Artificial Intelligence in Coronary Artery Disease Diagnosis and Intervention

The integration of artificial intelligence (AI) into the diagnosis and management of coronary artery disease (CAD) is no longer just an aspiration: it is already happening. Current technology offers clinicians new tools for early detection, risk stratification, procedural planning, and post-intervention prognostication. In his Sunday morning session and ensuing panel discussion at CRT 2025, Dr Ran Kornowski of the Rabin Medical Center in Petach-Tikva, Israel, detailed how AI is transforming clinical practice and research in interventional cardiology.

Pre-Intervention Screening and Diagnosis

As all practicing cardiologists know, early detection of coronary disease is crucial to preventing severe disease presentations in the catheterization lab. “Much too often the first encounter with patients with coronary disease is in the cath lab,” Kornowski says. “The question is, can we use AI capabilities in order to diagnose upfront and not [encounter] this patient [in] the cath lab with such a severe disease?”

To that end, AI-driven imaging analysis is emerging as a powerful tool for cardiovascular screening. In a study at Rabin Medical Center involving 631 patients undergoing non-cardiac chest computed tomography scans, Nanox.AI software (Nanox) identified significant coronary calcium in one-third of the patients, leading to earlier preventive intervention in a group of patients whose symptoms would have otherwise gone undetected until later evaluation.¹ Dr Kornowski is also working with Cleerly (Cleerly, Inc.), an AI-based coronary computed tomography angiography platform that measures plaque, stenosis, and ischemia,² and has demonstrated its ability to provide detailed information about plaque and calcification before patients even arrive at the cath lab. In addition, his team is collaborating with Heartflow, a medical technology company that integrates coronary physiology with deep learning-based plaque characterization—essential diagnostical data that enhances patient stratification and informs decisions regarding interventional vs conservative management strategies.

Procedural Integration

Once the treatment course is set, AI continues to aid the interventions themselves, with myriad technologies already incorporated into cardiology procedures. For one, there is CathWorks' FFRangio. This well-validated and globally utilized AI-powered angiographic-based processing unit enables intraprocedural coronary physiology assessment, lesion impact analysis, and virtual procedural planning. Plus, by eliminating the need for pressure wires, it allows for minimal invasiveness.

The integration of AI and deep learning into intravascular ultrasound (IVUS) and optical coherence tomography represents another promising advancement already being implemented in the cath lab. AI-driven enhancements in IVUS segmentation, as well as tissue and lesion characterization, offer significant potential for improved accuracy and automation in intravascular imaging.

Another notable mention is CathAlert, a real-time AI-based system that alerts operators to potentially dangerous situations during a procedure. Ideally, this kind of technology will be expanded further for comprehensive case planning.

Post-Procedure and Prevention

Dr Kornowski described operational machine learning models capable of conducting prognostic assessments of patients following myocardial infarction and acute coronary syndrome, and which are demonstrating strong predictive accuracy for 1-year outcomes. Additionally, algorithms powered by machine learning offer personalized primary and secondary prevention of coronary atherosclerosis. AI-based technologies also enable remote monitoring of ischemia, arrhythmia, and heart failure symptoms, along with AI-integrated patient medical records for enhanced clinical management.

Challenges to Implementation

Despite its promise, AI adoption in cardiology presents significant challenges. “We have to make sure that we understand the limitation,” Dr Kornowski cautions. Prevailing issues include data quantity and quality, analytical validity, bias, overfitting, and interoperability. Additionally, privacy, security, ethical regulations, and cost considerations remain key concerns to its implementation.

Panel Discussion

A panel discussion followed Dr. Kornowski’s presentation. While we continue to speculate on how AI will transform medicine in the years to come, said panel discussion members, we cannot ignore the salient fact that much of these imagined innovations are already possible; as far as AI is concerned, future is now. As proof of concept, Dr Kornowski detailed his center’s in-house AI lab, complete with dedicated computer scientist and analyst. His team demonstrates that this is “not theoretical, this is what we are actually doing on a daily basis.” Their goal is deceptively simple: to have an impact on the patient. The challenge here is to prove that AI can change practice for the better, not just makes change for its own sake. While some positive trends are emerging, Dr Kornowski acknowledged that much is still just “hype,” and some technologies will not make the final cut.

AI Optimization

One key panel discussion point was optimizing AI utility through data sharing. As the panelists rightly note, the application of AI is correlated to the data it receives; if users continue to build AI models solely from their own institutional data sets, the output will be correspondingly individualized. Here is where federated learning and federated server models come into play. "This is the only way to go," Dr Kornowski asserted, describing a system where institutions retain raw data but share analytic outputs, fostering collaboration without compromising privacy. He noted, however, that this approach depends on cooperation, trust, and regulatory agreements among institutions.

AI in Patient Care

When asked how he envisioned the patient journey in 5 years, panelist Rafael Beyar of Rambam Hospital proposed an “AI-triage system” where patients entering busy hospitals undergo AI-guided evaluations, from electrocardiogram (ECG) to imaging, with nurses and clinicians on hand to validate the computer’s recommendations.

Dr Kornowski echoed Dr Beyar’s prediction that AI-enhanced ECG, imaging integration, and electronic medical records will become standard within this time, optimizing efficiency while allowing clinicians to dedicate more time to direct patient care. However, he cautioned that AI should augment rather than replace human expertise. No matter the technological advancements we will see in the future, robots should never be the sole caregiver, and the need for patients to be treated by “good physicians” must continue to be central to the providers’ creed.

Takeaway

As it continues to unfurl its potential at an incalculable pace, AI is poised to revolutionize the management of coronary artery disease by enhancing early detection and diagnosis, refining personalized treatment, and optimizing intervention strategies. While the extent of its influence remains to be seen, we can be certain that AI’s role in interventional cardiology will only expand, necessitating ongoing research and clinician engagement to maximize its impact on patient care.

References

1. Innovative Data Featuring Nanox AI Cardiac Solution Showcased at SCCT 2024. Nanox. July 25, 2024. Accessed March 13, 2025. https://investors.nanox.vision/news-releases/news-release-details/innovative-data-featuring-nanox-ai-cardiac-solution-showcased
2. Cleerly Comprehensive Care Management Platform. Cleerly, Inc. Accessed March 13, 2025. https://cleerlyhealth.com/heart-disease-technology?gclsrc=aw.ds&&utm_term=cleerly%20heart%20scan&utm_campaign=cleerly_nationwide_brand&utm_source=adwords&utm_medium=ppc&hsa_acc=1461201274&hsa_cam=21252425692&hsa_grp=164211503680&hsa_ad=715436316401&hsa_src=g&hsa_tgt=kwd-1678369226334&hsa_kw=cleerly%20heart%20scan&hsa_mt=e&hsa_net=adwords&hsa_ver=3&gad_source=1&gclid=Cj0KCQjwhMq-BhCFARIsAGvo0KcVz0SHxlceOqjDXQn1x6IxMt7QQr2sDUC48tCYTlTf7jS6jSYE2ukaAla8EALw_wcB