Johns Hopkins Medicine and OpenEvidence announced a partnership on October 13 to rebuild both how clinicians are taught and how evidence reaches the bedside.
The collaboration has two pillars. One is a medical education model that leans on pathophysiology, evidence-based medicine and clinical reasoning from first principles. The other is a set of adaptive AI systems that pair current guidelines with a patient’s own history and Hopkins practice patterns.
Hopkins frames the deal as a return to its own history. The institution built the residency model that shaped 20th-century American medical training, and its education leadership argues AI creates room to push emphasis back toward the bedside rather than away from it.
“Physicians learn medicine at the bedside, grounded in science and reasoning,” said Sanjay Desai, vice dean for education. “The arrival of AI, in many ways, invites us back to the bedside.”
Travis Zack, OpenEvidence’s chief medical officer, framed the open question as who sets AI’s role in medicine, and on what principles, rather than whether the role changes at all.
The deal fits a widening pattern. OpenEvidence signed Penn Medicine on September 15 to reach more than 10,000 clinicians across a seven-hospital system, with decision support extended into low- and middle-income countries through the Botswana-UPenn Partnership. Memorial Sloan Kettering followed on September 16, bringing its OncoKB precision oncology knowledge base inside OpenEvidence.
For health system boards, the distinguishing feature is the education half. Most vendor deals sell retrieval. Hopkins is buying a redesign of how its trainees learn to reason with retrieval already assumed.
OpenEvidence has said its assistant is the most widely adopted medical AI in the U.S. The Hopkins arrangement tests whether that adoption can be steered toward training, not only toward answers.
