Inspiring symposium on unlocking the potential of AI in clinical care. Grateful to be a part of our pioneering institution at this pivotal inflection point.
wolterskluwer.com/en/news/up…
Well it’s about time. UpToDate now featuring built-in genAI CDS. Will be interesting to see how the competition between OpenEvidence (free, AI native, built on primary source literature) and UpToDate (costly, integrated, curated expert review) unfolds.
How good is AI at making medical diagnoses? Good enough to help—and to hurt.
For @NewYorker, I spent months talking to patients, doctors, and researchers about how to make the most of a powerful new technology and how to minimize the side effects:
newyorker.com/magazine/2025/…
Presented at #ADASciSessions:
Zimislecel is an allogeneic stem cell–derived islet-cell therapy. This phase 1–2 study supports the hypothesis that zimislecel can restore physiologic islet function and thus treat persons with type 1 diabetes. Full study: nej.md/3ZGupbv
arxiv.org/abs/2412.10849
o1 outperforms 2 IM docs in naming top diagnoses when given access to retrospective EHR data on 79 real ER cases. Physician 2 shall remain unnamed. A revolution in high-stakes diagnostics is underway.
A great example of what's possible when someone connects their passion with their purpose to make a real difference.
DxGPT, a rare disease diagnostic tool, is now live, showing how AI can be applied to help truly improve lives.
Original Article by A. Bhargava et al.: FDA-Authorized AI/ML Tool for Sepsis Prediction: Development and Validation nejm.ai/49dV42V#AIinMedicine
ALT Figure 3. Sepsis ImmunoScore Stratification for Sepsis-3 in All Cohorts.
Bar charts are shown for the derivation, internal validation, and external validation cohort datasets for the Sepsis-3 criteria within 24 hours predictive values for each Sepsis ImmunoScore risk stratification category. Dashed lines indicate the 95% confidence intervals.
Healthcare is personal, political, and incredibly complex. In the midst of recent tragedy, it’s worth taking time to understand how the system evolved, where it succeeds and fails, and how tradeoffs shape the challenges we face.
open.spotify.com/episode/3tN…
Can AI help find diagnostic errors? This team found "missed opportunities in diagnosis" (MOD) of abdominal pain at 30% PPV with a rules-based approach and 93% with machine learning of structured data. What would LLM's add? ⤵️
jamanetwork.com/journals/jam…
Some of medicine is complicated and will be helped along by computers and math. But anyone who practices in the real world knows that much of healthcare is actually complex, not complicated. Knowing the difference is vital.
In honor of #worldsepsisday a reminder that physician judgement is superior to many classically derived decision tools that have littered our workflows, EHRs. And why integrated outcomes-tested support tools (COMPOSER, TREWS etc) are desperately needed.
annemergmed.com/article/S019…