ARISING HEALTHALIVE. AWAKE. AWARE.Read the magazines →

← November 19, 2026 edited reader edition

THE ARISING · Longevity & Healthspan

In suspected heart attack, troponin beat an AI reading of the ECG

Putting the two together improved statistical performance slightly. Better emergency care still needs to be demonstrated.

In two European groups of emergency patients with a suspected heart attack, the standard high-sensitivity troponin T blood test distinguished cases better than this AI reading of the first heart tracing.

That is the finding of a prospective diagnostic cohort study in The Lancet Regional Health – Europe. It followed 12,630 emergency patients with suspected heart attack, drawn from two European cohorts, APACE (5,047 patients) and APACE-Light (7,583 patients). Heart attacks were confirmed by central adjudication under the fourth Universal Definition of Myocardial Infarction. They were diagnosed in 19% of APACE patients (968) and 18% of APACE-Light patients (1,386).

The tool, called AiTiA MI, was built using data from Korean patients. The researchers applied it to the 12-lead ECG recorded when each patient arrived. They compared it with high-sensitivity cardiac troponin T, the BASEL score, the recalibrated HEART score and physician judgment at 90 minutes. The last two were available only in APACE.

Accuracy was measured by the area under the curve, or AUC. A score of 1.0 means perfect separation of patients with and without a heart attack. The AI-ECG scored 0.78 in APACE and 0.84 in APACE-Light. The troponin T test scored 0.94 and 0.93.

Combining the two helped a little. The AI-ECG added some diagnostic value to troponin T, lifting the AUC to 0.95 in both cohorts. The authors describe the AI-ECG's accuracy as high, but not very high.

The authors conclude that a heart attack is unlikely to be ruled out on an ECG alone. Their view is that pairing AI-ECG with the blood test may improve decision making in the emergency department. They say this warrants prospective studies.

The work was funded by the Swiss National Science Foundation, the Swiss Heart Foundation, Innosuisse, the University Hospital Basel, the University of Basel and the Korea Health Industry Development Institute. Several diagnostics companies, including Abbott, Roche, Siemens and Beckman Coulter, also provided funding.

A higher AUC is not evidence that an AI alert leads to faster correct treatment or fewer missed heart attacks. The team calls for studies that test the combination within emergency care. This particular ECG model did not outperform troponin and the results do not support using it in place of standard evaluation.

Sources

THE ARISING NEWSLETTER

Science moves forward.
Come with it.

Research, clear explanations, and a little room for wonder.
Read and follow The Arising on Substack.

Find The Arising on Substack ↗