Boston Children's and OpenEvidence will study how doctors' AI use shapes care

A study of the tool's effects

Boston Children's Hospital and OpenEvidence, a platform doctors use to look up medical evidence, announced a multi-year collaboration on July 16, 2026. As openevidence.com reported, the aim is to examine how clinicians' real-world use of AI tools during patient care relates to the way they practice and deliver care.

That is a different kind of project from most AI health news. Instead of asking whether a chatbot gives a correct answer, the partners want to look at the bigger picture: once a tool becomes part of daily work, does the pattern of care change? "Practice patterns" simply means how doctors actually treat patients in routine settings, such as which approaches they choose and how often.

OpenEvidence says its adoption in clinics has been broad, but that little is known about how that adoption connects to practice across whole populations. The announcement claims the partnership can study this at a scale and level of detail not possible before. That is the partners' own framing, and no findings have been shared.

What data is used, and what is not

Privacy is the obvious concern, and the announcement addresses it directly. The teams say they will analyze only aggregated, provider-level data. Aggregated means combined rather than viewed one case at a time, and provider-level means the data is about clinicians rather than individual patients. The release states that no patient-level data will be accessed or used.

John Brownstein, Chief Innovation Officer at Boston Children's, described the approach as studying anonymous patterns in how clinicians ask and answer questions during visits. He said this could bring out "insights into health trends that would otherwise remain hidden."

In plain terms, the questions doctors put to a tool might work as a signal. If many clinicians start asking about the same condition or treatment, that could point to something changing in the community or in how care is given. The announcement does not describe the methods, so treat this as our reading of the idea, not a confirmed procedure.

A direct line from the health record

The collaboration also has a practical side. It includes an integration that lets Boston Children's clinicians open OpenEvidence directly inside the electronic health record, the system where patient charts and orders are kept. The release presents this as a sign of how central the tool has become to the hospital's workflows.

Mondira Ray, Senior Vice President of Clinical Informatics at OpenEvidence, is a pediatrician who trained and practiced at Boston Children's before joining the company. She said the tool has become part of how clinicians practice, and that a change this visible has implications the company wants to examine.

Easier access tends to mean more use, which makes the question of effects more pressing. If a tool sits one click away from the patient chart, its influence on everyday decisions may grow. Whether that influence is helpful is exactly what the project says it wants to explore.

The two partners

Boston Children's describes itself as ranked among the best children's hospitals in the nation by U.S. News & World Report and as a teaching affiliate of Harvard Medical School. It began in 1869 as a 20-bed hospital and is now a 491-bed center, with a research community of 3,000 researchers and scientific staff. It also runs a primary care network of 33 pediatric practices across Massachusetts.

OpenEvidence calls itself the most popular AI for doctors in the United States. It says independent studies show more American physicians use it than all other AI platforms combined, and that the majority of U.S. physicians use it daily on average. It also says its answers are sourced and cited from peer-reviewed medical literature, and that it is free for verified U.S. clinicians. These are company claims, made in the company's own announcement.

What we do not know yet

The announcement is a statement of intent, not a result. It does not say which practice patterns will be examined, what would count as better or worse care, or when any findings might appear. It also does not say whether the results will be published.

One structural point is worth keeping in mind. The study pairs a hospital with the company whose product is being studied. The release does not describe any outside oversight, so readers should watch for how the work is reported and whether independent researchers can check it.

Brownstein framed the project as a way to help clinicians deliver better care while protecting privacy. Whether it does depends on the findings, which do not exist yet. For now this is a project to watch, not a proven benefit.