In a clinically integrated network where nearly 70% of primary care physicians are independent, leaders cannot rely on a traditional management structure to drive change. Physician governance and trusted data become essential to engagement and to whether AI-powered insights earn clinicians’ confidence.
In a featured session sponsored by Arcadia and Navvis at Becker’s 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health on Sept. 15, three leaders described how Tufts Medicine built and runs its network.
The panelists were:
- Lisa Mann, MBA, vice president of population health management and optimization at Tufts Medicine Integrated Network
- Luke Hansen, MD, MHS, chief medical officer at Arcadia
- Sari Miettinen, MD, market president and physician executive at Navvis
Below are four takeaways from their conversation.
1. Physician governance is essential
Ms. Mann said Tufts’ network spans more than 3,000 clinicians, including roughly 600 primary care physicians. Close to half are independent and about 280,000 patients are in risk-based contracts.
Dr. Miettinen said engagement had to be designed deliberately. Physicians sit on the board, its performance and quality subcommittee and the network physician leadership council, supported by formal change management.
“I think it’s that combination of physician governance, data-driven performance improvement and change management that has really given us fabulous engagement,” Dr. Miettinen said.
2. Why trust matters
Dr. Hansen said the data foundation requires normalization, accuracy and timeliness; relying on the claims cycle alone is too slow for proactive action.
“The most precious commodity in the adoption of these changes is trust, and you really can’t rush that,” he said.
Dr. Miettinen also tied trust back to change management. When physicians challenge performance data, her teams document every concern, cross out what the physicians cannot control and focus on the one item they can influence.
3. How AI is earning its keep
Dr. Hansen said AI has delivered value through pattern detection and monitoring. Payers change claim-file specifications, data breaks and continuously monitoring models can catch the resulting disruptions.
AI models today are non-deterministic, he cautioned, and Arcadia’s heaviest users treat the technology as an accelerator for analysts and clinicians, not a replacement.
Ms. Mann cited ambient documentation, prior authorization and coding support as current use cases, with data reconciliation and patient attribution among the next opportunities.
4. Adoption depends on implementation
Ms. Mann recalled a physician saying an ambient documentation rollout felt like being taught to swim while drowning. The experience reinforced the need for teams to work directly with clinicians throughout implementation.
Dr. Miettinen described a similar approach on the analytics side. Tufts Medicine’s insights teams pair an analyst, a performance improvement leader and an operator within each performance pod to interpret reports in the context of local markets, workflows and relationships.
The speakers acknowledged that AI governance remains a work in progress, particularly across independent practices. Dr. Miettinen said physicians’ interest in feeding reports into consumer AI tools illustrates how quickly demand is moving, while Ms. Mann noted that system-level policies do not necessarily extend across the entire community network. Their advice was to connect each implementation to a clearly defined outcome. “If you cannot define what success looks like, we don’t kick off,” Dr. Miettinen said.
The post Inside Tufts Medicine’s physician-led approach to data and AI appeared first on Becker's Hospital Review | Healthcare News & Analysis.