Digital biomarkers of pediatric metabolic health in children with obesity: insights from wearable-derived heart rate data

Objectives: Pediatric obesity is a major public health concern associated with early development of metabolic syndrome (MetS) and autonomic imbalance. Wearable devices enable continuous, non-invasive monitoring of physiological signals under real-life conditions, but their use to investigate metabolic health in youth remains limited. We examined the association between wearable-derived heart rate (HR) measures and metabolic status in children with obesity, exploring the potential of wearable-based digital phenotyping to identify early markers of autonomic and metabolic dysregulation. Methods: Fifty children and adolescents with obesity (8-16 years) were recruited. Participants wore a Fitbit Charge 2 for continuous HR monitoring. HR indices, including All-day HR, Sleeping HR, Resting HR, Inactive HR, Minimum HR, and HR dip, were computed over ≥5 valid days (i.e., with ≥20 h of wear time). MetS was classified using Gurka's sex-specific criteria. Logistic regression models tested associations between HR indices and MetS status, with stepwise feature selection to identify the most informative predictors. Results: All-day and Inactive HR were significantly higher in Not-MetS participants compared with those with MetS (p<0.05). Stepwise logistic regression identified Sleeping HR (p=0.0435) and HR dip (p=0.0332) as significant predictors of MetS. Both showed inverse associations (odds ratio <1), indicating that lower Sleeping HR and a smaller nocturnal HR dip were related to higher odds of MetS. Conclusions: Fitbit-derived HR metrics, particularly Sleeping HR and HR dip, are associated with metabolic status in children with obesity. Continuous wearable monitoring can support early, personalized, and preventive strategies in pediatric cardiometabolic health through digital phenotyping of autonomic function.