11.1 Artificial Intelligence in Interventional Medicine

These proceedings summarize the educational activity of the 17th Biennial Meeting of the International Andreas Gruentzig Society held January 30 to February 2, 2024 in Chiang Rai, Thailand.

Faculty Disclosures     Vendor Acknowledgments

2024 IAGS Summary Document

Statement of the problem or issue

Artificial intelligence (AI) is actually a misnomer; it is more correct to call it augmented intelligence. It has been around for a very long time. A calculator is a form of AI, and we have had mechanical calculators for over 100 years. Nevertheless, the use of AI in medicine, cardiology, and interventional cardiology is increasing, as illustrated by publications on the subject (Figure).

Figure. Publications referencing AI in interventional cardiology.

Three broad areas of application of AI in medicine are: (1) medical image analysis; (2) genomics analysis; (3) natural language processing. For our practical purposes, it is useful to classify AI uses as listed in Table 1.

Table 1. Practical applications of AI in medicine.

On the topic of imaging, AI may offer the opportunity to have a completely objective analysis rather than a subjective analysis of angiograms, IVUS, and OCT. In fact, part of the underutilization of IVUS and OCT imaging may be that some or many operators are not comfortable interpreting the images. They are comfortable performing the procedures, but less comfortable interpreting the images, and AI may not only provide an educational tool but also increase operator comfort level and use. For many of the other areas, some clinicians are skeptical or hesitant (or downright fearful) of what has come down to them, because the products and systems were forced on them from a purely business-oriented model, or from other components of the healthcare system that either want to make money or streamline their processes at the expense of clinician-led processes. The concern here is that AI may overstep or contradict clinical judgment or test interpretation. None of these concerns would arise if AI is used appropriately.

Gaps in current knowledge

We have enormous gaps in our knowledge base regarding applications of AI in medicine, cardiology, and interventional cardiology. Some aspects are listed in Table 2.

Table 2. Knowledge gap areas.

Possible solutions and future directions

We are only at the very beginning of AI implementation in clinical practice. There is a great deal to learn. However, we must never lose sight of the fact that AI is just a tool, and like all tools it can have good or bad applications. There will always have to be human oversight and engagement. Change is always a challenge, but fear is not a productive response.

References

1. Subhan S, Malik J, Haq AU, et al. Role of artificial intelligence and machine learning in interventional cardiology. Curr Probl Cardiol. 2023;48(7):101698. doi: 10.1016/j.cpcardiol.2023.101698 PMID: 36921654.
2. Singh A, Miller RJH, Otaki Y, et al. Direct risk assessment from myocardial perfusion imaging using explainable deep learning. JACC Cardiovasc Imaging. 2023;16(2):209-220. doi: 10.1016/j.jcmg.2022.07.017 PMID: 36274041.
3. Williams MC, Bednarski BP, Pieszko K, et al. Unsupervised learning to characterize patients with known coronary artery disease undergoing myocardial perfusion imaging. Eur J Nucl Med Mol Imaging. 2023;50(9):2656-2668. Epub 2023 Apr 17. doi: 10.1007/s00259-023-06218-z PMID: 37067586.
4. Rudnicka Z, Pręgowska A, Glądys K, Perkins M, Proniewska K. Advancements in artificial intelligence-driven techniques for interventional cardiology. Cardiol J. 2024;31(2):321-341. Epub 2024 Jan 22. doi: 10.5603/cj.98650. PMID: 38247435