Jun 28, 2024 · 37m · a16z
Grand Challenges in Healthcare AI with Vijay Pande and Julie Yoo
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of 'Raising Health' by a16z Bio + Health, General Partners Vijay Pande and Julie Yoo discuss how artificial intelligence can address major healthcare challenges, from eliminating administrative waste and transforming clinical trials to enabling continuous care and autonomous medical co-pilots.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The host as informed peer 4.7 Guest teaching 3.7 Guest disagreement 1.1 The host pushing back 1.2
speaking balance: gold is the host, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 13:10 Heretical Reframe on Healthcare Technology
In a very agreeable dialogue, Vijay gently challenges the premise that software alone solves scheduling capacity, arguing that the way medicine is practiced fundamentally has to change first.
Hardest push from the host ▶ 19:26 Reframing AI Prior Authorization LawsuitsJulie reframes the narrative around AI prior authorization lawsuits, correcting the assumption that AI increases denial rates by clarifying that it merely accelerates existing human-written rules.
Biggest teaching moment ▶ 31:10 2x2 Matrix for AI Clinical ProgressionVijay lays out a clear conceptual 2x2 framework evaluating decision complexity and mistake tolerance to explain why AI adoption will move from nursing to general practice before reaching specialists.
The host holds their own ▶ 4:43 Claims as Units of Algorithmic LogicJulie demonstrates technical domain expertise by conceptualizing insurance claims as units of serialized logic, eliciting explicit acknowledgement from Vijay that he hadn't viewed claims that way.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The host as informed peer | Guest teaching | Guest disagreement | The host pushing back | Why |
|---|---|---|---|---|---|---|
| High-Level Optimism and Immediate Healthcare AI Impact | 4 | 2 | 1 | 1 | Julie sets the stage on healthcare AI optimism, and Vijay frames the immediate adoption challenge around 10x workflow improvements versus 10% incremental ease. The conversation is highly collaborative and aligned, setting up broader industry themes. | |
| B2B Administrative Automation and Claims Adjudication | 6 | 3 | 1 | 1 | Julie demonstrates strong domain knowledge by reframing insurance claims as units of algorithmic logic, surprising Vijay who notes he hadn't thought of it that way before. They discuss B2B administrative waste and contract digitization companies like Turquoise. | |
| Always-On Clinical Trials and Real-World Causal Data | 4 | 5 | 1 | 1 | Vijay educates on how Bayesian statistics and AI enable causal analysis in real-world clinical trial data, going beyond mere correlation. Julie complements this with insights from the VA CIO regarding millions of implicit trials occurring daily. | |
| Spot Market Pricing and Real-Time Economic Adjudication | 5 | 4 | 2 | 2 | When discussing spot pricing and provider scheduling, Vijay offers a mild reframe that the fundamental issue isn't software but changing the underlying way medicine is practiced. Julie connects this clean-sheet approach to Devoted Health and physician defensive scheduling. | |
| Post-Meaningful Use Era, Data Assets, and Value-Based Shifts | 6 | 3 | 1 | 2 | Julie offers deep sector context regarding post-Meaningful Use EHR data assets and value-based care shifts. She pushes back on popular commentary about AI prior-authorization lawsuits, clarifying that AI is merely increasing the speed of denials under human-written rules rather than the rate. | |
| LLM User Interfaces, Patient Storytelling, and Role Unbundling | 5 | 4 | 1 | 1 | Vijay frames LLMs primarily as natural language user interfaces rather than oracles, prompting a discussion on unbundling clinical roles. Julie adds depth around patient narrative synthesis and ambient scribing tools. | |
| AI Care Team Companions and Real-Time Patient Triage | 4 | 3 | 1 | 1 | Vijay shares a personal anecdote about cutting his hand with a chef's knife in the ED to illustrate the acute need for real-time AI triage. Julie expands on this by describing inpatient walkie-talkie AI companions and personal triage habits. | |
| Navigating Regulatory Guardrails for Healthcare AI | 4 | 4 | 1 | 1 | Julie probes regulatory boundaries for AI builders, and Vijay outlines strategies for collaborating with regulators like ONC. They agree that regulations should target specific healthcare use cases rather than underlying model technologies. | |
| The AI Doctor Horizon and Clinical Co-Pilot Progression | 4 | 6 | 1 | 1 | Vijay details a 2x2 decision framework based on complexity and mistake tolerance to map the progression toward an AI doctor, starting with nursing and general practice before reaching specialists. Julie queries the near-term adoption dynamics of physician co-pilots. | |
| Wishlist Startups: Clinical Trial AI and AI-Native Health Plans | 5 | 3 | 1 | 1 | Both speakers share wishlist startup ideas: Vijay envisions AI-optimized clinical trial selection and design, while Julie details a full-stack, AI-native health plan taking full population risk. |
Statements from this episode (22)
Pande: Healthcare AI needs 10x gains or zero-friction adoption
“So I think the immediate part is something that is so good, like not 10% better than what you have now, but like 10 X better than what you have now, that the adoption becomes natural. Or so easy to adopt that even 10% better could work.”
