Before Margaret Lozovatsky, MD, took over as chief digital information officer of Dayton, Ohio-based Premier Health earlier this year, she raised something with her CEO before she agreed to the role: She wanted to keep practicing medicine.
Dr. Lozovatsky is a pediatrician who now spends a couple of shifts a month caring for newborns at Premier, on top of running the health system’s technology strategy.
That arrangement is still unusual in health IT leadership. But as AI moves deeper into clinical workflows, a handful of physician technology executives across the country say staying in active practice — not just holding a medical degree — has become a genuine strategic advantage rather than a resume line.
Dr. Lozovatsky said the case for it came into focus almost immediately. She recently completed her first credentialed shift at Premier since taking the CDIO role and got to see the health system’s AI summarization tools in action herself.
“It really summarizes the chart for you,” she told Becker’s at its 11th Annual IT + Revenue Cycle Conference on Sept. 16 in Chicago. “The hours we spent digging through the information — it’s been really good. We have shown more than a full day’s worth of time saved in a month.”
That kind of firsthand test, she said, is what’s missing when technology leaders evaluate AI purely from a boardroom. “It has been a game changer for me to be able to see the tools used in practice.”
At Philadelphia-based Penn Medicine, Mitchell Schnall, MD, PhD, senior vice president for data and technology solutions, has continued to practice radiology since assuming the role in 2024. He said that schedule changes how clinicians receive him when he asks them to adopt new AI tools.
“As someone who one day a week is shoulder to shoulder with them practicing medicine … it brings a different message than if it’s someone who doesn’t have that experience,” he said.
Dr. Schnall pointed to a comment he heard from a health tech executive on a Becker’s panel discussion about the need for a “translator” between technology and clinical work — someone experienced enough in the actual job to bridge the two. Penn Medicine, he said, is now deliberately creating that kind of role.
“These tools are so complex and can be so intimately related to how we do our work that you really need people actually experienced in doing the work as part of the development team,” Dr. Schnall said. Penn is developing a “class of tech-savvy, tech-empowered clinicians to be that person … on the team that goes between the technology and … the clinical programs,” he said.
His own clinical practice also shapes what he’s trying to avoid. Dr. Schnall said Penn’s EHR rollout years ago quietly asked physicians to document the same information multiple times over — once for the patient’s chart, then again in structured fields to feed revenue cycle, quality reporting and other back-end systems.
“Every time we’re building something, I say we can’t make this mistake again,” Dr. Schnall said.
At Ann Arbor-based Michigan Medicine, Andrew Rosenberg, MD, who stepped down as CIO in March and now advises the health system’s new chief digital and information officer while teaching at the University of Michigan, said his own clinical and research background still shapes the kinds of questions he pushes the industry to ask about AI. He previously ran a large surgical intensive care unit, where his research focused on why patients ended up back in the ICU when they shouldn’t have.
“It was the unexpected predicting the unexpected,” Dr. Rosenberg said at the conference, describing what drew him toward technology in the first place. Predicting a readmission everyone already expects, he said, isn’t the hard part.
That instinct, he said, is what the industry needs to apply to AI adoption instead of defaulting to the same back-office use cases every health system already has in motion.
“What use cases in direct individual patient care do we think are likely to be good targets that are safe, and where the automation will do better than humans?” Dr. Rosenberg said.
