Send us Fan Mail
Clinical knowledge management decides whether clinical AI is safe at the bedside, and Dr. Declan Kelly explains why a hospital's own governed knowledge matters as much as the model.
Dr. Declan Kelly is a physician, software engineer, and founder and CEO of Eolas Medical, whose platform uses AI to index a hospital's own guidelines, policies and medicines information so clinicians get answers in seconds. He practiced for about five years, mostly in emergency medicine, before building Eolas, which is now live in roughly 500 sites, including most acute NHS trusts and U.S. hospitals such as Stanford and Mass General. He joins Chris Hutchins, Founder and CEO of Hutchins Data Strategy Consultants, to talk about the knowledge layer that clinical AI keeps skipping.
What We Cover Declan explains why external evidence gets a clinician only part of the way, while the last mile of a decision depends on local knowledge such as how to book a scan overnight or which drug sits on the formulary. By his estimate only about 10% of that knowledge is written down, and the rest lives in people's heads. Chris and Declan discuss why trust in clinical AI depends on context, transparency and the ability of a clinician to check where an answer came from. They also cover regulation in the U.S. and Europe, and how a physician's mindset has to change to build a company.
Key Takeaways Clinical AI should be judged like any other clinical tool, with safety nets built around an imperfect test, much as a troponin result is read alongside other checks and clinical judgment. Most health systems will not need to train their own models, since retrieval built on their governed data supplies the context that makes answers safe. Managing that knowledge is harder than accessing it, because ownership spans clinical, operational, legal and governance teams. Declan argues that healthcare needs technology to lower costs and widen access, with clinical safety as the condition for that progress.
Frameworks and Tools The conversation covers retrieval augmented generation versus fine-tuning, capture of tacit clinical knowledge, and connecting a knowledge layer to ambient listeners, the EHR and general AI assistants. Declan describes how Eolas has a separate answer engine review every question for clinical safety. He also notes that some European regulators classify ambient listeners as medical devices.
About Dr. Declan Kelly Dr. Declan Kelly trained as a medical doctor and worked clinically for about five years, mostly in emergency medicine, while also building software for his own hospital. He founded Eolas Medical, a Belfast-based healthtech company, after living the problem of reaching approved local guidance at the point of care. Eolas announced a $12 million Series A in January 2026.
Related Resources Episode page: https://signalroompodcast.com/episodes/clinical-ai-hospital-knowledge
Dr. Declan Kelly on LinkedIn: https://www.linkedin.com/in/declan-kelly-484bb7187/
Eolas Medical: https://www.eolasmedical.com/
Beneath the Signal, Chris's book on the human work behind trusted data and responsible AI in healthcare: Get it on Amazon
The Signal Room is featured in Feedspot's Best AI in Healthcare Podcasts, 2026: https://podcast.feedspot.com/ai_health_podcasts/
Support the show
About The Signal Room: The Signal Room is a podcast and communications platform exploring leadership, ethics, and innovation in healthcare and artificial intelligence. Hosted by Christopher Hutchins, Founder and CEO of Hutchins Data Strategy Consultants. Leadership, ethics, and innovation, amplified.
Website: https://www.hutchinsdatastrategy.com
LinkedIn: https://www.linkedin.com/in/chutchins-healthcare/
YouTube: https://www.youtube.com/@ChrisHutchinsAi
Book Chris to speak: https://www.chrisjhutchins.com
45 分