Medicare’s new payment model is built for AI, and most of the tech world has no idea

Neil Batlivala has spent seven years constructing a healthcare firm that many of the tech trade has by no means heard of and that serves a affected person inhabitants most of Silicon Valley ignores. However final month, that work put him on the middle of one thing a lot larger.

His firm, Pair Workforce, introduced on April 30 it had been accepted into ACCESS, a Medicare program — as considered one of 150 individuals chosen by the Facilities for Medicare & Medicaid Providers to check what AI-driven medical care may seem like at federal scale. This system goes reside July 5.

“The federal government is creating swim lanes for AI innovation in historically regulated industries,” he advised me over a Zoom name a number of days later. “The very best answer wins, which, in regulated industries like healthcare — that’s not been the case.”

ACCESS — Advancing Persistent Care with Efficient, Scalable Options — is a 10-year CMS program testing a cost mannequin that rewards well being outcomes quite than required actions (like a sure variety of check-ins). Taking part organizations like Pair Workforce obtain predictable funds for managing qualifying circumstances and earn the complete quantity solely when sufferers meet measurable well being targets, like decrease blood strain or diminished ache. It covers diabetes, hypertension, power kidney illness, weight problems, despair, and anxiousness.

That cost construction is the actual information.

Conventional Medicare reimburses based mostly on time spent with a clinician. There’s no mechanism to pay for an AI agent that displays a affected person between visits, calls to test in, coordinates a housing referral, or makes certain somebody picks up their treatment. ACCESS creates that mechanism for the primary time.

“It’s a cost mannequin transformation,” Batlivala stated. “You simply couldn’t do that earlier than.”

The primary cohort spans a variety of individuals — AI physician startups, digital diet remedy suppliers, linked machine firms, and wearable makers like Whoop. Batlivala is skeptical of a few of them.

“I am a giant fan of wearables, however for a senior who’s scuffling with meals insecurity, I do not know the way a lot Whoop goes to have the ability to do,” he stated, including of his personal firm, “We have been constructing towards this for five-plus years now.”

Pair Workforce launched in 2019 with a particular type of affected person in thoughts: individuals managing power circumstances who have been additionally coping with unstable housing, too little meals, or lack of transportation. A few third of Individuals fall someplace in that class.

The corporate’s premise was that you could’t enhance well being outcomes with out addressing the complete context of somebody’s life. It now employs roughly 850 medical professionals, runs what it describes as the most important neighborhood well being workforce in California, and, per Batlivala, generates income above 9 figures. It has raised about $30 million, backed by Kleiner Perkins, Kraft Ventures, and Subsequent Ventures.

The mannequin has peer-reviewed proof behind it. A research, co-authored by Pair Workforce researchers and peer-reviewed by the Journal of Common Inner Drugs, evaluated Pair Workforce’s community-integrated mannequin, which blends medical, behavioral, and social look after Medicaid members with excessive charges of homelessness, severe psychological sickness, and power illness and it confirmed sturdy affected person engagement and important reductions in avoidable emergency and inpatient utilization. Batlivala says one in 4 hospital visits and one in two ER visits do not occur when a affected person is in his firm’s care.

However for years, delivering that degree of care required human groups, which restricted how briskly and cheaply it may scale. Then, about 9 months in the past, Pair Workforce deployed a voice AI agent known as Flora as its main patient-facing interface. Flora is offered 24 hours a day, handles consumption, coordinates referrals, and does the check-ins that hold sufferers engaged between medical visits.

The primary name that shifted his considering was with a 67-year-old girl residing out of her automotive, managing PTSD and congestive coronary heart failure. She spoke with Flora for over an hour. “It was each unbelievable and miserable,” Batlivala advised me. “Flora was in all probability the one ‘individual’ she’d talked to in weeks about her scenario.” Now, hour-long conversations with Flora are routine. “That is the companionship piece,” he stated. “And it seems that’s actually an intervention.”

The architects of ACCESS are themselves former startup operators. This system was designed by Abe Sutton, Director of the CMS Innovation Middle, and Jacob Shiff, Chief AI and Expertise Officer of the CMS Innovation Middle. Sutton was beforehand a enterprise capitalist at a healthcare fund known as Rubicon Founders. Shiff is a former healthcare founder. Each joined CMS beneath the Trump administration and their startup backgrounds are mirrored in this system’s design: outcome-based funds, direct-to-consumer enrollment, and a deliberate push for competitors.

There are actual dangers. Contributors are feeding terribly delicate affected person knowledge — intimate conversations about housing and illnesses and psychological sickness — right into a federal infrastructure with a documented historical past of breaches, together with uncovered Social Safety numbers. For the susceptible populations ACCESS is designed to serve, that is not an impractical concern.

There are monetary dangers, too. The observe report of CMS innovation applications is blended. A 2023 Congressional Finances Workplace evaluation discovered that the CMS Innovation Middle elevated federal spending by $5.4 billion throughout its first decade quite than producing the projected financial savings. CMS can also be paying much less per affected person monthly than many individuals anticipated, which implies the mathematics solely works for organizations which have totally automated most of their affected person interactions.

Batlivala’s reply to the reimbursement concern is that it is a function, not a bug. “If you wish to construct a mannequin that actually incentivizes the usage of AI, the reimbursement charges should be low,” he advised me. “The economics solely work if you happen to’re working a lean, AI-first operation.”

Pair Workforce says it proper now has partnerships in place that give it entry to roughly 500,000 potential sufferers, and that it desires to achieve 1,000,000 inside three years.

Healthcare buyers have been watching this carefully. Digital well being funding hit its highest Q1 complete because the pandemic this yr, with AI firms capturing the majority of it. However ACCESS has barely registered outdoors well being tech commerce press.