Yoo: Healthcare labor crisis driven by technological administrative burdens
“One of the number one crises that our healthcare industry is facing right now is a labor crisis. And that we have both, both have a shortage of labor to do these kind of highly specialized jobs that we have, whether it be clinical or whether it be administrati…”
Yoo: 90% of healthcare payments are reimbursed revenue requiring claims
“90% of payments in healthcare are reimbursed revenue, where the provider has to submit literally a claim to the payer that is effectively an algorithm in many ways.”
Yoo: Automating healthcare claims could eliminate 30% of system waste
“You could eliminate 30% of the waste in our system if you were able to do that.”
Pande: Always-on clinical trial infrastructure is impossible without AI
“Yeah, the funny thing about that idea is, I mean, it's a very exciting idea, because we could gain so much knowledge, and we could improve healthcare so much, but like, it's a ridiculous thing to imagine doing, like, without something like AI. Without AI, I do…”
Pande: AI and Bayesian statistics enable causal understanding in health data
“With all this data AI is great, especially certain types of AI, like Bayesian statistics, are really good at causality. And so we could actually have causal understanding.”
Yoo: Monolithic contracts and claims systems prevent real-time healthcare spot pricing
“Today it's probably deemed illegal, honestly, by many of the contracts, because you're sort of bound by these, again, these monolithic agreements that highly specify, and the fact that you have this claim system there's no, really no notion of a real-time adju…”
Yoo: Long appointment wait times mask widespread wasted healthcare capacity
“You as a patient are told to wait weeks for a doctor's appointment, and you assume that that's because every doctor is booked out solid. But it actually turns out that a lot of the capacity in our system goes completely wasted.”
Yoo: Doctors design schedules defensively to protect against bad software
“You would see the way that doctors designed their schedule and very much to your point, they would design the templates of their schedules specifically in a very protective, like almost a defensive way. Because they felt wronged by the way that the system sent…”
Yoo: Fewer than 70% of doctors used EHRs five years ago
“Even five years ago, like, less than, you know, 70% of doctors had an electronic health record.”
Yoo: 100% of healthcare incumbents at JPM claimed active AI deployment
“At this last JP Morgan conference, like, a hundred percent of the incumbent, you know, payers and providers got up on stage and talked about not just what they want to do with AI, but how they're actually deploying AI in practice”
Yoo: AI insurance lawsuits stem from denial speed, not denial rates
“The rate of denials is not going up, it's the speed with which the denials are happening that's going up.”
Pande: LLMs' primary underappreciated value is serving as user interfaces
“Well, so I think the thing that's really underappreciated out of the LLM is like people think of it as like this Oracle or something like that, but I think it's maybe at least for us, I think of it as a UI.”
Yoo: Specialized AI models are required to accurately interpret medical information
“This is where I think one of the prime examples where we certainly believe that a specialist, you know, model is necessary to understand the specific nuances of how to interpret medical information versus general internet information.”
Pande: AI in healthcare should act as care team peer, not just co-pilot
“Where the AI comes in, one idea is a co-pilot, which is like each one of the team members has a co-pilot. But what's interesting about the data is this is like, The AI is a peer, you know, contributor, you know, more of the team and has its role that actually …”
Yoo: Startups are integrating LLMs into inpatient nurse walkie-talkie streams
“That actually reminds me of a company that we saw that It sort of reminded me of, like, what if every nurse in the inpatient ward, because the inpatient setting is very chaotic, very active, things are, like, surprises happen all the time, and a lot of nursing…”
Yoo: LLMs can unbundle care by automating triage, even without physical treatment
“Unbundling the role of a clinician. You know, there's one part which is actually the treatment part, so that's the part where maybe we can't necessarily today build an LLM that will stitch your finger. But the notion of triage, right and kind of getting you to…”
Pande: Healthcare regulators are eager to consult with AI startups
“In the cases where there is any gray zone, I think the regulators are I think eager to chat with startups.”
Yoo: Healthcare is ahead of tech regarding AI regulatory frameworks
“These are like some of the rare cases where healthcare is actually ahead of the curve as far as technology goes.”
Pande: AI primary care is viable because general practice relies on triage
“And I think the general practitioner, sort of concierge doctor, that tier is kind of a really interesting tier, because largely you're triaging and sending off to specialists. So the AI doesn't have to be an oncologist and a cardiologist and all these things. …”
Pande: Small AI efficiency gains in high-cash-flow clinical trials yield massive impact
“Like, I'm working on, like, something, you know, in drug design or whatever to make big leaps and bounds, and small things for big cash flows can have a huge impact. So something for clinical trials could be huge, or even just picking, like, the order of rank …”
Yoo: A 1% cost optimization in health plans unlocks hundreds of millions
“Where a one percent impact on the cost structure of a health plan or the way that you underwrite risk in a certain health plan could literally mean hundreds of millions of dollars of either cost savings or better economics to the providers who are part of thos…”