- The Filter for AI Healthcare Speakers
- Twelve Speakers Worth the Slot
- Topic-to-Speaker Match
- Red Flags That Should End the Conversation
- Format Choices That Land for AI Topics
- The Question Set That Sorts Strong Speakers From Weak Ones
- How to Brief an AI Speaker on Your Specific Stack
- What Comes After the Keynote
- The Bottom Line
The AI in healthcare keynote market has the worst signal-to-noise ratio of any topic on the speaker circuit right now. Half the speakers being marketed as “AI healthcare experts” are repurposed digital-transformation consultants with three new slides. Your audience of clinical informaticists, system CIOs, and chief medical information officers will spot it inside a deck.
This is a filtered list. Twelve speakers who actually deploy AI in clinical research or practice, plus the filter we use to keep them on the list.
The Filter for AI Healthcare Speakers
Three signals separate a real AI healthcare speaker from a slide-deck specialist. First, clinical or research deployment credibility, meaning they have actually built, validated, or used an AI system in a regulated clinical context. Second, fluency with the constraints, HIPAA, FDA clearance pathways, EHR integration, payor coverage decisions. Third, a track record at clinical or research audiences specifically, not just at corporate-tech conferences.
One additional disqualifier worth saying out loud. If the speaker is on the cap table of a startup they cite from the stage, that is a vendor pitch in a keynote outfit. Your clinical-informatics audience will spot it inside the first three slides and the room is gone. Ask the disclosure question during scoping.
Twelve Speakers Worth the Slot
1. Dr. Eric Topol. Founder of Scripps Research Translational Institute, cardiologist, author of Deep Medicine. The institutional voice on AI in clinical medicine. Topol is the rare speaker who is taken seriously by both the AI-research community and senior clinicians.
2. Dr. Daniel Kraft. Founder of Exponential Medicine, Stanford and Harvard trained, practicing physician. Kraft has been working at the intersection of AI, robotics, and clinical medicine for over a decade. Strong fit for innovation summits and health-system strategy retreats.
3. Dr. Karen DeSalvo. Chief Health Officer at Google, former National Coordinator for Health IT. Speaks at the intersection of population health, AI deployment, and platform-scale technology. The clearest voice on what AI looks like at planetary scale.
4. Dr. Bob Wachter. Chair of medicine at UCSF, author of The Digital Doctor. Wachter is the operational voice on the gap between AI promise and AI deployment in real clinical workflow. Strong fit for hospitalist medicine, IT governance, and CMIO audiences.
5. Dr. Atul Butte. Director of the Bakar Computational Health Sciences Institute at UCSF, founder of multiple AI-driven biotech companies. Butte is the speaker to book when your audience wants the data-science side of clinical AI explained without dilution.
6. Dr. Isaac Kohane. Chair of biomedical informatics at Harvard, co-author of The AI Revolution in Medicine. Kohane has been deploying AI in clinical research since before “AI in healthcare” was a panel topic. Authoritative on benchmarking, validation, and the limits of large language models in clinical practice.
7. Dr. Suchi Saria. Director of the Machine Learning and Healthcare Lab at Johns Hopkins, founder of Bayesian Health. Saria has built AI tools that are deployed in real ICUs. Strong fit for clinical informatics and quality-improvement audiences.
8. Dr. Pearse Keane. Consultant ophthalmologist at Moorfields Eye Hospital, lead on the Moorfields and DeepMind AI partnership. Keane is one of the few clinicians who can speak to a regulatory-grade AI deployment in a real clinical service line.
9. Dr. Nigam Shah. Chief Data Scientist at Stanford Health Care. Shah is one of the most cited researchers on AI deployment, validation, and the operational economics of clinical AI. Best for academic medical center and research-leadership audiences.
10. Dr. John Halamka. President of the Mayo Clinic Platform, former CIO of Beth Israel. Halamka is an operator who has actually run an EHR, governed AI, and built a multi-institution data platform. Speaks the language of CIOs and hospital boards.
11. Dr. Ziad Obermeyer. Acting Associate Professor at UC Berkeley, emergency physician. Obermeyer’s research on bias in clinical AI is the canonical reference. Essential for any program addressing AI fairness and equity.
12. Dr. Daniela Rus. Director of MIT CSAIL, robotics and AI researcher. Rus is the right pick when your audience wants the foundational AI research voice rather than the clinical-deployment voice.
