Artificial intelligence is no longer a future concept in echocardiography.

Presented exclusively to attendees at the American Society of Echocardiography (ASE) 2026 Scientific Sessions in Aurora, Colorado, the “Cor et Machina: AI BOOTCAMP” session, chaired by Lissa Sugeng, MD, MPH, FASE, of Yale University, explored how artificial intelligence (AI) is already being incorporated into clinical workflows while highlighting the challenges that remain before broader adoption.

Building Better AI

Minh B. Nguyen, MD, FASE, of Texas Children’s Hospital and Baylor College of Medicine, described a shift from traditional supervised learning, in which algorithms are trained to perform a single task, to self-supervised learning. He framed the newer approach as a way to “teach the model the heart, not just the answer,” allowing AI to learn the grammar of the heart from large collections of unlabeled echocardiograms before being fine-tuned for specific clinical tasks.

He described three families of self-supervised learning—compare (“same heart, different clothes”), reconstruct (“fill in the blanks”), and distill (“two views, one meaning”)—that enable models to learn from unlabeled data.

“When one model can cover many tasks…it becomes a foundation model,” Dr. Nguyen stated. “Self-supervised learning helps us build a more robust AI.”

He cautioned, however, that “a model teaching itself the patterns is not the same as teaching itself the truth,” emphasizing that new approaches must still answer three questions: Is it real? Does it help? Can we trust it?

Also during the session, Sreekanth Vemulapalli, MD, of Duke University School of Medicine, examined the current evidence supporting AI in echocardiography. He highlighted validated and emerging use cases, including image acquisition, clinical decision support applications, analysis of echocardiography reports to prompt appropriate follow-up, and analytics that use echocardiography data for population health management.

While acknowledging that much of the current evidence remains retrospective—particularly that used to inform clinical decision-making—Dr. Vemulapalli called prospective studies evaluating AI’s impact “the next frontier.” He also outlined a roadmap for implementing AI in the echo lab: preparing the lab, addressing workflow, piloting low-risk AI, adopting advanced AI, and expanding access, concluding that it is a “multistep, iterative process that has to have constant monitoring.”

Boon or Bane?

Jose Donato A. Magno, MD, of Philippine General Hospital, examined whether AI is a “boon or bane” across what he described as the “top 10 conundrums” surrounding its use in clinical echocardiography: measurement accuracy and reproducibility; workflow efficiency; disease detection and classification; generalizability and dataset bias; accessibility and point-of-care service; explainability; decision support and risk prediction; ethical, legal, and accountability issues; regulatory and clinical validation; and clinician autonomy and workforce deskilling.

Rather than arguing that AI is inherently a boon or a bane, he weighed its potential benefits against its limitations across each domain.

Expanding on the banes discussed during his presentation, Dr. Magno said in an exclusive interview with Conexiant that the greatest barrier to broader adoption is ensuring equitable access. “No matter how good the technology is, if it’s not serving the underserved, it’s not going to work,” he commented. “How do we address that? We need to put our heads together in societies all over the world, do more research on underrepresented populations, and share information—work toward a common goal.”

Dr. Magno emphasized during his presentation that “[AI] will only truly flourish if it rests on a bedrock of genuine intelligence.” He concluded, “No digital or artificial force can ever truly match the human spirit,” a message he reiterated during the interview.

Putting AI Into Practice

Karen Zimmerman, BS, ACS, RDCS, RVT, FASE, of the University of Michigan, used a clinical case to illustrate the real-world use of AI in echocardiography. Rather than advocating for AI adoption, however, she concluded by asking whether AI assistance had meaningfully changed patient management—or whether it justified the extra time, cost, and effort required.

Elsewhere in the session, Steven J. Lester, MD, FASE, of Mayo Clinic in Scottsdale, Arizona, described what he called a “missed opportunity” in cardiovascular imaging. Thousands of studies are acquired each day and typically interpreted for the reason they were ordered, he said, yet AI could help answer additional clinically relevant questions. To illustrate that vision, he presented CardioAID (Cardiology Artificial Intelligence Dashboard), showcasing AI applications for screening, diagnosis, and prognostication.

Dr. Lester suggested that the dashboard reflects a broader goal for AI in cardiovascular care, stating, “This is what AI is going to do: democratize access to health care, provide faster, more accurate diagnosis, and improve patient outcomes.”

Disclosure: Dr. Vemulapalli reported grants or contracts from the National Institutes of Health, the US Food and Drug Administration, the American College of Cardiology, the American Heart Association, Edwards Lifesciences, and Abbott Cardiovascular; advisory board or speaker’s bureau relationships with Cytokinetics, Edwards Lifesciences, Medtronic, Abbott Cardiovascular, and Eli Lilly; and that the Duke Heart Center and Duke Echo Lab have a codevelopment agreement with Us2.ai. All other speakers declared no relevant conflicts of interest at the time of presentation.

Source: ASE 2026 Scientific Sessions