Overview
The American Medical Association (AMA) introduced a proposed billing framework called “clinically meaningful algorithmic analyses,” or CMAA. The framework would allow health care facilities to seek reimbursement for AI-driven clinical outputs that algorithms generate without physician involvement at the point of care. Nursing organizations and professional advocacy groups strongly oppose the proposal. They contend that it creates a revenue path for AI-generated analyses while continuing to exclude bedside nurses from standalone billing recognition.
This alert summarizes the CMAA framework, related CMS and Food and Drug Administration (FDA) regulatory developments, the nursing profession’s objections, key legal risks, and practical guidance for health care clients. Notably, the CMAA proposal coincides with the Centers for Medicare & Medicaid Services' (CMS's) 2027 Medicare Hospital Outpatient Prospective Payment System and Ambulatory Surgical Center Proposed Rule (Proposed Rule), which introduces a temporary payment category for clinical AI tools called "Software as a Medical Service" (SaMS). Together, these developments signal a multi-front regulatory shift in how AI-driven clinical services will be recognized and reimbursed.
The CMAA Proposal
Under the CMAA framework, hospitals could bill for algorithmic services that interpret clinical data—including diagnostic imaging, lab panels, and cardiac rhythm recordings—and generate actionable clinical outputs. The services would not require a physician to perform the underlying analysis. The AMA’s Current Procedural Terminology (CPT) Editorial Panel created the initiative after receiving an increased number of applications for fully algorithm-dependent services. This prompted the AMA to form a dedicated Digital Medicine Coding Committee to standardize the approach.
CPT codes provide the standard language that the U.S. health care system uses to identify and reimburse clinical services. Without a code, payers cannot recognize a service for reimbursement. The AMA uses Category III CPT codes—temporary codes for emerging technologies—to collect data on clinical efficacy, utilization, and outcomes before a service meets the higher evidentiary threshold for Category I status. Category III codes are provisional and subject to sunset if they do not meet the evidentiary requirements for permanent Category I designation. Health care organizations should be aware that building revenue cycle infrastructure around Category III codes carries inherent uncertainty and should plan for the possibility that specific CMAA codes may not achieve permanence.
Intersection with FDA Oversight
The CMAA billing framework does not address federal regulatory approval requirements for AI-enabled clinical tools. Many autonomous AI applications, particularly those that analyze medical images, laboratory data, or cardiac rhythm recordings, to produce diagnostic or treatment recommendations, may meet the FDA's definition of "Software as a Medical Device" (SaMD) and require premarket clearance or authorization under the FDA's regulatory framework.
Billing eligibility under CMAA and FDA regulatory approval operate on separate tracks. A clinical AI tool may qualify for a CMAA billing code but still require FDA clearance before it can be lawfully marketed or deployed. Conversely, FDA-cleared AI tools may not automatically qualify for reimbursement under the CMAA framework. Health care providers should evaluate both regulatory and reimbursement pathways before deploying AI clinical tools.
Appendix S of the AMA's CPT code set classifies AI-enabled services as assistive, augmentative, or autonomous based on the amount of clinical work the algorithm performs. The table below summarizes Appendix S.
| Service Components |
Assistive | Augmentative | Autonomous |
| Primary objective | Supports a physician's or other qualified health professional's (QHP's) clinical work by providing clinically relevant data,
such as results described as “likelihood of," “suggestive of," or “risk for." |
Produces clinically meaningful output parameters that differ from the input and identify a particular clinical condition, including, for example, “predictive of” or “prognostic of." | Produces clinically meaningful parameters and uses them to generate interpretations or conclusions in the form of a diagnosis or intervention. |
| Requires documentation showing clinical meaningfulness | No | Yes | Yes |
| A physician or other QHP must interpret and report the output | Yes | Conditional (see Note) | No |
| Examples in the CPT code set | Algorithmic electrocardiogram risk-based assessment for cardiac dysfunction (0764T, 0765T) | Noninvasive estimate of coronary fractional flow reserve (FFR) (75580) | Retinal imaging (92229) (Level I) |
| Note: Existing codes may capture physician work related to services that use augmentative software. For example, an output parameter may serve as a data element in an evaluation and management service or as a factor in pre-surgical planning. It may also form part of another service that includes physician or other QHP interpretation.
