The heart speaks in signals. Every beat leaves an electrical trace: a waveform that a trained eye can read like a sentence, full of meaning about how a person's heart is working and where it might be failing. For a century, reading that language has been the work of clinicians, in a field where decisions have long rested on established rules, accumulated guidelines, and the practical experience of those who have read thousands of these traces before. Cardiology is, by temperament, a careful and conservative discipline, and for good reason: the stakes are measured in lives. Now, machines are learning to read this language too, and increasingly, to write it.
This talk is about what happens when we turn the attention of intelligent machines to the signals of the human heart. These are systems that recognize and interpret, learning to spot patterns a clinician might take years to master. A family of these, generative AI, is now creating images, voices, and text, and they can do something unprecedented: not just classify a recording as healthy or abnormal, but generate realistic heartbeats, fill in what noise has erased, reconstruct a clean signal from a messy one, and flag how confident they are while doing it. In a domain where a sensor on the skin captures only a faint, noisy echo of a complex biological process, these abilities turn out to be powerful.
No background in cardiology or AI is assumed. We'll see how these intelligent systems work in settings like detecting arrhythmias or stratifying the risk of sudden cardiac death, why the heart is such a compelling place to put them to the test, and what they can already do: recovering signals, quantifying uncertainty, and generalizing across very different kinds of cardiac data. We'll also look at where they help, where they mislead, and what has to be true before tools like these belong anywhere near a patient.
The bigger picture is a shift that reaches well beyond cardiology: machines are moving from recognizing patterns to generating them, and medicine is one of the first places that shift will be felt. In the heart, this story is starting to get actionable.
Speaker: Samuel Ruipérez Campillo (PhD candidate in Computer Science and Artificial Intelligence at ETH Zürich; currently Visiting Graduate Fellow at the Institute for Medical Engineering and Science at Massachusetts Institute of Technology, MIT)
LOCATION:RCCHU Conference Room STATUS:CONFIRMED DTSTART:20260714T210000Z DTEND:20260714T220000Z END:VEVENT END:VCALENDAR