Part 3 – Artificial Intelligence in the Cath Lab: A Pragmatic Current-State Analysis for the Practicing Interventional Cardiologist
© 2026 HMP Global. All Rights Reserved.
Any views and opinions expressed are those of the author(s) and/or participants and do not necessarily reflect the views, policy, or position of Cath Lab Digest or HMP Global, their employees, and affiliates.
Jacob McAuliffe, MD is a cardiology fellow at Inova Schar Heart & Vascular in Falls Church, Virginia. He completed his residency training at Eastern Virginia Medical School. Prior to medical school, he completed a Master of Science & Technology Policy degree through the Consortium for Science, Policy, and Outcomes and worked as a Health Policy Fellow in Washington, D.C., cultivating an interest in advocacy, governance and responsible integration of emerging medical technologies into clinical practice. His scholarly and clinical interests converge at the intersection of interventional cardiology, health policy, and the responsible deployment of AI-enabled clinical tools.
Disclosures: Dr. McAuliffe reports no conflicts of interest regarding the content herein.
Jacob McAuliffe, MD, can be contacted at Jacob.McAuliffe@inova.org
PART 3 – Where Will AI Take Us?
The first two parts of this series established a working vocabulary, introduced important foundational concepts for appraising AI, and reviewed examples of AI-empowered clinical tools already in daily cath lab use across the full procedural workflow. This article will conclude the series by shifting focus toward future considerations regarding the impacts of AI within the field of interventional cardiology.
V – A Profession at an Inflection Point, Without a Clear Roadmap
The historical context of our field is worth considering. The introduction of coronary angiography did not simply add a new technique to the cardiologist’s toolkit; widespread use of the technology reorganized the epistemology and practice of the field. Coronary angiography fundamentally changed what it meant to know something about a patient’s coronary anatomy, and changed how that clinical knowledge was generated. The stethoscope, the electrocardiogram, and advanced imaging modalities like CMR each represent not merely new tools, but technologies that have previously unlocked and continue to unlock new realms for scientific inquiry and clinical understanding. So too will this new generation of AI applications fundamentally reshape the fabric of clinical knowledge.1
The current AI movement carries that same quality of fundamental reorganization, one that evokes a renaissance- or enlightenment-level paradigm shift. As with other instances of transformative technology such as the release of the personal computer or the smartphone, the pace of adoption is explosive, outpacing the practical education of those already encountering generative AI features in clinical settings.
The vast majority of medical schools still lack an integrated AI curriculum. Cardiology fellowship training does not include formal instruction in algorithmic validation, training data bias, or the distinctions between AI architectures that would allow a trainee to critically evaluate newly proposed AI-empowered tools. Interventional cardiologists are learning in the same way most physicians are, through conference presentations, industry relationships, journal publications of variable quality, and direct experience with products whose limitations may not be prominently disclosed. This is in lieu of insightful societal guidance and in a space largely devoid of regulatory clarity.
This is not a failure of individual initiative. It is a structural reality, and it carries a serious implication. When physicians are not engaged during the design, validation, and implementation of clinical AI tools — to a widespread degree, not just a select set of industry-appointed liaisons — the priorities that shape those tools are not always clinical priorities. Training datasets might be assembled from institutionally convenient data rather than from representative patient populations. Performance metrics risk being optimized for regulatory submission and billing capture rather than real-world clinical utility. Failure modes get characterized in controlled validation environments that may bear limited resemblance to the actual complexity of a cath lab managing a hemodynamically compromised ST-elevation myocardial infarction patient.
As introduced in the first part of this series, a cautionary tale can be seen in the state of our nation’s EMR landscape. The fractured, commercially driven landscape of electronic health records in the United States exists from how the technology was first incentivized, deployed, and entrenched. This enterprise was driven more by the promise of EMRs to serve as enhanced billing and documentation platforms, rather than for the core purpose of serving as a clinical knowledge repository and care infrastructure. The resulting EMR landscape has functioned as a persistent barrier to an integrated, interoperable patient data ecosystem so many of us in the field desire (i.e., a universal health record). The fragmented state of our EHR systems has unintentionally created architectural barriers limiting deployment of AI, and thus simultaneously created the potential niche for AI to help bridge and aggregate the data captured in these discrete systems. Reforming the EMR technological and commercial landscape would now demand an extraordinary expenditure of political, regulatory, and financial capital. We must avoid such a bottlenecked, entrenched ownership paradigm for any future AI application destined to become as so deeply embedded in the daily workflow of our healthcare system.
The interventional cardiology community has historically not been passive in the face of paradigm-shifting technological advance. The insistence on rigorous trial data to inform adoption of revascularization equipment and strategies, the development of operator and institutional volume standards, the creation of quality benchmarks through the American College of Cardiology’s NCDR and affiliated registries all reflect a culture of critical, evidence-based engagement with innovation. That same culture is now urgently needed in the AI space.
