Overview | Step 1: Fit the survival sub-model with survivalSub() | Step 2: Fit the longitudinal sub-models with longitudinalSub() | Optional: imputing interrupted follow-up before fitting | A diffusion-based alternative | Propagating imputation uncertainty with poolLongitudinalSub() | Step 3: Predict event risk with predictRisk() | Step 4: Predict a future biomarker value with predictLongitudinal() | Choosing bandcount1, bandcount2, bandcount3 | Friendly error messages | Where to go next

Overview | Terms that are always safe | Explicit polynomials and transforms | Interactions | Terms that are safe only in a continuous biomarker's long_sub_fixed | Orthogonal poly(), splines, and factor() | What goes wrong elsewhere | Precomputed time transforms with survivalTrans() | Putting it together: a spline term through to prediction