An AI that triages the deadliest heart attacks in the ER just won the FDA's rare tougher pathway

A 55-year-old arrives at the ER with crushing chest pain, and the EKG looks borderline. In most US emergency departments, that borderline tracing triggers a cascade: activate the cath lab, mobilize the team, and try to reopen the blocked artery in under 90 minutes. By Powerful Medical's accounting, the false-positive cath-lab activation rate runs from roughly 15% to 35%, depending on the case mix.

The FDA's newest decision targets that tradeoff. On September 11, STAT+ reported that the agency issued a de novo classification to Powerful Medical's PMcardio STEMI algorithm, nicknamed Queen of Hearts, an AI that reads 12-lead EKGs to flag the deadliest kind of heart attack, the ST-segment elevation myocardial infarction, plus a broader pattern called occlusion MI, in which the artery is shut but the classic STEMI criteria don't show up.

De novo is not the FDA's everyday route. Most AI devices reach the market through 510(k) clearance, where a manufacturer shows a new product is "substantially equivalent" to an already-cleared device. De novo requires more clinical evidence: the device has no clear predicate, and the agency writes a new regulatory classification. Fewer than 10 such AI clearances happen in a typical year, according to STAT+.

Powerful Medical had been building toward this. In March 2025, the company announced FDA Breakthrough Device Designation for the same algorithm. Breakthrough status is the agency's fast-track lane for devices that could offer a meaningful advantage over existing options for irreversibly ill patients. The de novo authorization is the next step, the actual marketing clearance.

The model is cleared to triage chest-pain patients at risk of STEMI or OMI on a 12-lead ECG. It is not a screening tool for structural heart disease, not a general diagnostic chatbot, not a replacement for the clinician. As the company's product page describes it, the algorithm is designed to flag the patterns that need a cath lab, and, just as importantly, the patterns that probably don't.

The supporting evidence is what makes the de novo pathway meaningful. At TCT 2025, the Transcatheter Cardiovascular Therapeutics meeting, late-breaking science presentations reported a 1,032-patient retrospective cohort at primary-PCI-capable centers in which the AI-ECG workflow hit a sensitivity of 92% against 71% for standard processes, and a specificity of 81% against 29%. The American College of Cardiology's summary of the session framed the gain as a meaningful reduction in false-positive cath-lab activations. The PubMed-indexed OMI accuracy study and a separate JACC: Advances paper on OMI-specific AI performance add the peer-reviewed backbone.

Trade-press coverage confirmed the milestone. Cardiovascular Business and Medical Device Network both reported the authorization in the same window, situating it as a category marker for ECG-AI: most cleared AI in cardiology runs on 510(k) predicates, and a de novo clearance raises the evidence bar.

The TCT 2025 numbers are retrospective, single-system, and conference-reported; the primary publication should be confirmed before any of these figures is treated as load-bearing in a clinical decision. EKG-only triage has documented real-world sensitivity gaps, including in early presentations and in patients with conduction abnormalities. The clearance is for a specific indication, not a category-wide endorsement of cardiology AI. And clinician override is non-negotiable: the algorithm is meant to inform the team's call, not make it. The false-positive cost the AI is meant to reduce, unnecessary cath-lab activation with its anxiety, radiation exposure, and resource use, is the entire reason the de novo bar was worth crossing.

The FDA clears AI devices regularly. The news this week is the pathway: de novo says the agency required enough clinical evidence to write a new regulatory class for this device, and that the device had no clean predicate to lean on. Whether real-world deployment replicates the TCT 2025 specificity gains is the next thing to watch.