The question
How do clinical measures change over time, and how much of that comes from the individual rather than the group? An average alone hides what’s happening in a single person.
What the data showed
Hierarchical models captured both the shared trajectory and the variation between people. The non-linear relationships between growth and clinical variables became clear, and the uncertainty was shown with credible intervals instead of tucked away.
What it could be used for
A picture of developmental trajectories across clinical subgroups, where you could see how certain the numbers were before making a decision.
Tools
R with brms, lme4, and tidybayes. Posterior predictive checks and model diagnostics. Visual communication with ggplot2 and posterior, and reproducible scripts throughout.
The same approach works whenever you want to follow a group over time and separate the general pattern from the individual — for example pupils’ progress, customer behaviour, or measurements across sites.