Dr. Mazin Shikara on How MedFlorida Medical Centers Uses AI to Grow Primary Care Without Burning Out Its Physicians

The founder of MedFlorida Medical Centers says the case for artificial intelligence in primary care is capacity, not cost cutting. The evidence on claim denials and clinician burnout suggests he has a point.

Two numbers frame the argument Dr. Mazin Shikara makes about artificial intelligence in primary care.

The first is 5.9. That is how many hours of an 11.4-hour workday family physicians spent inside the electronic health record in a widely cited time-motion study published in the Annals of Family Medicine, including nearly an hour and a half after clinic closed. Close to half of that EHR time went to clerical and administrative work such as documentation, order entry, billing and coding.

The second is roughly 15%. That is the share of claims private payers initially denied, according to a Premier survey of 516 hospitals. More than half of those denials, 54.3%, were eventually overturned and paid. Providers spent an estimated $19.7 billion a year fighting denials, about $10.6 billion of it on claims that should have been paid the first time.

“Both numbers describe the same problem,” said Shikara, founder, President and CEO of MedFlorida Medical Centers, a multi-location primary care network headquartered in Deerfield Beach, Florida. “Physician time and practice revenue are leaking out through administrative work that doesn’t make anyone healthier. If AI can stop that leak, the practice can grow. That’s the only reason I’m interested in it.”

Capacity, not headcount

Shikara, a board-certified internist who has practiced since 1991 and has led MedFlorida since 2007, said the most common framing of health care AI gets the goal wrong.

“A lot of the industry talks about AI as a way to do the same work with fewer people,” he said. “In primary care, that’s the wrong target. The country does not have enough primary care capacity. The question is whether these tools let the team we already have care for more patients, open more locations and stay in the profession.”

That philosophy shaped MedFlorida’s adoption of AI tools from eClinicalWorks for its revenue cycle. In an interview with Healthcare IT Today, MedFlorida EMR Innovation Administrator Robert DeLuca said the objective was to make billing efficient enough to support growth. The organization now manages a higher volume of claims without adding billing staff, a meaningful advantage in a labor market where experienced revenue cycle professionals are difficult to recruit and retain.

Moving the claim edit to the point of care

For physicians, the most tangible change is where documentation problems get caught.

In a conventional workflow, a note that does not fully support what is billed travels downstream. Billing staff may catch it and route the progress note back to the clinician days later, or the payer may deny the claim and start a rework cycle. Either way, a physician is asked to reconstruct a visit long after the patient has left.

At MedFlorida, the system now alerts clinicians during the encounter when information needed to bill properly may be missing. As DeLuca told Healthcare IT Today, clinicians fix it in the moment.

Shikara said physicians underestimate how much that late rework costs them. “Every note that comes back to you a week later is a small interruption with a large price,” he said. “You have to reopen the chart, remember the conversation and defend your own documentation. Correcting it in the room, while the clinical reasoning is still fresh, produces a better note and a cleaner claim at the same time.”

He added a caution for colleagues: the prompt is a check on completeness, not a substitute for clinical judgment. “The documentation should reflect the medical decision making that actually happened,” he said. “A good prompt reminds you to record what you did. It should never push you to record something you didn’t.”

The burnout evidence is getting stronger

The other half of MedFlorida’s AI strategy is ambient clinical documentation. The organization was an early adopter of Sunoh.ai’s ambient scribe, which drafts visit notes from the clinician-patient conversation for physician review, as described in Sunoh’s announcement of the deployment.

Independent research is now catching up with practice-level experience. A quality improvement study of 263 ambulatory clinicians across six U.S. health systems, published in JAMA Network Open in October 2025, found that burnout fell from 51.9% to 38.8% within 30 days of adopting an ambient AI scribe. Participants also reported lower cognitive task load, less after-hours documentation, more focused attention on patients and an improved ability to fit urgent patients into their schedules.

That last finding is the one Shikara watches most closely. “Access is the whole game in primary care,” he said. “If a doctor has room in the day to see the patient who called this morning with a fever, that patient doesn’t end up in an emergency department tonight. That is where documentation burden turns into a public health issue.”

A recruiting advantage MedFlorida did not plan for

The tools have also changed how MedFlorida competes for clinicians. DeLuca told Healthcare IT Today that clinicians who leave the practice now seek out ambient listening at their next job because they have grown accustomed to it.

Shikara sees that as a signal for every independent and community practice. “Physicians compare notes with each other constantly,” he said. “Once a doctor has finished charts before dinner for a year, a job offer without that workflow looks like a step backward. Practices that treat these tools as optional are going to lose candidates without ever knowing why.”

The test MedFlorida applies

After three decades in practice, Shikara said he has watched many technologies promise to rescue primary care and fail to deliver. His filter for new tools is deliberately simple, and he applies it to AI the same way.

“We ask three questions,” he said. “Does it give physicians time back with patients? Does it make the record more accurate, not just faster? And does it let us take care of more people with the team we have? If the answer to all three is yes, we scale it. If not, we stop.”

By that measure, MedFlorida’s AI tools have earned a permanent place in how the practice runs. The larger question, in Shikara’s view, is whether the rest of primary care will judge them by capacity and quality rather than by how many positions they eliminate. On that choice, he said, depends whether the next decade of health care AI expands access to a primary care doctor or quietly shrinks it.