What’s Trending
Artificial intelligence (AI) is increasingly being incorporated into clinical workflows, raising the stakes for how health systems integrate and govern these tools. The central question is whether AI can be embedded in ways that meaningfully improve care delivery while maintaining trust.
Recent activity is bringing AI closer to clinical decision-making and the electronic health record (EHR). The latest example is Memorial Sloan Kettering Cancer Center’s (MSK’s) recently announced partnership with OpenEvidence. The collaboration will integrate the clinical AI platform into MSK’s Epic workflow and pair patient context with cancer genomics expertise and medical evidence at the point of care. MSK’s OncoKB precision oncology knowledge base will also become more broadly available through the partnership.
The announcement follows an earlier partnership between Cedars-Sinai and OpenEvidence that gives clinicians enterprise access to the platform within the EHR. Clinicians can query medical literature in the context of an individual patient’s record and longitudinal health data. Cedars-Sinai also plans to incorporate its own care pathways and protocols into the platform.
These partnerships are part of a broader effort to connect AI more directly to the clinical environment. Earlier this month, OpenAI announced an Epic connection for ChatGPT for Healthcare, which gives authorized clinicians read-only access to patient information. Epic continues to expand its own AI capabilities as well.
Interoperability will become increasingly important as AI vendors seek deeper access to health system data and workflows. The proposed federal HTI-5 rule would strengthen access for third-party developers and extend information-blocking protections to automated technologies, including AI. The rule has not yet been finalized, and recent reporting suggests some of those provisions may be scaled back. That leaves open important questions about the final impact of the rule and how easily AI platforms will be able to integrate with incumbent EHRs.
Clinician and consumer demand is helping push this activity forward. More than 80% of physicians now report using AI professionally, more than twice the share in 2023, although concerns about privacy and validation remain significant. Physician-facing AI is also evolving beyond discrete tools for documentation or research support. Newer capabilities are beginning to combine patient context, clinical evidence, and workflow support in a more integrated way. That shift is visible in platforms moving from ambient documentation into visit preparation and decision support, as well as in partnerships like MSK’s work with OpenEvidence.
Consumer adoption is growing as well. New research found that 41% of consumers now incorporate AI and/or social media into their non-emergency healthcare journeys. At the same time, 85% of consumers who use AI and/or social media for health information verify the information they receive before acting on it, reinforcing that growing use has not eliminated concerns around trust.
Epic has announced a 6-week pause in product development to focus on cybersecurity. The move comes as AI capabilities are expanding rapidly across healthcare, highlighting the need for guardrails and oversight to keep pace with that growth. Health systems are facing the same tension: AI adoption is accelerating while the governance, technology infrastructure, and operating models needed to support broader use are still taking shape.
Why it matters
As AI becomes more integrated into clinical decision-making and day-to-day physician work, health systems are feeling growing pressure from both clinicians and consumers to engage. A prolonged wait-and-see posture is becoming increasingly difficult. Health systems will need to respond in ways that improve care delivery and fit the organization’s broader strategy while building the trust and oversight needed for broader use.
That pressure is increasingly evident in how patients and clinicians are using the technology. Patients are turning to AI to seek and interpret health information. Clinicians are looking for tools that reduce administrative burden and make it easier to practice medicine. The real test is whether AI can be integrated in ways that materially improve care delivery and fit naturally into existing workflows, rather than simply adding another layer of technology.
The current wave of pilots and partnerships is an important testing ground, particularly as physician-facing AI expands beyond documentation into broader workflow and decision support. Results are likely to vary, depending on how clearly the technology is tied to a specific problem and how well it fits into the surrounding workflow.
These efforts can help health systems understand where AI can expand, where workflows may need to change around it, and where the technology may not yet add enough value. Those lessons should inform broader decisions about workforce, technology investment, operating priorities, and the role AI plays in the organization’s broader strategy.
As use expands, trust will become even more important. Health systems will need governance, guardrails, and validation models that keep pace with the technology. Patients, physicians, and communities will need confidence that clear boundaries are in place around where AI can act and where human judgment remains essential. They will also need confidence that performance is being actively monitored as capabilities evolve.
The technology environment in which AI operates will shape health systems’ decisions. Epic’s embedded position gives it a structural advantage, while healthcare-focused and horizontal AI platforms may offer capabilities that extend beyond the EHR. The value of such platforms will depend in part on how deeply they can connect to the existing technology environment. Interoperability will therefore be a key consideration for health systems as they assess where these platforms can scale.
Waiting for the technology, regulatory environment, and governance model to fully settle is becoming less realistic. Health systems will need to move deliberately while staying responsive to what clinicians and patients are already asking for. The strongest approaches will connect AI decisions to enterprise strategy and build the trust and infrastructure needed to expand responsibly.
Related Links
Fierce Healthcare:
OpenEvidence deepens oncology push, launches new AI model family
Becker’s:
The new AI deal-breaker for health systems
STAT:
STAT Health Tech: OpenEvidence launches new family of AI models for clinicians
Becker’s Health IT:
25 health systems build AI agents with Epic
