HIMSSCast

The episode surfaces a handful of genuinely useful operational insights - the cost economics of AI enabling scope expansion, the reinforcement-learning degradation risk, and the three-part utilization-management framework - but roughly half the runtime is scene-setting, framing, and truisms about 'prior auth sucks.' A smart healthcare operator learns a few things but there is meaningful padding.

“if you're an insurance company now, to do a prior auth determination doesn't cost you the 40 or 50 bucks you used to have to pay a human. You can do it practically for free. So why not do it for more services?”

“if they can't be good correctors of errors, the algorithm thinks it's right when it's not. And so through reinforcement learning, it might just get worse over time”

A few genuinely non-obvious angles appear - AI making prior auth cheap enough to expand its scope, the perverse WISE-R contractor incentive structure, and the reinforcement-learning degradation argument - but the overarching arms-race framing is the paper's own thesis and much of the rest is standard health-policy commentary. Not recycled punditry, but not frontier thinking either.

“it might be the best bad idea we've had for keeping healthcare costs down”

“AI might be used as a reason to expand the scope of prior auth”

Michelle Mello is a legitimately credentialed empirical health law scholar at Stanford with a primary-research paper directly on the topic; she is not a podcast-circuit thought leader. However, she is an academic-researcher rather than a payer or provider operator who has run these processes at scale, which limits the practitioner depth a B2B operator would most value.

“She's an empirical health law scholar who's a professor of law at Stanford Law School and a professor of health policy at Stanford School of Medicine”

“I've been working with other colleagues at Stanford here, Elise Adams, and a, uh, med student, Stephen Gong, on understanding the history of prior authorization”

The episode provides a handful of real numbers - 12% traditional Medicare prior-auth rate, 85 - 93% approval rates, $40 - 50 per human determination cost - and names a specific federal program (WISE-R). This is above average for a 19-minute interview but the numbers are offered briefly without sourcing depth, and no named company examples or outcome data from specific deployments are cited.

“Less than 12% of services in that world historically have been subject to prior auth”

“the lowest scoring plans, we might have no's 20% of the time, but more typically we're talking 85, 90, 93% of the time it's a yes”

The host's questions are structurally reasonable but mostly broad scene-setters; he never probes a specific claim, challenges the guest's framing, or follows up on a weak answer. When the guest flatly said nothing surprised her, the host simply moved on, which is a clear missed opportunity for a tighter interview.

“Was there anything in your research that maybe took you by surprise a little bit?”

“I really don't think anything comes to mind that I find surprising. No.”