AI Classified CPT Codes: What Private Practices Need to Know

Physician using a tablet to review AI classified CPT codes in an exam room.

The CPT Editorial Panel has accepted 43 AI classified CPT codes so far.
AI classified codes fall into three categories: assistive, augmentative, and autonomous.
2027 Appendix S revisions clarify physician oversight requirements by AI category.
Radiology, cardiology, and ophthalmology see the most AI CPT code activity today.
Not all AI tools qualify for a billable code; scribes and schedulers generally don’t.
A CPT code doesn’t guarantee reimbursement; payer coverage still varies.

TL;DR

Check whether your practice's AI-enabled tools fall under an AI classified CPT code, document according to the AI's role, and confirm payer coverage before you bill, since a code alone doesn't guarantee reimbursement.

To date, the CPT Editorial Panel has accepted 43 AI classified CPT codes (i.e., CPT codes for medical services and procedures that incorporate AI in a clinically meaningful way)—and the number continues to grow as AI-enabled medical technologies move from research into routine clinical practice. As more FDA-authorized AI-enabled medical technology enters the market, providers will increasingly adopt AI-assisted imaging, diagnostics, clinical decision support, and treatment planning in everyday practice. Practices that code and document these technologies correctly get paid accurately for them — and avoid the denials and payment delays that come with miscoding AI-enabled services.

What are AI classified CPT codes?

AI classified CPT codes represent medical services and procedures where AI may detect abnormalities, analyze complex medical data, generate clinically meaningful findings, or even produce autonomous diagnostic results.

AI classified CPT codes are located throughout the CPT code set; however, each AI classified code is assigned one of the following three classifications based on the role AI plays in the service:

Assistive AI: AI identifies or highlights clinically relevant information for the clinician to interpret.
Augmentative AI: AI analyzes data and provides clinically meaningful information that supports the clinician’s decision-making.
Autonomous AI: AI independently generates a clinical conclusion or recommendation that can directly inform patient care.

Consider these AI CPT coding examples:

AI Classification	Example of Artificial Intelligence CPT Codes	Service	How AI Contributes
Assistive	0764T (professional component) and 0765T (technical component)	Algorithmic ECG risk assessment for cardiac dysfunction	AI analyzes ECG data and identifies clinically relevant risk information, but it does not make the diagnosis. The physician interprets the AI output, incorporates it into the clinical assessment, and determines the final diagnosis and treatment plan.
Augmentative	75580	Noninvasive estimate of coronary fractional flow reserve (FFR) derived from coronary CT angiography	AI analyzes imaging data and generates a new clinically meaningful parameter (estimated FFR) that helps the physician evaluate coronary artery disease.
Autonomous	92229	Retinal imaging with point-of-care autonomous AI for diabetic retinopathy	AI independently analyzes retinal images and generates a diagnostic result without requiring a physician to interpret the images before the result is produced.

What revisions did the AMA make to CPT’s Appendix S?

The AMA recently revised Appendix S (AI Taxonomy for Medical Services and Procedures) to provide a clearer framework for classifying clinical software-enabled medical services and procedures. The revisions make the framework more precise, scalable, and easier to apply as AI becomes a routine part of healthcare. For private practices, the updates should ultimately lead to:

Clearer CPT code descriptors
More consistent documentation expectations
More predictable reimbursement for AI-enabled services

Appendix S (2022–2026)	Appendix S (2027 Revisions)	Why it matters to private practices
Defined three AI categories (assistive, augmentative, and autonomous), but the boundaries between them were sometimes unclear.	Refined the definitions to create clearer distinctions, particularly between assistive and augmentative AI.	Practices can better understand how much physician involvement and oversight an AI-enabled service requires.
Focused on what the machine did during the clinical service.	Focuses on the software output(s) and how those outputs contribute to patient care.	The taxonomy better reflects today’s software-enabled technologies and is flexible enough to accommodate future AI innovations.
Did not formally define key terminology used throughout the taxonomy.	Adds formal definitions for derived parameters, clinically meaningful, and automated/automatically.	Standardized terminology helps reduce confusion among providers, coders, AI developers, and payers.
Less guidance on when physician interpretation was required versus when software output could stand on its own.	Clarifies the role of physician review, interpretation, and oversight for each AI category.	Practices can better align documentation and coding with physician responsibilities.

How should providers document AI-assisted services?

Clear documentation is what stands between your practice and a denied claim. To get paid for AI-enabled services on the first submission, your records should generally:

Demonstrate medical necessity
Reflect the physician’s level of involvement based on whether the AI is assistive, augmentative, or autonomous
Support the AI healthcare CPT codes reported

Documentation by AI classification

As AI classified CPT codes become more common, providers should ensure the medical record clearly reflects how AI contributed to the service, what role the clinician played, and how the findings informed diagnosis or treatment. Consider the following:

Assistive AI: With assistive AI, providers should document that they personally reviewed and interpreted the clinical information. The AI may identify or highlight abnormalities, but it does not replace the physician’s interpretation.
Augmentative AI: For augmentative AI, documentation should reflect how the physician incorporated the AI-generated analysis into clinical decision-making.
Autonomous AI: With autonomous AI, the medical record should identify the autonomous AI service performed, the AI-generated result, and any follow-up clinical management.

Documentation best practices

Consider the following best practices for documentation of AI-assisted services:

Avoid relying solely on automatically generated AI reports without documenting the clinician’s involvement when required. The medical record should demonstrate that the physician independently reviewed the findings or exercised appropriate clinical judgment, consistent with the AI classification and CPT code requirements.
Clearly identify the physician’s role in reviewing, interpreting, or acting on the AI output. This is especially important for assistive and augmentative AI services, where physician oversight and clinical judgment remain essential components of the service.
Document medical necessity. Clearly explain the patient’s symptoms, diagnosis, risk factors, or clinical circumstances that justify using the AI-enabled service.
Include clinically relevant AI-generated findings that influenced patient care. Document how the AI output affected diagnostic assessment, treatment recommendations, follow-up testing, or referral decisions.

