Programa del congreso

Obstructive sleep apnea (OSA) is a prevalent sleep disorder linked to cardiovascular risk, with significant clinical heterogeneity in its clinical manifestations. Traditional metrics such as the apnea-hypopnea index often fail to capture this complexity. This study introduces a novel phenotyping methodology using subject-based weighted correlation networks and heart rate variability (HRV) features derived from polysomnography. Data from 2,641 adults of the Sleep Heart Health Study were analyzed, yielding 21 HRV metrics from time and frequency domains. Modularity analysis of the HRV-based correlation networks identified three distinct OSA phenotypes without prior assumptions on the number of clusters to explore. Statistical analysis confirmed significant differences in clinical, anthropometric, and sleep-related parameters among these 3 subgroups. Concretely, the first cluster comprised middle-aged individuals with mild OSA, higher diastolic blood pressure (DBP), and intermediate anthropometric measures. The second cluster included older individuals with mild OSA, narrower neck circumference, shorter stature, lower weight, and reduced total sleep time. The third cluster featured younger individuals with moderate OSA, fragmented sleep, higher DBP, heavier weight, and reduced slow-wave sleep duration. By leveraging inter-subject HRV-based relationships, this approach addresses the drawbacks of traditional clustering methods and highlights the potential of HRV alongside network analyses for uncovering clinically relevant OSA phenotypes. Future work should incorporate longitudinal data and other physiological signals to enhance clinical applicability.