Ai-based Ecg Interpretation Outperforms Standard Diagnosis Of Occlusive Myocardial Infarction

Artificial intelligence (AI)-based ECG interpretation outperformed modular pathways for nan discovery of occlusive myocardial infarction (MI), according to a study presented today at ESC Acute CardioVascular Care 2026, the yearly legislature of the Association for Acute CardioVascular Care (ACVC), a branch of the European Society of Cardiology (ESC). 

In patients with suspected acute coronary syndrome (ACS), a circumstantial alteration connected an ECG, called an ST elevation, is an parameter that nan patient may person an occlusion in a coronary artery. This type of bosom onslaught is known as an ST-elevation myocardial infarction (STEMI) and it requires immediate percutaneous coronary intervention to restore the heart's blood flow. In patients who do not person an ST elevation, nan origin of nan thorax pain can be less certain and further tests are needed to corroborate if nan MI is owed to an occlusion. 

Many patients without an ST elevation have an occlusive MI, but it tin beryllium difficult for clinicians to quickly and accurately recognize this, leading to delays in providing emergency treatment. We investigated whether AI-based interpretation of the initial ECG could improve the accuracy of detecting occlusive MIs successful nan absence of an ST elevation to optimize diligent management." 

Presenter, Doctor Federico Nani from Central Hospital Bolzano, Italy

This single-centre prospective study included 1,490 patients who had symptoms suggestive of ACS but without an ST elevation on the first ECG. The mean property was 63 years and 42% were female. Clinicians interpreted nan first ECG, tested levels of nan cardiac biomarker troponin, and performed coronary angiography, when required, to diagnose occlusive MI based connected ESC Guidelines. In parallel, nan first ECG was interpreted by a smartphone-based CE-certified AI-ECG algorithm. 

AI-based ECG interpretation ruled out occlusive MI in 1,382 patients and detected it in 108 patients (7%). The AI-based method correctly identified obstructive MI in 84% of cases. Sensitivity was 77%, specificity was 99% and the negative predictive value was 98%. There were 27 mendacious negatives (2%) and 17 false positives (1%). 

According to nan modular diagnostic pathway, occlusive MI was ruled retired successful 1,207 patients based connected troponin levels and 283 patients underwent coronary angiography to confirm or exclude the diagnosis. Overall, human ECG-interpretation correctly identified occlusive MI successful 42% of cases. 

Doctor Nani concluded: "This simple, accessible AI-based approach demonstrated superior accuracy successful identifying and excluding occlusive MI compared with conventional diagnostic pathways in patients without an ST elevation. The results of our single-centre study require further validation, but these findings propose that AI ECG interpretation is a valuable addition to existing decision-making devices to improve early recognition and timely, effective treatment." 

The powerfulness of AI to support nan guidance of cardiovascular illness will beryllium further explored arsenic nan spotlight taxable of this year's yearly ESC Congress, taking spot from 28–31 August 2026 successful Munich, Germany.