Topic-to-Speaker Match
| Audience Topic | Best-Fit Speakers | Format Recommendation |
|---|---|---|
| AI in clinical workflow | Wachter, Halamka, Saria | Keynote plus moderated panel |
| AI in imaging and diagnostics | Topol, Keane, Butte | Main-stage keynote |
| Validation, bias, equity | Obermeyer, Shah, Kohane | Closed-door briefing |
| Strategic AI roadmap | Topol, Kraft, DeSalvo | Fireside chat with CEO |
| Foundational AI research | Rus, Kohane, Butte | Workshop or master class |
| Population health AI | DeSalvo, Halamka, Saria | Plenary plus breakout |
Red Flags That Should End the Conversation
- The speaker cannot name a specific deployed model. Vague references to “AI tools” without a specific model, validation cohort, or institution are a tell.
- Equity and bias get a single slide. Any AI healthcare keynote that addresses bias as a footnote is missing the dominant question your audience has.
- The slides are full of stock futurism. Self-driving car analogies, sci-fi imagery, and “the doctor of the future” decks indicate a speaker who has not been in a clinic in years.
- Vendor disclosure is hand-waved. If you cannot get a clean disclosure on cap-table positions and consulting relationships, the room will spot it for you.
Format Choices That Land for AI Topics
For HIMSS-class events, the strongest format is a 40 to 50 minute main-stage keynote followed by a 30-minute moderated panel with two clinicians from your own system. The panel is where the AI speaker has to defend the talk against operators who actually deploy the technology, and that defence is where the room learns the most. For payor and policy summits, a closed-door briefing is more candid than a public keynote. For clinical innovation summits, a 90-minute workshop with case-based examples beats a one-way keynote.
The Question Set That Sorts Strong Speakers From Weak Ones
During the scoping call, run the speaker through these five questions. Strong speakers answer all five with specifics inside two minutes. Weak speakers will deflect or generalize on at least three.
- Name a clinical AI tool currently deployed in your institution that you would not recommend, and explain why. A speaker who cannot name one is either too credentialed or too commercial.
- Which AI deployment in the last 24 months has surprised you the most, in either direction? The answer should be specific, recent, and slightly uncomfortable.
- What is your read on the FDA pathway for adaptive AI models? A non-answer here disqualifies the speaker from regulatory or governance audiences.
- How do you talk about bias and equity in this audience? A speaker without a specific approach will produce the single-slide bias treatment that loses the room.
- What disclosures should I share with the audience before you take the stage? The speaker should volunteer commercial relationships without prompting.
How to Brief an AI Speaker on Your Specific Stack
Generic AI keynotes age fast. The fix is a specific-stack briefing memo, two pages, delivered seven to ten days before the event. The memo should cover three things. First, the AI tools currently deployed in your system, by clinical service line, with deployment status (pilot, scaled, retired). Second, your governance posture, who chairs AI governance, what the validation and monitoring cadence is, and what the system’s stance is on adaptive models. Third, the two questions your CMIO and CNIO most want addressed on stage. Speakers who use that briefing produce keynotes that feel custom-built. Speakers who ignore it produce decks that could have been delivered at any conference in any city.
What Comes After the Keynote
The strongest AI in healthcare programs treat the keynote as the opening of a longer conversation. Three follow-throughs consistently produce ROI. First, a 90-minute working session 30 days after the keynote with the speaker and the system’s CMIO, CNIO, and a clinical-informatics governance lead. Second, a written follow-up memo from the speaker addressing two or three of the questions raised on stage that the keynote did not have time to answer. Third, an internal post-mortem 60 days out, where the program committee assesses whether any of the keynote’s recommendations actually entered the system’s AI roadmap. Speakers who push for that follow-through are speakers who treat the engagement as a partnership. Speakers who do not are speakers whose work ends when the audio engineer kills the lectern mic.
The Bottom Line
The AI healthcare keynote market will keep getting noisier. The filter does not. Clinical or research deployment credibility, fluency with regulatory and operational constraints, and a clean disclosure on commercial relationships. The twelve speakers above pass that filter today. Use the topic-to-speaker table to narrow further, and run a scoping call before you commit.
For broader healthcare keynote selection guidance, see our 2026 healthcare keynote guide and our feature on the 25 hospital CEO and leadership speakers reshaping the field.
Browse our healthcare speakers roster or filter by artificial intelligence and innovation.
