Webinar–From Cleared to Clinical: Making AI Work at the Bedside

A health system can do everything right and still watch the rollout stall, because even a cleared tool with a clinical champion behind it has to fit how clinicians work and prove itself on the patients it will actually see. The systems that get past that are pulling ahead, cutting backlogs and catching disease earlier, turning one-off pilots into a platform they can scale. The gap between those two outcomes is what this hour is about.

The regulatory path runs through all of it. Clearing a tool is hard, as it should be, but the framework is also maturing in ways that can speed good products to the bedside and even let a model keep improving after approval. How it develops will shape what gets built and funded, and how fast any of it reaches patients, especially across academic medical centers and integrated delivery networks where the stakeholder count only climbs.

Our panel has worked these questions from different angles. Dr. Robert Califf ran the FDA twice and led medical strategy at Verily and Google Health between terms, and he’s blunt that cleared models drift in use and that no agency can recheck them at scale, which leaves the monitoring to the systems running them. Dr. Caroline Chung, MD Anderson’s first Chief Data and Analytics Officer, argues the real investment is the data beneath the models, and that the discipline is asking whether AI is the right answer to a defined problem, not switching it on because it’s there. Dr. Tessa Cook built Penn Medicine’s radiology AI governance over nearly seven years, across eight hospitals and 2.5 million imaging exams a year, and her lesson is that deployment is mostly people and process, and that a model no one watches is one no one should trust.

‍