Heatmap of binary feature prevalence by cluster

Shows, for each cluster, the percentage of participants with each binary feature (for example, a marker beyond a clinical cutoff, a diagnosis, or a medication), alongside the whole cohort for reference. Use it to describe phenotypes in clinical terms ("share with albuminuria") rather than by means or z-scores, or to compare outcome prevalence across phenotypes.

Usage

PlotClusterPrevalence(
  data,
  ClusterVar,
  variables,
  positive = "Yes",
  Relabel = TRUE,
  show_cohort = TRUE,
  digits = 0,
  title = "Prevalence by cluster"
)

Arguments

data

A data frame.

ClusterVar

Name of the cluster column in data.

variables

Binary variables to show (factor, character, logical, or 0/1 numeric).

positive

Value counted as present for factor and character variables. Default "Yes". Logical TRUE and numeric 1 are always counted as present.

Relabel

Logical; if TRUE (default), rows use variable labels.

show_cohort

Logical; if TRUE (default), add a column for all participants with a cluster.

digits

Decimal places for the cell percentages. Default 0.

title

Plot title.

Value

A ggplot object. The underlying summary is attached as the "prevalence" attribute.

Details

Percentages use participants with a non-missing value for that feature within each cluster, so features measured in only some participants or cycles are summarized over those who have them. Cells are shaded by prevalence; the cell text gives the percentage and, in parentheses, the number of participants with a value.

Examples

df <- data.frame(
  Cluster = rep(c("A", "B"), each = 50),
  high_bp = rep(c("Yes", "No", "Yes", "No"), times = c(40, 10, 5, 45)),
  anemia = rep(c(TRUE, FALSE, TRUE, FALSE), times = c(5, 45, 20, 30))
)
attr(df$high_bp, "label") <- "Systolic BP >= 130 mmHg"
attr(df$anemia, "label") <- "Anemia"
PlotClusterPrevalence(df, "Cluster", c("high_bp", "anemia"))