Of the three classifications, autonomous AI services are the most directly affected by the CMAA. |
|||
Nursing Organizations Push Back
Nursing leaders say the CMAA proposal exposes a persistent inequity in clinical labor compensation. Nurses account for roughly 30 percent of hospital labor spending, approximately $266 billion annually, yet hospitals bundle nursing services into facility room rates and do not provide a dedicated billing mechanism. One price transparency analysis of an academic medical center with more than 1,000 beds found 158,475 billable line items but not a single inpatient nursing charge.
Rebecca Love, MSN, RN, founder of the Commission for Nurse Reimbursement, argues that the same coding infrastructure now being built for algorithms could serve as a template for nursing reimbursement. Love warned that if AI shows up as a billable line item, but nursing stays buried in the room charge, the money will naturally follow what is visible on the ledger. Over time, that imbalance could make nursing look like a cost center rather than a value driver — even though it remains essential to patient care.
Nursing advocates also highlight a structural irony: autonomous AI cannot function without human intervention. Nurses must respond to algorithmic alerts, implement recommendations, and monitor outcomes. Yet under the emerging model, only the algorithm’s contribution generates revenue.
The reimbursement question also raises unresolved liability concerns. If autonomous AI generates billable clinical outputs, the question of who bears responsibility when those outputs are inaccurate or harmful remains largely unanswered. Hospitals adopting AI billing codes will need to assess malpractice exposure, review indemnification provisions in AI vendor contracts, and confirm whether their professional liability insurance policies cover algorithm-driven clinical decisions.
Additionally, the characterization of AI services as "autonomous" under the AMA's taxonomy should not be interpreted as requiring zero human oversight. Federal and state regulators are increasingly proposing AI governance frameworks that mandate human-in-the-loop requirements for high-risk AI applications in health care. Even where CMAA codes classify a service as autonomous for billing purposes, applicable law may still require meaningful human oversight of the AI's clinical outputs.
The American Society of Registered Nurses has stressed that AI is a tool that cannot replace a nurse's clinical judgment or ability to detect subtle but critical changes in a patient's health status.
Critics of the CMAA framework also say the longstanding justification for excluding nursing from standalone billing—that nursing work is too complex to separate—no longer holds because the AMA has solved the same problem for algorithms. They argue that CMAA’s own definition—producing “medically actionable outputs” from clinical data—describes work that nurses perform every shift.
The AMA will soon consider stakeholder comments submitted on the proposed CMAA framework. Meanwhile, nurses and other health care providers may continue weighing in on AI-related reimbursement by submitting comments on CMS's Proposed Rule through August 31, 2026.
Practical Guidance for Health Care Clients
- Monitor the AMA's CMAA proposal and CMS rulemaking closely, including the August 31, 2026 deadline for comments on the Proposed Rule.
- Consider submitting formal comments on CMS's Proposed Rule to advocate for your organization's interests in the AI reimbursement framework.
- Evaluate current and planned AI clinical tools to determine whether they qualify as assistive, augmentative, or autonomous under the AMA's Appendix S taxonomy.
For tools classified as autonomous, also evaluate whether they require FDA clearance or authorization as SaMD.
- Assess whether your revenue cycle and billing infrastructure is ready to support AI-related CPT codes and CMS's SaMS payment category.
- Develop documentation protocols now to demonstrate the clinical validity of AI-generated outputs and the appropriateness of billing, as payers will likely audit AI-billed services.
- Review existing nursing labor cost structures and documentation practices in light of the growing debate over nursing reimbursement parity.
- Engage compliance, legal, and clinical leadership in cross-functional planning for AI billing integration. Compliance programs should specifically address AI-related risks, including Anti-Kickback Statute implications if AI vendors offer volume-based pricing and False Claims Act exposure if AI-generated outputs used for billing are inaccurate.
- Review AI vendor agreements with particular attention to data ownership, indemnification for erroneous clinical outputs, representations regarding FDA clearance status, and termination rights if applicable reimbursement codes are not finalized or are subsequently revoked.
- Assess state-level AI-in-health-care legislation that may impose requirements beyond the federal billing framework, including transparency obligations, algorithmic bias auditing requirements, and patient notification mandates.
- Watch for developments from the AMA's Digital Medicine Coding Committee and the CPT Editorial Panel on finalizing CMAA codes.
If you have questions about CMAA's proposal, CMS's rulemaking, or the potential impact of AI clinical tools for your organization, please contact Jade Davis or Kate Crawford.