An article published in Nature Medicine on March 20, 2026 describes the remote monitoring of heart failure exacerbations using a smart watch. A deep learning model uses data from smart watches to predict peak oxygen consumption. The model also predicts unplanned medical events in patients with heart failure in the prospective TRUE-HF cohort. It also predicts unplanned health care utilization in patients with heart failure in the All of Us Research Program. The study confirmed the effectiveness of this approach in real conditions. The model focuses on early detection of heart failure deterioration. The results support the use of smart watches in clinical practice.