Performs comprehensive ROC analysis for single or multiple continuous indicators and custom combination models, including AUC calculation, optimal cutoff determination, and DeLong test for AUC comparisons.
Usage
roc_analysis(
data,
outcome,
predictors,
combined_models = NULL,
output_dir = NULL,
save_format = c("none", "plot", "data", "all"),
delong_test = FALSE,
colors = NULL,
legend_labels = NULL,
seed = 123,
plot_width = 2000,
plot_height = 2000,
plot_res = 300,
direction = "<"
)Arguments
- data
Data frame containing outcome and all predictor variables
- outcome
Character scalar, binary outcome variable name (0/1)
- predictors
Character vector, continuous predictor variable names
- combined_models
Named list of combined models, e.g., list(Combined = c("X1", "X2"))
- output_dir
Output directory for saving results, default NULL
- save_format
Save format: "none", "plot", "data", "all", default "none"
- delong_test
Logical, whether to perform DeLong test for pairwise AUC comparisons, default FALSE
- colors
Custom color vector for ROC curves; if NULL, colors are automatically sampled
- legend_labels
Named vector for custom legend labels, e.g., c(PIV = "Systemic Inflammation Index")
- seed
Random seed for reproducibility, default 123
- plot_width
Plot width in pixels, default 2000
- plot_height
Plot height in pixels, default 2000
- plot_res
Plot resolution in DPI, default 300
- direction
roc direction, "auto" for automatic selection, default
<