CPT code set facilitates development of useful AI-powered devices

Andis Robeznieks , Senior News Writer

Technology that may come into focus as part of the 2024 digital medicine picture includes:

  • Assistive AI- rules-based algorithms that help physicians choose a patient’s best cancer treatment.
  • Augmentative AI-software that helps assess the severity of coronary artery disease.
  • Basic telephones and treadmills linked to virtual-reality environments.

Half the dues, all the AMA benefits!

AMA membership offers unique access to savings and resources tailored to enrich the personal and professional lives of physicians, residents and medical students.

“Assistive,” “augmentative” and “autonomous” are the terms used to describe health care services or procedures delivered via augmented intelligence (AI), often called artificial intelligence, in the Current Procedural Terminology (CPT®) code set.

The AI taxonomy can be found in Appendix S of the CPT code set and it provides uniform terminology to describe the impact of AI on the service or procedure. The differentiation between the work performed by the machine and the work performed by the physician or qualified health professional (QHP) creates a path to payment for AI-related medical services and procedures.

In late November, the AMA released new principles for augmented intelligence (AI) development, deployment and use (PDF)—a critical step toward fostering a consistent governance structure for advancements in health care technology. 

Richard Frank, MD, PhD, a member of the CPT Editorial Panel and the former chief medical officer for the German devicemaker Siemens Healthineers, co-chaired the AI Working Group of the AMA-convened Digital Medicine Payment Advisory Group, which generated the content for Appendix S.

During the AMA CPT® and RBRVS Annual Symposium, held virtually this year, Dr. Frank highlighted additions to the CPT code set related to services using digital health tools, including revisions to the appendices for the AI taxonomy and telemedicine services.

“We took a view to what we considered then to be the ‘foreseeable future,’ which we defined as 2025,” Dr. Frank said, describing the thinking of the working group.

“The taxonomy as it is built today is intended to help describe the relationship between the work which is done by the machine and the work that is done by the human,” he added. “It’s also intended to facilitate the valuation of the procedure relative to the work being done by the human.”

He offered a brief description of the AI taxonomy’s categories:

  • Assistive AI identifies clinically relevant data and brings it to the attention of the physician. There is no analysis involved.
  • Augmentative AI analyzes data for output that is “clinically meaningful,” and the physician will understand how to use the device’s output in the patient’s care pathway.
  • Autonomous AI interprets data and makes recommendations.

“These are to be distinguished from software which has been around for a long time—particularly in radiology in optimizing the image acquisition or reconstructing the image—which are simply processing but not actually analyzing data,” Dr. Frank said. “The terms ‘assistive,’ ‘augmentative’ and ‘autonomous’ represent more sophisticated work being done by the machine than in the past.”

Dr. Frank highlighted an example of a service using assistive AI, “Pharmaco-oncologic Algorithmic Treatment Ranking,” that is in the 2024 CPT Codebook as Category III code 0794T. (Category III CPT codes are temporary and used for emerging technologies, services and procedures.)

“The rationale for this new Category III code has been established to report an assistive-AI, rules-based algorithm for ranking a specific patient’s pharmaco-oncologic treatment options,” Dr. Frank explained.

“Assistive” is the appropriate term, Dr. Frank said, because the algorithm simply detects relevant data. It doesn’t provide any diagnosis or management decisions or make recommendations. “It simply presents options to the physician for their consideration,” he said.

The 2024 codebook also includes the first Category I code using the “augmentative” descriptor.

Cardiology code 75580 for fractional flow reserve with CT, includes the augmentative AI software analysis of the data derived from a coronary computed tomography angiography with interpretation and report by a physician or other QHP. The code replaces Category III codes 0501T–0504T.

The software analysis is described as “augmentative” because it provides clinically meaningful output that is predicated on the standard of care for the physician to interpret in assessing the severity of coronary artery disease, Dr. Frank said.

Defining what telemedicine must do

The CPT Editorial Panel recently convened a telemedicine work group, which Dr. Frank co-chaired, to establish new criteria for services that would be included in CPT Appendix P, which covers services delivered by synchronous audio-video technology, and Appendix T, which covers services delivered via synchronous audio-only technology.

“This criterion requires that the totality and quality of the communication must be sufficient to meet the requirements for the same service if services were to be rendered during an in-person, face-to-face interaction,” Dr. Frank said.

The work group produced a list of nine factors that could be offered as evidence of the adequacy of the service, and includes these elements that Dr. Frank said are important to patients, physicians and even payers. The telemedicine service should:

  • Facilitate a diagnosis or treatment plan that may reduce complications.
  • Lower the need for diagnostic or therapeutic interventions.
  • Lower hospitalizations.
  • Lower in-person visits to the emergency department.
  • Lower in-person visits to physician or other QHP offices, including urgent-care centers.
  • Hasten resolution of the problem.
  • Lower quantifiable symptoms.
  • Reduce recovery time.
  • Enhance access to care, such as for rural and vulnerable patients.

The CPT® 2024 Professional Edition and CPT® Changes 2024: An Insider’s View are available from the AMA Storefront on Amazon.