Loading
Obesity, Physical Activity, Diet · Journal article
European Journal of Pediatrics · August 6, 2026
Encouraging direction, but not yet definitive.
This cross-sectional study applied unsupervised machine learning (DDRTree) to 7-day thigh accelerometry in 91 children with obesity and identified three PA phenotypes with significantly different metabolic profiles. The Higher Activity phenotype showed clinically meaningful improvements in fat mass, fat-free mass, insulin sensitivity, and metabolic risk score compared to the Sedentary phenotype, independent of MVPA duration alone.
Cross-sectional observational study. Children and adolescents with obesity aged 6–17 years (43% girls).. Intervention: PA phenotypes identified from 7-day thigh-mounted accelerometry data using DDRTree. n = 91.
Three PA phenotypes identified: Sedentary (24%), Light Activity (55%), Higher Activity (21%) Only 3 of 91 participants met MVPA recommendations Higher Activity vs Sedentary: 122 min/day less sedentary time, 16.5 min/day more MVPA (p < 0.001)
Safety was not reported in the material analysed. Check the source before drawing any conclusion about harm.
These results suggest that phenotype-based stratification of PA behavior using multidimensional accelerometry features may better capture metabolic risk heterogeneity in pediatric obesity than MVPA duration alone, and may inform more targeted intervention design. However, cross-sectional design prevents causal inference and the findings require replication in independent cohorts before clinical implementation.
A sound cross-sectional analysis using unsupervised machine learning on accelerometry to identify clinically meaningful PA phenotypes and their metabolic associations in children with obesity, but limited by single time-point design, small sample, and lack of a comparator arm.
As stated by the source record.
Quoted from the source exactly as published.
These results suggest that phenotype-based stratification of PA behavior using multidimensional accelerometry features may better capture metabolic risk heterogeneity in pediatric obesity than MVPA duration alone, and may inform more targeted intervention design. However, cross-sectional design prevents causal inference and the findings require replication in independent cohorts before clinical implementation.
Graded across the dimensions that decide whether you should act, each from what the source actually supports. There is no single score, and where a dimension was not assessed it says so.
Abstract Pediatric obesity is associated with early cardiometabolic risk, yet MVPA alone may not capture meaningful variation in movement behavior. We applied DDRTree to accelerometry data to identify multidimensional physical activity phenotypes in children and adolescents with obesity and assessed their associations with body composition and cardiometabolic. This cross-sectional study included 91 children and adolescents with obesity (43% girls), aged 6–17 years, who wore a thigh-mounted accelerometer for 7 days. Multidimensional accelerometry features were extracted and PA phenotypes identified using DDRTree, an unsupervised reversed graph-embedding method that maps high-dimensional data onto a low-dimensional branching manifold. DDRTree identified 3-PA phenotypes. Anthropometry, metabolic outcomes, and a continuous metabolic syndrome risk score were compared across phenotypes using ANCOVA adjusted for age and sex. DDRTree identified three PA phenotypes: Sedentary (24%), Light Activity (55%), and Higher Activity (21%); only three participants met MVPA recommendations. Compared with the Sedentary Pattern, the Higher Activity Pattern accumulated 122 min/day less sedentary time and 16.5 min/day more MVPA ( p < 0.001) and showed lower fat mass percentage, higher fat-free mass percentage, higher insulin sensitivity (SPISE), and a lower metabolic risk score ( p ≤ 0.05). Conclusion: Multidimensional PA phenotypes derived from accelerometry identify clinically meaningful differences in body composition and metabolic risk beyond MVPA duration, supporting phenotype-based approaches for pediatric obesity risk stratification and intervention design. What is Known? • Most children and adolescents with obesity do not meet moderate-to-vigorous physical activity recommendations and are at increased risk of adverse cardiometabolic outcomes. • Physical activity volume alone may not fully capture health-relevant movement behaviour; sedentary accumulation, light activity, activity fragmentation, and transitions between activity states may also be associated with metabolic health. What is New? • Multidimensional analysis of thigh-worn accelerometry data identified three distinct physical activity phenotypes, Sedentary, Light Activity, and Higher Activity, even though only three participants met recommended moderate-to-vigorous physical activity levels. • The Higher Activity phenotype was associated with lower fat mass percentage, higher fat-free mass percentage, greater insulin sensitivity, and a lower metabolic risk score than the Sedentary phenotype, supporting assessment beyond moderate-to-vigorous physical activity duration alone.
Taken from the source record, never inferred. Follow any of these and new work involving them reaches your briefing.