Who gets paid when the AI does the work? with Caroline Pearson of the Peterson Health Technology Institute

A clinical artificial intelligence (AI) tool costs little to run and can be billed many times over, while fee-for-service rates are set on the time and effort a service is expected to take. Caroline Pearson says that gap is what makes technology-based care lucrative under today’s payment system and precarious under the next one.

Medical Economics Managing Editor Todd Shryock sat down with Pearson, executive director of the Peterson Health Technology Institute, a nonprofit that evaluates the clinical and economic value of emerging health technologies.

Pearson separates assistive AI, which a supervising physician deploys, oversees and bills for, from autonomous AI, which delivers some care independently. She gets into why liability for an autonomous prescribing tool sits with the physician’s malpractice coverage today, how Medicare’s new Advancing Chronic Care with Effective, Scalable Solutions (ACCESS) Model pays enrolled technology companies directly for chronic care management, why accountable care organizations (ACOs) have not found the evidence on these tools compelling enough to adopt them, and why she argues the harder work is change management, meaning redesigned workflows and a rethink of how clinical staff spend their time.

Pearson’s recommendations are laid out in the institute’s report on paying for clinical AI. She has also detailed what the findings mean for practices weighing these tools against their own payer mix.

Don’t miss our recent episodes on state corporate practice bans, disclosing medical errors, physician corporatization and malpractice rates.

Music Credits:

Paper Cranes by Buurd - stock.adobe.com
A Textbook Example by Skip Peck - stock.adobe.com

Editor’s note: Episode timestamps and transcript produced using artificial intelligence (AI) tools.

0:00 – 0:27 | Cold open Pearson on why technology doesn’t need a living wage.

0:27 – 1:35 | Introduction Austin Littrell introduces the episode and the guest.

1:35 – 1:55 | Meet Caroline Pearson Todd Shryock introduces Pearson and the Peterson Health Technology Institute’s report on paying for clinical AI.

1:55 – 2:24 | The takeaway for practices The potential is real, the incentives to adopt are not, and the payment models available now risk raising health care costs.

2:24 – 3:37 | Why fee for service inflates the bill Rates are set on clinician time and effort, but the marginal cost of running a tool is low and it can be billed many times over. Pearson points to remote patient monitoring as a live example.

3:37 – 4:18 | More patients, or more revenue per visit Seeing more patients would stretch the workforce further. The concern is revenue per visit rising without more patients seen or better outcomes in those visits.

4:18 – 5:47 | Assistive AI vs. autonomous AI Assistive tools are deployed, overseen and billed by the supervising physician. Autonomous tools deliver facets of care on their own, with medication prescribing and hypertension titration as the nearest examples.

5:47 – 6:38 | P2 Management Minute Keith Reynolds shares practice management tips and invites listeners to submit their own workflow ideas.

6:38 – 7:47 | Getting paid to review the algorithm Remote patient monitoring codes already pay physicians to review AI analysis of patient data. Pearson on setting the right value for oversight payments and keeping clinicians cognitively sharp as they use the tools more.

7:47 – 8:39 | Who is liable when it goes wrong Prescribing tools generally operate under the physician’s malpractice insurance, which makes it the physician’s liability. Pearson on the push to move that to the vendor and the legal infrastructure that does not exist yet.

8:39 – 10:20 | Cutting the physician out of the payment Medicare’s ACCESS Model pays enrolled organizations directly for chronic care management, and Doctronic is piloting prescribing in Utah. Pearson on preserving the physician’s role and setting a payment level that reflects what the technology actually costs.

10:20 – 11:25 | Why ACOs haven’t bought in Risk-based models need evidence of better outcomes and lower total cost rather than a revenue play, and Pearson says the evidence available has not been compelling enough.

11:25 – 12:26 | Change management is the other half Redesigning workflows and rethinking how clinical staff spend their time, at the system level and the individual level.

12:26 – End | Closing thoughts and outro Shryock thanks Pearson, and Littrell wraps the episode.