I’m Mike Dulin. I built and led the analytics organization at Atrium Health—work Harvard Business School turned into a case study—and then went back to seeing patients. Now I help health systems and health-tech teams decide which AI is worth adopting, prove that it works, and get it into the workflow.
Every engagement starts with a conversation and a short diagnostic. No slide decks about the future of AI—just the decision in front of you.
Which tools deserve a pilot, how to test them against your own data and workflows, and what governance has to be in place before go-live.
Physician judgment inside the model lifecycle: the right outcome to predict, the data that can and can’t support it, and the evidence a health-system buyer will ask for before go-live.
Dashboards and workflows that show a team what needs attention today—referrals, value-based-care gaps, quality measures—rather than a report nobody opens.
As Chief Clinical Officer for Analytics I stood up the center of excellence for data science and outcomes research at one of the largest health systems in the Southeast—hiring more than 60 data scientists and embedding analytics in clinical operations and population health. Harvard Business School published the work as a teaching case.
Read the HBS case (Carolinas HealthCare System: Consumer Analytics) ↗A referral only counts if it leads to care. Working alongside the referral coordinators, I designed the dashboards that show where referrals stall and led a process-redesign workshop that turned those findings into an agreed future-state workflow.
Talk about your analytics challenge ↗Product strategy, clinical validation, and client relationships for a value-based-care analytics company—clinical judgment inside the product cycle, not bolted on at the end.
Faculty for the NAM / RWJF Health Policy Fellows program on practical, hands-on use of AI—including a course on using AI to design your own work and life.
See the course ↗Health records are scattered across portals, PDFs, and memory. Folio brings them into one story, keeps every source attached, and helps a person walk into their next visit prepared. I started it after years of stitching together my own records as a patient—and watching my patients do the same.
Looking for a few concierge and direct-primary-care practices to pilot with. If you run one and want your patients to arrive with a clearer picture, I’d like to talk.
A weekly note on what AI actually does in a primary care visit, what I’m learning building Folio in public, and the occasional reflective piece.
I speak on clinical AI validation, population health analytics, and what changes when a physician is also the patient.
Evidence, local validation, and the gap between an impressive model and a useful clinical tool.
Disease-burden evaluation, predictive analytics, and what health systems can do now.
Population health analytics and primary care.
Panel discussion at the HIMSS 2022 Global Conference.
On genomics and precision medicine from both sides—as a physician and as a patient.
Short bio. Michael Dulin, MD, PhD, is a family physician and healthcare AI advisor. He built Atrium Health’s analytics organization—the subject of a Harvard Business School case—served as a Robert Wood Johnson Foundation Health Policy Fellow, and teaches AI for the National Academy of Medicine. He is the founder of Folio.
I trained as a family physician and a scientist (MD, PhD, UTHealth Houston) and spent the first stretch of my career in community health research—more than $10 million in NIH and AHRQ funding—before running analytics for a large health system. I spent eight years as a professor of health management and policy at UNC Charlotte, working through its Academy for Population Health Innovation, and two as a Robert Wood Johnson Foundation Health Policy Fellow in Washington.
In 2024 I went back to full-time primary care. That turned out to be the most useful thing I could have done for my AI work: I now judge every tool by whether it survives a 20-minute visit. I live in Greensboro, North Carolina.
An AI tool to evaluate, an analytics team to build, a product that needs a physician in the room, or a practice that wants to pilot Folio. Thirty minutes is enough to know whether I can help.
Consulting, model validation, speaking, and Folio pilots.