Uyku sırasında alınan EKG’ler uzun vadeli kalp riskini gösterebilir

Electrocardiograms are routinely recorded during overnight sleep studies, but their information is rarely used beyond monitoring the test. An NIH-supported team developed a deep learning model that combines single-lead ECG signals with expert-labelled sleep stages to estimate ten-year cardiovascular risk. The system was refined with records from 15,809 patients and evaluated on two independent hospital datasets containing 9,810 and 12,576 people. Higher model scores corresponded to greater long-term risk, and the signal remained informative after researchers accounted for age, sex, body mass index, diabetes, hypertension and features such as sleep-apnea severity. It predicted atrial fibrillation, heart failure and death from any cause, although further optimisation is needed for heart attacks and strokes. Because a single electrical lead can be collected with a simple patch, the approach might eventually turn existing sleep tests into a practical form of early cardiovascular screening. Researchers must still establish whether it improves conventional risk assessment in clinical use.