Accelerometry-derived activity phenotypes discriminate metabolic risk in pediatric obesity

Pediatric obesity is associated with early cardiometabolic alterations; however, moderate-to-vigorous physical activity (MVPA) duration alone may insufficiently describe behaviorally relevant variability. To identify multidimensional physical activity (PA) phenotypes in children and adolescents with obesity using discriminative dimensionality reduction via learning a tree (DDRTree) applied to accelerometry data, and to examine associations with body composition and cardiometabolic risk beyond MVPA. This cross-sectional study included 91 children and adolescents with obesity (39 girls; 43%), aged 6-17 years. Participants wore a thigh-mounted accelerometer for 7 consecutive days. A multidimensional set of accelerometry-derived features was extracted. DDRTree identified PA phenotypes. Anthropometry, insulin sensitivity indices, and a continuous metabolic syndrome risk score were compared across phenotypes using Analysis of Covariance (ANCOVA) adjusted for age and sex. DDRTree identified three PA phenotypes: Sedentary Pattern (24%), Light Activity Pattern (55%), and Higher Activity Pattern (21%). Only 3 participants (3.3%) met MVPA recommendations. Compared with the Sedentary Pattern, the Higher Activity Pattern accumulated approximately 122 min/day less sedentary time (p