A university hospital in Rome has restructured how it responds to suspected heart attacks, routing nearly every emergency room ECG through an AI model that flags dangerous patterns within seconds of the recording being taken. Sant’Andrea University Hospital, one of two university hospitals affiliated with Sapienza University of Rome, published results from a six-month deployment of the PMcardio platform’s “Queen of Hearts” AI-ECG model in an October 7, 2026 case study released by Powerful Medical, the Slovakia-founded company behind the technology.
The Diagnostic Gap the AI Is Designed to Close
The project targets a well-documented blind spot in emergency cardiology: standard STEMI criteria, based on measuring ST-segment elevation on an ECG, miss a meaningful share of occlusive myocardial infarctions — heart attacks caused by a blocked artery that still need urgent reperfusion even though the ECG doesn’t meet textbook STEMI thresholds. Patients with these “STEMI-equivalent” patterns can be misclassified as lower-urgency non-STEMI cases and wait longer for treatment, a gap cardiologists have worried about for years. Queen of Hearts, Powerful Medical’s AI model, is trained specifically to catch those equivalent patterns as well as classic STEMI, holding both FDA De Novo marketing authorization and FDA Breakthrough Device Designation.
How the System Was Built Into Daily Workflow
Sant’Andrea began evaluating the AI in its emergency department in February 2026, and the project grew from a narrow test into an integrated pathway linking emergency medicine, cardiology, and the catheterization lab. Every eligible 12-lead ECG is now analyzed automatically: recordings flow from the hospital’s Cardioline ECG infrastructure straight into PMcardio with no manual upload step, and any high-risk result is pushed directly to on-call cardiologists through the hospital’s existing Ascom DECT communication system. Patients flagged with STEMI-equivalent or occlusive-MI patterns can be escalated through the same urgent pathway as classic STEMI cases — though hospital staff emphasized that the AI only supports the decision, with cath lab activation remaining a clinical judgment call.
The Numbers From Six Months of Real Use
Between February 13 and August 13, 2026, the hospital screened 5,629 patients through the platform, generating 138 STEMI alerts that the case study reports at 96% accuracy, though the release doesn’t specify exactly how that figure was calculated. Notably, 69% of those alerts were generated automatically without a clinician initiating the analysis, and STEMI-equivalent patterns — the very cases that standard criteria would likely have missed — made up 64% of all alerts. The hospital reported a median time of under 17 seconds from alert generation to a clinician opening the AI report, and a median door-to-balloon time (the clinical benchmark for how quickly a blocked artery gets reopened) of 56 minutes. Eighty-four clinicians used the pathway over the six-month period.
Voices From the Cardiology Team
“Previously, an ECG could be acquired on time but still had to wait for a physician to review it. Now it is analyzed immediately and the cardiology team is automatically alerted,” said Prof. Emanuele Barbato, Director of Cardiology at Sant’Andrea, who also sits on Powerful Medical’s Scientific Advisory Board and has co-authored validation studies of the Queen of Hearts model — a dual role worth noting when weighing his enthusiasm. He described the tool’s effect on catching subtle cases bluntly: “nearly no fish is escaping from the net.” Dr. Marta Belmonte, an interventional cardiologist at the hospital, offered a more measured framing, saying classic STEMI criteria alone aren’t enough to identify every patient with an occluded artery, and that the AI provides “an extra layer of support regardless of the reader’s experience” — while stressing that “the final decision to take the patient to the cath lab remains clinical.” Viktor Jurasek, Powerful Medical’s chief product officer and co-founder, framed the deployment as a template for how clinical AI should work generally: “Clinical AI delivers the greatest value when it becomes part of the care pathway,” rather than a separate tool clinicians have to remember to open.
Reading the Results With Caution
It’s worth noting plainly that this case study comes from the vendor itself, publicized on Powerful Medical’s own newsroom rather than through an independent, peer-reviewed outcomes study — the 96% accuracy figure and the door-to-balloon benchmarks haven’t been published in a journal with outside statistical review, at least not yet. Powerful Medical does point to prior independent validation, including more than 25 peer-reviewed studies behind the Queen of Hearts model and a comparable deployment at Belgium’s AZORG Hospitals, where the company says PMcardio cut heart attack treatment delays by an average of 48 minutes. Still, outside cardiologists have generally welcomed AI-ECG tools cautiously, noting that automation bias — clinicians over-trusting an algorithm’s output — remains a real risk in high-stakes, fast-moving emergency settings, even when the tool’s intent is to add a safety net rather than replace clinical judgment.
What’s Next
Sant’Andrea’s six-month data point follows Queen of Hearts’ FDA De Novo authorization, announced by Powerful Medical on September 15, 2026, and a separate October 1, 2026 partnership with ambulance-monitor maker corpuls to extend the same AI into prehospital care — meaning paramedics, not just emergency room physicians, could eventually see AI-flagged ECG results before a patient even reaches the hospital. With more than 150 hospital deployments and over 2.5 million ECGs analyzed on the platform, according to the company, the next real test will be whether independent, multi-center outcome data — not vendor case studies — confirms that faster alerts actually translate into fewer deaths and less heart damage at scale.
Photo: Stephen Andrews / PEXELS via Pexels