Resistance is neither realistic nor appropriate. Powerful, novel AI applications are coming regardless, and many will represent genuine advances in patient care. Passive adoption, using whatever your employer provides or the next vendor presents without the analytical framework to evaluate it, is not a safe alternative. It is simply a different form of failure. Active, critical engagement is the only posture that will serve our patients.
VI – Future Directions and the Questions Every Operator Should Ask
The near-term pipeline of AI applications for the cath lab is substantial. These use cases will start to push the bounds of what was once conceived of as more science fiction than plausible patient care. Augmented reality already holds the technical ability to develop personalized 3D coronary maps from angiographic data, similar to how potential mapping is done in EP cases; this model could then be projected or overlayed through a headset, allowing real-time augmented visualization of equipment in situ. Just as fluoroscopy operates in real time, augmented reality applications have the potential to live unobtrusively in the operator’s visual field, vastly expanding the degree and type of clinical data accessible during a procedure. LLMs integrated into clinical decision support platforms may soon enable operators to query a patient’s relevant clinical history out loud, then receive a natural language response in return. In effect, this would enable a voice-activated chart review agent readily accessible within the procedural environment. These use cases represent just two potential ways AI applications may transform from tools to agents within a clinical context. In the far future, an interventionalist may confer with AI agents as we would with peers for a second opinion.
Applications of AI in research are numerous and far reaching. Well beyond the scope of this article, it suffices to say that AI has the potential to greatly empower academic inquiry and revitalize our paradigm of evidence-based medicine. Federated learning architectures, enabling AI models to train across multiple institutional datasets without centralizing patient data, offer the prospect of tools with meaningfully better generalizability than those built on proprietary, single-institution or industry-curated datasets.18,19 Trial enrollment monitoring for safety and power considerations can help improve the quality of RCTs.20 Automated analysis of imaging endpoints can reduce manual measurement variability across large multicenter trials.20 Natural language processing applied to the published literature can facilitate systematic reviews and meta-analyses at a scale and speed far beyond traditional methods.21
As AI tools enter practice, more so as they are being conceived and developed, a set of questions should become reflexive for any operator. These are a few questions to help you start building your own assessment questionnaire:
• What patient population was this model trained on, and does that accurately reflect the demographics and comorbidity profile of my patients?
• What was the FDA clearance pathway, and what specific indication does it cover? (Clearance for a narrow use case does not constitute validation for broader or merely similar applications — a principle that should feel familiar to any operator experienced with the landscape of off-label device use.)
• Where does the algorithm fail, and how would I recognize failure during a live case? What are the clinical consequences of that failure mode?
• Who conducted the validation studies, and what was their financial relationship with the manufacturer?
• Does the tool use a locked algorithm — validated once — or an adaptive algorithm that continues to learn post-deployment? If the latter, what safeguards govern that ongoing learning?
These are not hostile or obstructionist questions. They are the same rigorous questions that a thoughtful interventionalist would apply to any new device, drug, or technique before integrating it into practice, and we must maintain precisely the same disciplined approach toward AI.
VII – Conclusion
Artificial intelligence is already engrained within the cath lab. It has been for years, operating under the guise of less prominent monikers such as machine learning. The current AI craze represents an acceleration of that foundation, a blooming of capacities and the start of a new chapter, not the authoring of the first.
What is genuinely new is the breadth of AI’s potential impact across the full spectrum of care delivered through the cath lab, as well as the rate at which new tools will be developed and deployed. What is not new with regard to paradigm-shifting technologies, and what merits consequential deliberation and action, is the absence of any recognized framework for preparing physicians to engage critically with the emerging generation of AI applications. We are at an inflection point in medicine analogous in character, if not in pace, to the great reorganizations of clinical knowledge that have preceded us.
Interventional cardiology has always defined itself by a willingness to push boundaries, embrace innovation, and insist on rigorous evidence for what enters practice. That same identity is now the basis for a clear obligation: to understand AI not as a marketing category, but as a set of distinct tools with specific architectures, specific capabilities, and specific limitations. We must bring this understanding to bear on every new platform that arrives in the lab. The operators who understand these tools will shape them. Those who don’t will be shaped by them.
