DiscoverPractical AI in Healthcare
Practical AI in Healthcare
Claim Ownership

Practical AI in Healthcare

Author: Steven Labkoff, MD and Leon Rozenblit, JD, PhD

Subscribed: 10Played: 49
Share

Description

AI promises to transform healthcare—but real, scalable impact remains rare. Practical AI in Healthcare cuts through the noise to showcase real-world use cases delivering business value today. Hosted by senior leaders— former VPs of life science technology groups, clinical informatics professionals from top-tier organizations, and a former Big Four consultant—each episode features candid conversations with the people making AI work inside the healthcare enterprise
64 Episodes
Reverse
When better information reaches a physician, what actually changes—and how would we know it helped?Emergency physician Brian Fengler joins Steven Labkoff and Leon Rozenblit to explore the distance between collecting healthcare data and using it during care. The discussion examines why feedback arriving months after a decision has different consequences from information available in the moment, and how peer comparisons can influence clinical behavior.The conversation also turns to a harder question: when performance improves, how much credit belongs to software, training, staffing, or the work of implementation? This episode of Practical AI in Healthcare considers what useful AI contributes inside an existing workflow, what a better number can leave unexplained, and why physician trust remains part of the problem.
John Halamka, president of Mayo Clinic Platform, joins Steve Labkoff and Leon Rozenblit to ask what large health datasets can answer and where they run out. A registry will record a patient's toenail length; a medical record won't. John makes the case for testing ideas in real-world data before spending on a trial, and Leon presses him on where randomization still has to happen. They also get into governing thousands of AI skills, why Steve hears an echo of MYCIN, and why John thinks people trust large language models too much.
Adam Blum returns to talk about what sits underneath his clinical-trial matcher: PRomop, an open-source patient record built on OMOP. Now CTO of HealthKey.ai, which acquired CancerBot this year, Adam argues that organizations can replace their transactional, analytics, and decision-ready stores with one, and that a single flat record per disease is enough. Leon Rozenblit pushes back on whether one record can serve many questions, and Steve Labkoff presses on staleness and where the AI actually lives. Steve advises HealthKey and co-authored the PRomop paper with Adam.
In 1973, Charlie Safran wrote programs for an MIT group that expected a computer doctor by the year 2000. Fifty years later, the former chief of clinical informatics at Beth Israel Deaconess and past AMIA chair walks Steve Labkoff and Leon Rozenblit through what changed and what didn't: Bayesian decision models for Hodgkin's disease, rules of care in early electronic health records, why those rules were so hard to maintain, and why clinical data will always reflect what gets paid for. His conclusion: reasoning turned out to be the easier part, and technology should make care more human, not less.
Six conversations into Season 2, Steve Labkoff and Leon Rozenblit look back at what they added up to. They revisit Raihan Faroqui on AI agents at the front desk, Greg Raiz on data over gut feel in pre-seed investing, Randi Seigel and Jared Augenstein on the state-by-state map of AI law, Travis Zack on an evidence engine that grades its own sources, Sir Peter Donnelly on genetics as nature's drug trial, and Latha Valluripalli Soorya and Erica Crawford on smart glasses in the autism evaluation. Their read: the field is moving from hype into implementation, efficiency keeps arriving before accuracy, and "AI" may be too broad a word to regulate or even to argue about.
loading
Comments