This week’s episode of “WisBusiness: the Podcast” is with Dr. Sean McCormick, CEO of Atrility.
This Madison-based company is developing an AI platform for better cardiac patient monitoring, based on its AtriAmp device. The tool is used to improve diagnosis of arrhythmias for post-operative cardiac surgery patients while they’re in the intensive care unit.
“Now we’re using some of the information and data that we’re getting through that device to build some software and AI models that will really expand our impact,” he said.
McCormick provides technical insights on the device, which was invented by UW-Madison Prof. Nicholas Von Bergen, a doctor in the university’s Division of Cardiology. By providing more clear, high-quality data on the patient’s heart, the AtriAmp can help care teams identify abnormal heart rates following surgery.
“As you can imagine, we have very clear electrical signals from the heart, these are literally wires directly from the heart,” he said. “Since all ECGs, or all electrocardiograms, are based upon electrical signal from the heart, that data that we get through that wire is extremely high-fidelity and clear information. The foundation of building any good AI model is good data.”
Using the heart “waveform” data from the device, the company is training AI models to automatically detect arrhythmias in patients, both for post-surgery care and elsewhere. This has implications for cardiac monitoring in other patients as well being tracked with simple surface leads.
“Our data will be good enough, we believe, to be able to help even clean up that signal, to improve arrhythmia diagnosis for all patients in the hospital, not just heart patients,” McCormick said.
He discusses how better arrhythmia identification can help improve the picture for patients, noting the AI models can even help predict the likelihood of future arrhythmias, giving care providers a chance of preventing them from happening.
The podcast also details the path ahead for the company, including software development and prototyping of AI models as well as fundraising.
“The early proofs of concept for the AI models will really be dependent upon the amount of data we have and some of what the early signs show in terms of accuracy for the predictive models and other things that we’re doing,” he said. “We’re going to do those proofs of concept in early 2026, probably some additional ones later in the year with more advanced AI techniques and use cases.”
Listen to the podcast below, sponsored by UW-Madison.