References
1. Khera R, Oikonomou EK, Nadkarni GN, et al. Transforming cardiovascular care with artificial intelligence: from discovery to practice. J Am Coll Cardiol. 2024;84(1):97–114. doi:10.1016/j.jacc.2024.05.003
2. Adler-Milstein J, Jha AK. HITECH Act drove large gains in hospital electronic health record adoption. Health Aff. 2017;36(8):1416–1422. doi:10.1377/hlthaff.2016.1651
3. Chen M, Zhang B, Cai Z, et al. Acceptance of clinical artificial intelligence among physicians and medical students: a systematic review with cross-sectional survey. Front Med. 2022;9:990604. doi:10.3389/fmed.2022.990604
4. Omiye JA, Lester JC, Spichak S, et al. Large language models propagate race-based medicine. NPJ Digit Med. 2023;6:195. doi:10.1038/s41746-023-00939-z
5. Yang Y, Liu X, Jin Q, Huang F, Lu Z. Unmasking and quantifying racial bias of large language models in medical report generation. Commun Med. 2024;4:202. doi:10.1038/s43856-024-00601-z
6. U.S. Food and Drug Administration. Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan. January 2021. Accessed June 9, 2026. Available at: https://www.fda.gov/media/145022/download.
7. Hanneman K, Playford D, Dey D, et al. Value creation through artificial intelligence and cardiovascular imaging: a scientific statement from the American Heart Association. Circulation. 2024;149(6):e296–e311. doi: 10.1161/CIR.0000000000001202
8. Armoundas AA, Narayan SM, Arnett DK, et al. Use of artificial intelligence in improving outcomes in heart disease: a scientific statement from the American Heart Association. Circulation. 2024;149(14):e1028–e1050. doi:10.1161/CIR.0000000000001201
9. Douglas PS, De Bruyne B, Pontone G, et al. 1-year outcomes of FFRCT-guided care in patients with suspected coronary disease: the PLATFORM study. J Am Coll Cardiol. 2016;68(5):435–445. doi:10.1016/j.jacc.2016.05.057
10. Patel MR, Nørgaard BL, Fairbairn TA, et al. 1-year impact on medical practice and clinical outcomes of FFRCT: the ADVANCE registry. JACC Cardiovasc Imaging. 2020;13(1):97–105. doi:10.1016/j.jcmg.2019.03.003
11. Niimi N, Shiraishi Y, Sawano M, et al. Machine learning models for prediction of adverse events after percutaneous coronary intervention. Sci Rep. 2022;12:6262. doi:10.1038/s41598-022-10346-1
12. Fearon WF, Achenbach S, Engstrøm T, et al. Accuracy of fractional flow reserve derived from coronary angiography (FAST-FFR). Circulation. 2019;139(4):477–484. doi:10.1161/CIRCULATIONAHA.118.037350
13. Fearon WF, Kirtane AJ, Jeremias A, et al. Angiography-derived fractional flow reserve to guide PCI (ALL-RISE). N Engl J Med. 2026. doi:10.1056/NEJMoa2600949
14. Abbott Vascular. Ultreon™ software for OCT intravascular imaging. Abbott Laboratories; 2025. Accessed June 9, 2026. Available at: https://www.cardiovascular.abbott/us/en/hcp/products/percutaneous-coronary-intervention/intravascular-imaging/ultreon-software.html
15. Consolo F, Tessari M, Morici N, et al. The intelligent Impella: future perspectives of artificial intelligence in the setting of Impella support. ESC Heart Fail. 2024. doi:10.1002/ehf2.14865
16. Deisenhofer I, Albenque JP, Busch S, et al. Artificial intelligence for individualized treatment of persistent atrial fibrillation: a randomized controlled trial (TAILORED-AF). Nat Med. 2025;31(4):1286–1293. doi:10.1038/s41591-025-03517-w
17. Olson KD, Meeker D, Troup M, et al. Use of ambient AI scribes to reduce administrative burden and professional burnout. JAMA Netw Open. 2025. doi:10.1001/jamanetworkopen.2025.34976
18. Teodoro D, Naderi N, Yazdani A, et al. A scoping review of artificial intelligence applications in clinical trial risk assessment. NPJ Digit Med. 2025;8:384. doi:10.1038/s41746-025-01886-7
19. Rahman MS, Karmarkar C, Islam SMS. Application of federated learning in cardiology: key challenges and potential solutions. Mayo Clin Proc Digit Health. 2024;2:590–595. doi:10.1016/j.mcpdig.2024.09.005
20. Olawade DB, Fidelis SC, Marinze S, et al. Artificial intelligence in clinical trials: a comprehensive review of opportunities, challenges, and future directions. Int J Med Inform. 2026;206:106141. doi:10.1016/j.ijmedinf.2025.106141
21. Pilz M, Zimmermann S, Friedrichs J, et al. Semi-automated title-abstract screening using natural language processing and machine learning. Syst Rev. 2024;13:265. doi:10.1186/s13643-024-02688-w
22. Narula J, Stuckey TD, Nakazawa G, et al. Prospective deep learning-based quantitative assessment of coronary plaque by computed tomography angiography compared with intravascular ultrasound: the REVEALPLAQUE study. Eur Heart J Cardiovasc Imaging. 2024;25(9):1287–1295. doi:10.1093/ehjci/jeae115
