US regulators have authorized an artificial intelligence model designed to detect heart attacks directly from electrocardiogram results, according to a report published this week by Cardiovascular Business. The system, known as Queen of Hearts, is an AI-powered ECG platform developed by medical technology company Powerful Medical, and its clearance marks another step in the quiet migration of machine learning from research papers into the emergency departments where minutes determine outcomes. For more on AI in medicine, follow our latest AI developments.
According to the trade publication's September 4 report, the FDA authorization covers the model's use in detecting heart attacks from ECG results — putting the tool in the company of a small but growing set of AI systems that clinicians can legally fold into cardiac triage workflows.
The evidence behind the authorization
The authorization follows a major safety study led by UC Davis Health, published in JACC: Cardiovascular Interventions and presented at the 2025 Transcatheter Cardiovascular Therapeutics conference, that tested the Queen of Hearts platform against standard triage at three US hospitals. The results were striking: across more than 1,000 patients suspected of suffering the most severe form of heart attack, the AI outperformed the existing process on both ends of the ledger — catching more true heart attacks while generating dramatically fewer false alarms.
The study reviewed records from emergency departments at UC Davis Medical Center and two other geographically diverse US hospitals, using ECGs collected between January 2020 and May 2024. Follow-up testing confirmed that 58 percent of the suspected patients were genuinely experiencing a STEMI — an ST-elevation myocardial infarction, the type of heart attack caused by a fully blocked artery. The remaining 42 percent were false alarms.
On the initial ECG, the AI model correctly identified 553 of the confirmed cases, compared with 427 detected through traditional methods. Just as important, it reduced false positives to roughly 8 percent, against nearly 42 percent under standard triage.
Why minutes matter in a STEMI
STEMI is among the most time-critical diagnoses in medicine. The definitive treatment is percutaneous coronary intervention, a procedure that restores blood flow through the blocked artery — and the window is brutally short. When treatment takes longer than 90 minutes, the risk of death is about three times higher, according to the UC Davis researchers.
That arithmetic is what makes triage accuracy so consequential. Every false activation of the cardiac catheterization lab consumes resources and, at hospitals without redundant teams, can delay care for the next patient. Every missed STEMI risks the worst outcome of all.
"These results demonstrate the potential of AI-based applications to transform emergency cardiovascular care," said Bryn Mumma, a professor of emergency medicine at UC Davis Health and the study's primary investigator, at the time of publication. Her team concluded that using the model at first medical contact could shorten time to treatment and reduce false activations of emergency protocols.
A tool, not a replacement
Mumma and her colleagues were careful to frame the platform's role. The AI, she noted, should be interpreted with caution and must serve as a support tool rather than a substitute for clinical judgment — a caveat that applies with equal force now that the model carries regulatory authorization.
The clearance also lands in a regulatory environment that is actively warming to medical AI. The FDA has spent recent months soliciting public input on generative AI in medical devices while clearing a steady stream of algorithms for imaging, screening and monitoring tasks. Heart attack detection from a routine 12-lead ECG — a test performed millions of times a day with equipment already installed in every hospital and ambulance — is among the highest-leverage applications, because it requires no new hardware, only better interpretation of data clinicians already collect.
If the Queen of Hearts platform's trial performance carries into routine use, the arithmetic suggests a meaningful dent in both missed STEMIs and wasted cath-lab activations. The study's own authors, for now, keep the framing conservative: technology works best when it works alongside clinicians, not instead of them.
Part of a broader FDA pattern
The authorization also fits a clear pattern at the agency. Earlier this year, the FDA cleared HeartLung's opportunistic imaging platform, which screens routine chest CT scans for cardiovascular and other conditions patients weren't being scanned for — the same "find disease in data already collected" logic that makes an ECG-based heart attack detector so attractive. And the agency has an open docket soliciting public input on how generative AI should be regulated in medical devices, signaling that the current run of clearances is a prelude to a more formal framework rather than an ad hoc sprint.
For Powerful Medical, the practical effect is commercial: US hospitals can now evaluate Queen of Hearts not as a research curiosity but as an authorized device in the cardiac triage pathway. For the dozens of other teams training models on ECG archives, it sets a precedent worth studying — clinical evidence from a multi-site safety study, published in a major cardiology journal, followed by regulatory authorization.
Stay Ahead of AI
Follow the AI tools moving from lab to clinic. Read more AI news every day.
Read more AI news →
