Longitudinal prediction of body mass index (BMI) using explainable AI: integrating polygenic scores, maternal, early-life and familial factors
Authors' conclusion
The interpretable KAN model revealed the dynamic interplay of childhood BMI z-score and PGS emerging as key drivers of BMI trajectories across life stages. The finding underscores the potential of BMI at critical time in early childhood as a biomarker for obesity risk. Our interpretable model offers actionable insights for targeted obesity prevention strategies.
Study context
- Population
- Raine Study Gen2 cohort; BMI measured at ages 8, 10, 14, 17, 20, 23 and 27 years; predictors: maternal and paternal anthropometrics, parental education, early-life skinfold measurements and seven BMI-related polygenic scores (PGS)recruited
- Participants
- 2,868 participants
- Outcome
- Prediction of BMI trajectories from childhood to early adulthood
- Follow-up
- Ages 8 to 27 years
Methodological detailsprediction modelling with explainable AI, longitudinal, observational, Kolmogorov-Arnold Networks (KAN) alongside conventional machine learning models
- Prediction modelling with explainable AI
- Longitudinal
- Observational
- Kolmogorov-Arnold Networks (KAN) alongside conventional machine learning models
Main finding
As reported in the abstract
Predictive performance of KAN model (epidemiological + PGS data) for BMI
R² for BMI at age 8
0.81
R² for BMI at age 27
0.34
Observational data: associations, not effects.
Original publication
Longitudinal prediction of BMI using explainable AI: integrating polygenic scores, maternal, early-life and familial factors. Int J Obes (Lond). Published 16 March 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →