CIRSE Insider: What do you see as the biggest potentials and risks of AI’s application to IR? How could the government and other regulatory institutions regulate AI to mitigate these risks?
Lojo Lendoiro: The greatest potential of AI in IR lies in improving precision, efficiency, safety, and personalization of care. AI may help optimize patient selection, support procedural planning, reduce radiation exposure, predict complications, and streamline repetitive tasks that currently consume valuable clinical time.
However, the risks are equally important. Overreliance on algorithms, lack of transparency, biased datasets, unequal access to technology, cybersecurity concerns, and unclear legal responsibility are all major challenges. There is also the danger of adopting technologies faster than we can properly validate them clinically.
Regulation should therefore focus on transparency, explainability, independent clinical validation, continuous post-market surveillance, and clear accountability frameworks. Importantly, regulation should not only evaluate whether an algorithm technically works, but also whether it safely integrates into real clinical workflows and benefits patients equitably.