Will AI replace cardiologists?
Artificial intelligence (AI) is getting used more and more by cardiologists in addition to becoming one of the hottest topics at cardiology meetings. As AI-based tools increasingly assist with image analysis, risk assessment, clinical decision-making and procedures, a question has emerged: Could the technology eventually replace cardiologists?
Sanjiv Narayan, MD, MBChB, MSc, an American Heart Association volunteer and co-founder of the Stanford Arrhythmia Center at Stanford Medicine, sees AI as a major inflection point for cardiovascular care. But he does not expect it to eliminate the need for physicians. He shared his views in a video interview with Cardiovascular Business.
Narayan said AI is augmenting cardiologists and will allow them to accomplish more, potentially expanding their capabilities and improving the efficiency and success of care.
“It's allowing us to do more and expand the scope and improve success with what we have,” he said.
He pointed to radiology as an example. Radiologists have worked alongside remote interpretation services and computational AI tools for years, yet the specialty has continued to grow.
“People predicted the end of radiology 10 to 15 years ago,” he said. “There are now more radiologists and it is flourishing.”
Narayan expects a similar trajectory in cardiology. Straightforward cases may increasingly be handled with greater automation, but complex cases will continue to require physicians who can integrate imaging and other data with the patient's clinical circumstances.
The value of AI, he said, may be particularly apparent in helping clinicians avoid overlooking important information. AI systems can potentially identify relevant risk factors, incorporate newly published guidelines and provide additional insights that may be difficult for a physician to recall or process during a busy clinical day.
“I think it will just be an enhancement so that things that we forget because we're human and we're tired won't be forgotten,” Narayan said.
AI is already becoming part of cardiovascular diagnostic and decision-making workflows. Tools such as fractional flow reserve derived from coronary CT angiography (FFR-CT) and automated coronary plaque analysis illustrate how computational technologies can provide additional information to clinicians.
Narayan expects the use of such technologies to continue expanding and becoming increasingly integrated into clinical guidelines and practice.
However, he said physician oversight remains critical. Current limitations of AI, including errors and hallucinations produced by large language models, make fully autonomous clinical care difficult to envision.
“I don't think it will replace us,” Narayan said. “We just want someone to be accountable, and we want that someone to be somebody who's seen patients for many years.”
Rather than replacing cardiologists, Narayan expects AI to change how they work. The physicians who effectively incorporate AI into their practices may be able to handle more information, answer clinical questions more efficiently and better navigate increasingly complex procedures.
“AI-based tools are here to stay and will grow,” he said. “I can see there will be changes in the way we do our jobs.”
The future of cardiology, therefore, may be less about humans vs. AI and more about how effectively physicians can work with these advanced technologies. Narayan said clinical guidelines to continue emphasizing clinician oversight of AI-supported decisions, preserving the physician's role while allowing technology to expand what cardiologists can accomplish.