Cardiology groups to Trump administration: AI still has a long way to go
While artificial intelligence (AI) holds a lot of promise to help streamline clinical workflows and improve patient outcomes, there is also some apprehension among physicians about its ability to operate unsupervised. There are also concerns that AI could end up making healthcare less efficient and that health systems are implementing these advanced algorithms without ensuring they are effective. So when the U.S. Department of Health and Human Services (HHS) called for public comments about AI in late 2025, the agency received a mix of reactions; most were cautiously optimistic, but some responses highlighted the potential downsides to these technologies.
HHS received more than 7,300 comments received from to medical societies, health systems, medical device companies and associations, and individuals. The American College of Cardiology (ACC) and the Society for Cardiovascular Angiography and Interventions (SCAI) both shared detailed thoughts about this topic.
The ACC's opening statement summarized the concerns of most health systems and medical groups.
"AI-enabled systems provide well-established benefits by leveraging computing power to enhance patient care. However, the deployment of new tools also comes with great risks that could just as easily put patients in harm’s way and threaten the clinician-patient relationship that has been the central tenet of medicine," ACC President Christopher Kramer, MD, said in the ACC's response.
HHS outlined its ambitions to harness the transformative potential of AI in clinical care, a stated goal in the Trump Administration. The department was looking for public feedback on the actions it can take to establish a forward-leaning, industry-supportive and secure approach to accelerate the adoption and use of AI as part of clinical care, but needs to better understand the barriers to larger scale adoption. For example, while the U.S. Food and Drug Administration (FDA) has cleared more than 1,200 AI algorithms for direct clinical care, adoption has been slow.
The response from the SCAI Advocacy Committee AI Task Force outlined numerous barriers:
• Medicare and Medicaid reimbursements are lacking for the vast majority of FDA-cleared AI. Payment policy needs to recognize that AI creates new interpretive work for physicians rather than eliminating existing work. Payment reductions such as the "efficiency adjustment" in the 2026 Physician Fee Schedule are based on assumptions that AI replaces professional effort. These reductions threaten the sustainability of outpatient practices, particularly in rural and underserved areas, SCAI said.
"Payment and programmatic structures should not incentivize AI systems that disintermediate clinicians or weaken the patient–provider relationship. Alignment should favor AI that demonstrably enhances clinician-supervised care," SCIA added in its letter
• Data fragmentation where patient data remains siloed across EMR systems, claims databases, imaging archives and research repositories.
• Regulatory and liability ambiguity that does not clearly delineate responsibility when AI influences a clinical decision that results in patient harm. This ambiguity discourages adoption, SCAI said.
• Acting on opaque algorithmic recommendations is fundamentally at odds with how professional clinical judgment is exercised.
• Workflow integration is a issue. AI systems are often designed without adequate understanding of clinical workflows. Tools that generate excessive alerts, require parallel documentation, or interrupt established care pathways create friction rather than efficiency. Poorly integrated AI could compromise rather than enhance patient safety.
• The absence of recognized validation standards is a major concern. SCAI noted that every licensed professional must demonstrate competency through examination, supervised training and ongoing certification. AI, SCAI argued, should be held to that same standard.
• Privacy and data use concerns and also present. SCAI said the potential for third-party monetization, re-identification through metadata aggregation and the secondary use of protected health information without clear consent boundaries erodes the trust necessary for responsible data sharing.
ACC calls autonomous AI use into question
The ACC added additional concerns over HHS interest in the use of autonomous AI in patient care. Many clinicians do not feel this technology is ready to be regularly used in healthcare.
"Currently the U.S. healthcare system does not have the infrastructure to support autonomous AI. Therefore, the systems established for implementation in the near future are best based in assistive AI or collaborative intelligence, with the clinician as the responsible entity, using AI as a tool," Kramer wrote.
He noted that AI is trained on large amounts of data, deriving insights from patterns, averages and variations across populations. While this can help show meaningful trends, it does not take into account the specific traits or clinical context of each unique patient.
"While the college is acutely aware AI capabilities will continue to evolve, it is essential that any policy frameworks must reaffirm that clinicians—not algorithms—remain at the center of patient care," Kramer explained. "While AI is a powerful tool, it serves to support clinical decision-making, not replace it."
AI requires regulatory clarity and flexibility
The ACC letter also called for regulatory flexibility to distinguish between AI-enabled healthcare applications that actively participate in the practice of medicine and other AI systems that assist in the practice of medicine through administrative simplification or other essential tasks.
"Future AI systems may blur the line between regulated and non-regulated activities. Any actions should allow agencies sufficient regulatory flexibility to provide clear guidance to clinicians, developers and stakeholders, so they are clear on what is and is not regulated. Regulatory ambiguity hampers innovation, promotes distrust in systems, and results in the development of fewer applications of this promising technology," Kramer wrote.
He added that the federal government needs to align any AI-related statutes, regulations or policies between its various agencies so that they are on the same page. This includes HHS, FDA, the Federal Trade Commission (FTC), Medicare and Medicaid Services (CMS), and the Assistance Secretary for Technology Policy/Office of the National Coordinator for Health IT (ASTP/ONC).
At the same time, the growing patchwork of state-based AI policies poses challenges for consistent innovation and implementation. Divergent requirements across jurisdictions can create uncertainty for developers, increase administrative complexity for clinicians and delay the safe and equitable deployment of AI tools.
The ACC also said funding needs to be provided for comparative effectiveness research and to delineate real-world case studies to guide clinicians and patients. Clinical adoption of new techniques, drugs and devices are all driven by clinical data and comparative effectiveness studies. In the absence of this data, it is unclear if health systems will adopt embrace AI.
This has been a big issue with the majority of AI algorithms not seeing wide adoption, and is a primary reason why they are not reimbursed. The notable, rare exceptions to this in cardiology are use of computed tomography fractional flow reserve (CT-FFR) and CT coronary plaque analysis. Both AI technologies were pushed forward by companies that understood the need for clinical data. In both instances, this hard work eventually led to category 1 CPT codes for reimbursement.
Dave Fornell has covered healthcare for more than 17 years, with a focus in cardiology and radiology. Fornell is a 5-time winner of a Jesse H. Neal Award, the most prestigious editorial honors in the field of specialized journalism. The wins included best technical content, best use of social media and best COVID-19 coverage. Fornell was also a three-time Neal finalist for best range of work by a single author. He produces more than 100 editorial videos each year, most of them interviews with key opinion leaders in medicine. He also writes technical articles, covers key trends, conducts video hospital site visits, and is very involved with social media. E-mail: [email protected]