For practices already using an integrated tool like Tebra's AI Note Assist, documentation habits built for assistive AI can extend naturally to CPT-relevant services as more become billable.

Monitor payer policies, as coverage and documentation requirements for AI-enabled services continue to evolve. Staying current here is the difference between clean claims and reworked, delayed payments.
Verify that the AI-enabled service meets the requirements of the applicable CPT code. Review the code descriptor and any related CPT guidance to align the workflow with the code's intended use before you submit — so the claim is paid rather than kicked back for review.

What specialties will be most affected by AI classified CPT codes?

Although AI-enabled services are expanding across healthcare, adoption has been greatest in specialties where AI can improve diagnostic accuracy, efficiency, or clinical decision-making. That includes the following specialties:

Specialty	How AI Is Being Used	Potential Impact
Radiology	Image detection, measurement, workflow prioritization, and diagnostic support	More AI-enabled imaging services, evolving documentation requirements, and additional reimbursement opportunities.
Cardiology	Coronary CT analysis, plaque characterization, ECG analysis, cardiac imaging	AI can help quantify disease severity, assess cardiovascular risk, and support treatment planning.
Ophthalmology	Autonomous diabetic retinopathy screening and retinal image analysis	AI enables earlier disease detection and expands screening in primary care and community settings.
Pathology	Digital slide analysis, cancer detection, biomarker quantification	AI assists with identifying abnormalities and generating quantitative information that supports diagnoses.
Endocrinology/Primary Care	Diabetes management and automated insulin dosing	AI supports chronic disease management and may improve glycemic control while reducing clinician workload.
Neurology	Stroke detection, brain imaging analysis, seizure monitoring	AI can help identify time-sensitive conditions and support faster clinical decision-making.
Pulmonology	Lung nodule detection, pulmonary imaging analysis	AI assists clinicians in identifying abnormalities that might otherwise be overlooked.
Dermatology	Skin lesion analysis and melanoma risk assessment	AI may help prioritize suspicious lesions for further evaluation, although physician review remains essential.

What should private practices do now?

Following are some strategies to consider:

Assess your current technology

Determine whether any current or future AI-enabled tools have associated AI classified CPT codes. If codes exist, identify AI tools you're already using that now carry a billable code — that's revenue your practice may be leaving on the table today.

May Have an AI classified CPT Code	Does Not Have an AI classified CPT Code
Autonomous AI diabetic retinopathy screening (CPT 92229)	AI clinical decision support that recommends antibiotic therapy based on a patient’s symptoms and lab results.
AI coronary plaque analysis from coronary CT angiography (CPT 75577)	AI that predicts a patient’s risk of sepsis from electronic health record data and alerts clinicians.
AI-derived fractional flow reserve analysis (CPT 75580)	AI that flags patients at high risk for hospital readmission or clinical deterioration.
AI-enabled digital pathology analysis with an associated AI classified CPT code	AI that suggests differential diagnoses based on the patient’s history, symptoms, and imaging findings.

Educate providers and coding staff

Ensure clinicians, coders, and billers understand the differences between assistive, augmentative, and autonomous AI and how those distinctions affect coding and documentation.

Review payer coverage policies

Before implementing an AI-enabled service, verify coverage and prior authorization requirements — and make sure your practice's electronic claim submission workflow is set up to flag denials tied to newer or less familiar codes

Will payers recognize AI classified CPT codes?

Coverage varies by payer. Medicare, Medicaid, and commercial insurers may differ in whether they cover a particular AI classified CPT code or require prior authorization. However, as more evidence emerges and additional AI-enabled services receive CPT codes, insurers are expected to expand and refine their coverage policies. Experts say autonomous clinical AI requires entirely new payment models, not incremental modifications to existing ones.

Frequently asked questions

Appendix S is the AMA’s taxonomy for classifying AI-enabled medical services and procedures within the CPT code set. It provides a standardized framework for describing how AI contributes to patient care, helping providers, coders, payers, and technology developers apply AI-related CPT codes more consistently. As more AI-enabled services receive CPT codes, understanding Appendix S can help private practices prepare for evolving documentation, coding, and reimbursement requirements.

No, CPT code provides a standardized way to report a service, but each payer determines whether the service is covered, what documentation is required, and how much it will reimburse. Before implementing an AI-enabled technology, practices should verify payer coverage policies, prior authorization requirements, and payment rates.

Only AI that contributes directly to a reportable clinical service may have an AI classified CPT code. Administrative tools—such as AI ambient scribes, coding assistants, scheduling software, and prior authorization automation—as well as many clinical decision support tools, may improve efficiency but generally do not qualify as separately billable AI-enabled services.

Start by identifying whether any AI-enabled technologies your practice currently uses or plans to purchase have associated AI classified CPT codes. Then educate providers and coding staff on documentation requirements, review payer coverage policies, and monitor annual CPT updates as new AI-enabled services become available.

Written by

Lisa Eramo, freelance healthcare writer

Lisa A. Eramo, BA, MA is a freelance writer specializing in health information management, medical coding, and regulatory topics. She began her healthcare career as a referral specialist for a well-known cancer center. Lisa went on to work for several years at a healthcare publishing company. She regularly contributes to healthcare publications, websites, and blogs, including the AHIMA Journal. Her focus areas are medical coding, and ICD-10 in particular, clinical documentation improvement, and healthcare quality/efficiency.

Learn more about Lisa Eramo

Tags