There is no single "Medicare reimbursement pathway" for AI-enabled healthcare software — there are several, they don't share a coverage decision-maker, and FDA clearance doesn't automatically unlock any of them. Getting paid for an AI tool is a separate project from getting it cleared, and the two run on different timelines with different evidence standards.
Current as of October 2026 — CMS rulemaking in this area is active and changing; verify the current status of any specific proposal before relying on it.
A new AI diagnostic tool usually enters the CPT code set as a Category III code — a temporary tracking code with no fixed national Medicare payment, typically sunsetting around five years unless enough adoption evidence supports graduating to a permanent Category I code. HeartFlow's AI-derived coronary blood-flow analysis followed exactly this path: four Category III codes in 2018, replaced by a single Category I code (75580) effective January 2024 once the evidence base was there. Autonomous AI retinal screening took a more direct route — CPT 92229 launched as a Category I code in 2021, with CMS setting its national payment amount in that year's Physician Fee Schedule rule. Neither path is fast, and a Category III code alone tells a buyer nothing about what, if anything, a payer will actually pay.
For inpatient hospital stays, New Technology Add-on Payments (NTAP) can cover up to 65% of a qualifying technology's cost above the standard DRG payment — 75% if the technology carries FDA Breakthrough Device designation. Qualifying requires the technology to be genuinely new, to exceed a cost threshold, and generally to demonstrate substantial clinical improvement over existing alternatives (Breakthrough-designated products skip the newness and improvement criteria). Two real AI examples: Viz.ai's stroke-triage software was the first AI product ever granted an NTAP, in 2020, worth up to $1,040 per eligible patient; Caption Health's AI-guided cardiac ultrasound software received one effective October 2021. Both required FDA marketing authorization already in hand, or a submission FDA had formally accepted, by a fixed annual deadline — NTAP is one of the few pathways where the FDA link is explicit and contractual, not just practical.
CMS tried to make this automatic once, and reversed course. The Medicare Coverage of Innovative Technology rule would have granted any FDA Breakthrough Device four years of automatic Medicare coverage the moment it cleared FDA review. CMS rescinded the rule in November 2021, stating plainly that FDA authorization evidence — often generated in younger, healthier trial populations — doesn't necessarily establish that a technology is reasonable and necessary for the Medicare population specifically. FDA clearance remains a practical prerequisite for nearly every reimbursement conversation, and an explicit, contractual one for NTAP. It is not, on its own, a coverage decision.
National Coverage Determinations — issued by CMS, binding nationwide — exist for only a small slice of Medicare's coverage policy. For a specific new AI diagnostic tool, the operative document is far more often a Local Coverage Determination, issued by the Medicare Administrative Contractor for that region, filling the gap where no national policy exists. This means coverage for the same tool can genuinely differ between a hospital in one MAC jurisdiction and one in another — a fact worth confirming directly with the relevant contractor rather than assuming national uniformity.
CMS's proposed CY2027 outpatient payment rule would formally define a new category, "Software as a Medical Service" (SaMS), and assign 36 HCPCS codes covering AI diagnostic software — including AI retinal screening, echocardiogram-based heart-failure detection, and coronary blood-flow estimation — to new, separately payable technology classifications under a new status indicator. CMS describes this as an interim policy while it develops a longer-term framework. As of this writing the rule is proposed, not final — confirm whether a final rule has since been issued before treating any of this as settled.
Where Eclypse sits: Eclypse doesn't bill Medicare or secure codes on a client's behalf — it's a platform vendor, not a billing entity. What a deployment produces is the evidence a reimbursement conversation actually needs: documented clinical validation, a reconstructible audit trail, and the kind of substantiation an NTAP "substantial clinical improvement" case or an FDA submission requires, built in from the start rather than assembled after the fact.
See the FDA's AI credibility framework for the clearance side of this, and what HIPAA actually requires of an AI system for the other half of the US compliance picture. For how both fit together, see Eclypse for the United States.
Category I codes are permanent, with a nationally set Medicare payment amount. Category III codes are temporary tracking codes with no fixed national payment, typically sunsetting around five years unless they graduate.
No. CMS rescinded its own rule that would have made FDA Breakthrough status an automatic coverage trigger, specifically because FDA evidence doesn't necessarily establish Medicare-population reasonableness and necessity.
A supplemental inpatient Medicare payment covering up to 65% of a qualifying new technology's cost above standard payment, requiring newness, a cost threshold, and generally substantial clinical improvement.
Usually a Medicare Administrative Contractor via a Local Coverage Determination, not CMS nationally — coverage for the same tool can differ by region.
No. Reimbursement rules change frequently — this is a starting point for internal discussion, not a substitute for qualified reimbursement counsel.
Our proprietary AI orchestration platform for healthcare: one engine, a registry of reusable task modules and domain agents, and a governed knowledge base — deployed inside your walls and run by your team.