Cardiovascular Risk Assessment via Sleep Patterns and ECG-Based Biological Age Estimation

Background: Understanding the intricate relationship between sleep quality and cardiovascular outcomes opens new avenues for risk stratification in cardiovascular diseases (CVDs). This study aims to evaluate the prognostic potential of biological age estimates derived from sleep-stage analysis and nocturnal heart rhythm patterns. Methods: Using polysomnographic data from 1149 patients, we extract ECG signals and use an unsupervised clustering approach to generate time-series clusters that capture dynamic fluctuations in heart rhythms. A subsequent deep learning model then estimated individual biological ages from these clusters, revealing associations between the predicted age, sleep patterns, and cardiac function. Results: In an independent test set of 736 patients, the predicted biological age was significantly associated with increased mortality (Hazard Ratio [HR] 2.27, p < 0.05) and elevated CVD risk (HR 3.56, p < 0.001), while models based solely on nocturnal heart rhythms yielded HRs of 2.29 (p < 0.05) for all-cause mortality and 3.13 (p < 0.01) for CVD risk. Conclusions: These findings demonstrate that integrating sleep stage and ECG offers a robust biomarker for cardiovascular risk stratification, paving the way for earlier interventions and more personalized healthcare strategies.