Comment on CMS-2026-2377-0002

Medicare's CPT and coverage structure treats AI-assisted clinical decision-making as physician/QHP work product. Codes like the CPT Category III codes for AI/ML-based algorithm analysis (e.g., 0721T–0724T and similar augmented intelligence codes), and CMS's broader billing guidance, route reimbursement and liability to whoever is credentialed to "interpret" or "order" based on the AI output which today is exclusively physicians and other qualified health professionals with independent ordering authority. Nurses, meanwhile, are frequently the ones actually executing on that AI output: triaging based on an early-warning sepsis algorithm, adjusting a care pathway per a predictive readmission-risk score, responding to a deterioration alert. The clinical judgment embedded in the AI tool has already happened. What's left is implementation which is squarely within nursing scope in every state. Yet because nurse practice acts were written before AI-driven clinical decision support existed, they don't distinguish between "a physician's independent judgment" and "a physician co-signing an AI-generated recommendation that nursing staff already acted on." The result is a signature requirement that adds no clinical value — it's a paperwork gate, not a safety gate. Why this matters beyond administrative burden Care delay: If a nurse must wait for physician co-signature before acting on an AI-flagged deterioration, the AI's speed advantage is partially negated by a workflow bottleneck the AI didn't create and doesn't need. Misallocated liability: The physician is nominally accountable for a decision that, functionally, an algorithm generated and a nurse executed — while the nurse practice act gives nurses no independent authority to act on AI output even within their existing competency. Regulatory lag becomes a scope-of-practice ceiling: This isn't really a CMS billing problem in isolation — it's what happens when reimbursement policy (which moves relatively fast, via CPT updates) outpaces practice-act statutes (which move slowly, state by state). CMS sits at the point where this gap is most visible, because it's the one federal actor touching both the coding side and, indirectly, the conditions of participation that shape scope-of-practice expectations in Medicare-certified facilities. The ask Acknowledge the gap explicitly, that current AI-related CPT and coverage policy assumes a physician/QHP as the decision-making node, while implementation increasingly happens at the point of AI output, not point of physician review. Commission or fund a study (via CMS Innovation Center or in coordination with HRSA/state boards of nursing) on whether AI-supported clinical decision support changes the risk calculus for expanding nursing scope — specifically for protocol-driven, algorithm-flagged actions that don't require new diagnostic judgment. Consider a pilot pathway — similar to how CMS has piloted expanded scope for advanced practice providers in other contexts — allowing nurses to act on specific categories of AI-generated alerts without physician co-signature, under defined guardrails (e.g., algorithm validated for the specific clinical scenario, nurse practicing within an existing protocol, physician notified but not required to co-sign in real time). Coordinate with state boards of nursing, since CMS can't unilaterally change nurse practice acts, but can signal via conditions of participation, quality measures, or innovation model design that current signature requirements are a friction point CMS wants states to reexamine.