One much discussed topic as artificial intelligence (AI) is increasingly introduced onto diagnostic radiologists'  workstations has been whether an economic model can help support the investment made by health care providers in these software platforms. In the just published final fiscal year (FY) 2025 Medicare Hospital Inpatient Prospective Payment Systems (IPPS) rule,  the Centers for Medicare & Medicare Services (CMS) has created a temporary add on payment to hospitals for one such product, an obstructive hydrocephalus (OH) software tool from Annalise.ai (Annalise) used to triage suspected OH cases. The final rule notes that Annalise's software device - as described by the company - analyzes studies using an AI algorithm to identify suspected OH findings in non-contrast computed tomography brain scans to provide study-level output for an order and imaging management system for worklist prioritization or triage.

The add on payment arises from a CMS program to temporarily assist in the launch of such products through new technology add-on payments (NTAPs). CMS provides an overview of these add on-payments here. 

The IPPS rules at 42 CFR 412.88(a)(2) limit new technology add-on payments for discharged Medicare inpatients to the lesser of 65 percent of the average cost of the technology, or 65 percent of the costs in excess of the MS-DRG (Medicare Severity Diagnosis Related Groups) payment for the case.  The IPPS 2025 final rule finalize the maximum new technology add-on payment for a case involving the use of Annalise Enterprise CTB Triage - OH at $241.39 for FY 2025 (that is, 65 percent of the average cost of the technology).

No similar add on payment exists for services using the AI software device paid under the Medicare physician fee schedule system. 

In 2023, we reported that the U.S. Food and Drug Administration (FDA) had granted a breakthrough designation for the first time for this same radiology triage software medical device.  FDA continues to clear artificial intelligence/machine learning (AI/ML) devices, and radiology devices have consistently constituted the majority of those AI/ML devices being cleared by that agency.