Longitudinal Biomarker Trajectories by Cohort / Outcome — plot_dynamic

Plots mean trajectory over time with standard error bars, stratified by outcome or treatment cohort (e.g. Survived vs. Died), supporting arbitrary color scales.

Usage

plot_dynamic_trajectory(
  data,
  value_col,
  time_col,
  group_col,
  facet_col = NULL,
  colors = c(Survived = "#6F6F6F", Died = "#9D2235"),
  base_size = 12,
  palette = NULL,
  group_levels = NULL,
  ribbon_type = c("se", "quantile", "iqr"),
  quantile_probs = c(0.025, 0.975),
  central_tendency = c("mean", "median")
)

Arguments

data: Long-format data frame containing longitudinal measurements.
value_col: Character name of measurement column.
time_col: Character name of time point column.
group_col: Character name of cohort/grouping column.
facet_col: Optional character name of domain or biomarker column to facet wrap.
colors: Named or unnamed character vector of colors matching group levels, or `NULL` to use default CBE palette. Defaults to c(Survived = "#6F6F6F", Died = "#9D2235"). When `colors` is left at its default and the data's groups are not Survived/Died, the CBE palette is used instead of failing, so any grouping works without extra arguments.
base_size: Base font size (default: 12).
palette: Optional colors for the groups, used instead of `colors`. A **named** vector must name every group (extra names are ignored), so a group keeps its color whichever subset of groups is plotted. An **unnamed** vector is assigned to the groups in level order and must be at least as long as the number of groups. Anything short of that is an error, not a recycled color.
group_levels: Optional character vector giving the groups and their order (legend and color assignment). Every group in the data must be listed; list the levels in the order you want. By default a factor keeps its levels (unused ones are dropped) and other columns are sorted. Use this for multiclass phenotypes, tertiles (`c("T1", "T2", "T3")`) or cluster labels.
ribbon_type: How the spread around the central line is drawn: `"se"` (default; error bars of the standard error of the mean, as before), `"quantile"` (a ribbon between the empirical `quantile_probs` of the values at each time point) or `"iqr"` (a ribbon from the 25th to the 75th percentile). Quantiles use [stats::quantile()] type 7. A time point with one value has no spread.
quantile_probs: Two probabilities (lower, upper) for `ribbon_type = "quantile"`; default `c(0.025, 0.975)`.
central_tendency: `"mean"` (default) or `"median"`, the line and points. With `ribbon_type = "se"` the error bars are centred on the line, so with `"median"` they are median plus or minus the standard error of the mean; prefer a quantile ribbon with the median.

Value

A ggplot2 object.