a vulnerable moment. What they needed was not always what I was trained to provide, and it was almost never what the system was designed to deliver.

The community health worker model at the center of IMPaCT Care's work exists precisely because of this gap. CHWs are members of the communities they serve. They share lived experience with patients. They hear things in a kitchen or on a front porch that never make it into a clinical encounter. They hold a kind of knowledge that no amount of medical training produces — and for years, the healthcare system treated that knowledge as supplementary. Nice to have. Adjacent to the real work.

That inversion — where physician advocacy stands in for patient voice — has real consequences. It shapes what gets funded, what gets researched, what gets built. It explains why we have extraordinarily sophisticated tools for managing acute illness and extraordinarily underdeveloped infrastructure for the social conditions that generate it.

The public doesn't fully know yet how much of what gets called patient advocacy in medicine is actually professional protection. The fix isn't physicians becoming better advocates. It's building systems where patients — particularly patients from communities historically excluded from those rooms — have direct, structural power over the decisions that shape their care.

7. If you had a magic wand for one thing in medicine — not your whole career, just one thing — what would you fix tomorrow?

If I could change one structural feature of American medicine tomorrow, it would be the reimbursement architecture for primary care — specifically its failure to value the work that most directly determines whether underserved patients receive adequate care.

This isn't a novel observation. The primary care workforce crisis is well-documented. The relative value unit system's bias toward procedural specialties is well-documented. What's less discussed is how this structural misalignment cascades into virtually every other major problem we're trying to solve.

The value-based care transition offers a pathway out of this misalignment, and I've seen it work in practice. When you're paid for outcomes — for keeping patients healthy, out of the emergency department, connected to their care team — suddenly the economics of investing in community health workers, in care coordination, in relationship depth, actually pencil out.

But the transition is slow, uneven, and still insufficient for the most complex populations. The magic wand isn't a technology. It's the political will to acknowledge that we have systematically undervalued the work of keeping people healthy — particularly people who are poor, Black, brown, unhoused, or incarcerated — and to pay for it differently.

Until we do that, everything else is optimization within a broken frame.