SuperLearning of child growth trajectories

  • Empirical Model

To describe variability in longitudinal growth trajectories.


Most work with HAZ, but some with HTCM. General approach can be applied to any longitudinally measured anthropometric measure. R package available on github.


Predictors used were study specific: all baseline and time varying covariates available in the study were typically included, unless the predictor was known to be highly correlated with the outcome of interest. In addition, the following summaries of observed child-specific growth measurements were used as predictors: total number of measurements available and mean / median / SD / min / max of available growth measurements.