How should health care pay for clinical AI? New report raises tough questions

Clinical artificial intelligence is moving into everyday care faster than the payment systems built to support it. That mismatch is at the center of new research from the Peterson Health Technology Institute (PHTI), which asks a question the industry has largely avoided: How should health care actually pay for AI?

According to PHTI, applying today's reimbursement frameworks to AI-enabled care carries real risk. Fee-for-service could reward extra billable activity and inflate costs, while pay-for-performance and capitated models may fail to reward adoption of AI that genuinely improves outcomes. The report uses hypertension management — an area where AI-driven monitoring is already reshaping primary care — to illustrate how differently AI tools would need to be paid depending on how much clinical judgment they take on.

That distinction, between AI that assists a clinician and AI that acts more independently, turns out to matter enormously for payment design — and gets to the core of whether clinical AI ends up lowering costs or adding a new layer of reimbursement complexity.

Caroline Pearson, PHTI's executive director, spoke with Medical Economics about what the findings mean for physicians, practices and the payment models shaping the next phase of AI adoption.