Who Should Get Paid for AI Used in Healthcare

A recent opinion piece, co-authored by Colin Rowell, MD, and Ronnie Sebro, MD, PhD., radiology researchers at Mayo Clinic Florida, explored an emerging question — who should get paid for AI used in healthcare? Their collaboration was published online in the August 3 Radiology: Artificial Intelligence.

The authors addressed this complex question by addressing the 7 stakeholders in the AI system. Not only are there multiple stakeholders, but there are legal issues. And even more fundamental issues to address — the moral and ethical factors to consider. Drs. Rowell and Sebro comment, “There is no consensus on who owns medical data, or for how long. There are multiple stakeholders and multiple individuals who are essential when creating an AI system.”

It’s time to consider the potential compensation claims each of the 7 stakeholders may have.

1. Without patients — their cells, images, demographic information, and outcomes — there could be no AI. The argument is convincing that these patients from whom data is collected are due some compensation. One only has to remember the case of Henrietta Lacks. The HeLa cell line has played an instrumental role in cancer research. Yet, neither she or her family had any knowledge or were asked permission regarding the cells of her cervical tumor.
2. Healthcare professionals play a critical role from ordering lab testing, imaging, and diagnostic procedures to interpretation of data leading to diagnosis; healthcare professionals create patient data that correlates and trains AI systems. Given their lengthy and expensive education, omitting those who created the data would seem problematic.  
3. Healthcare systems possess the infrastructure that helps build AI systems. Data storage facilities, EHR hardware, software, and labs are part of that infrastructure, and these entities bear the costs of a data breach.
4. Health insurance companies store patient and healthcare professional data for business operations. They may claim ownership of that data because of their indirect investment in the infrastructure to store the data. The researchers note that as data is now a new commodity, health insurance companies may lay claim to that data because it’s become lucrative.
5. S. taxpayers pay millions of dollars in taxes towards the Centers for Medicare and Medicaid Services’ budget — which, in turn, sends patient data into large databases used for AI development. According to the researchers, an argument can be made that these databases should be free and accessible to the public since taxpayers paid for them.
6. AI companies expend capital to develop AI systems, so they should have ownership claims on the AI systems they create. Most AI companies have software license agreements saying they own the AI system.
7. AI developers directly responsible for an algorithm that could generate income for many years stand to benefit from the coming surge in AI. It remains to be seen how this income would be distributed, but it is reasonable to believe that software developers and shareholders have the potential for monetary gain.

The development of AI from concept to deployment is lengthy, with multiple stakeholders potentially entitled to compensation. In 2020, the Centers for Medicare & Medicaid Services (CMS) established payment for AI — they established a new Current Procedural Terminology code for an AI tool for diagnosis of diabetic retinopathy.

CMS observed during their process, “Sustained adoption of AI through the current reimbursement framework may be challenging in a fee-for-service environment. As value-based payment models mature, in which measuring improvement in quality becomes increasingly important at decreased costs, AI becomes a valuable tool for radiologists and health care systems.”

Researchers Rowell and Sebro wind up their study by saying, “Multiple stakeholders and multiple individuals are essential when creating an AI system. Dissecting the individual contribution of each stakeholder and each individual to the development of an AI system is difficult and, in some cases, intractable. An urgent discussion is required in the scientific community to really understand data ownership as it pertains to medical AI and how its use will be reimbursed.”