MAC2: Internal function for the Meta-Analytical Calibration Curve (MAC2)

Description

Computes meta-analytical calibration curves using multiple methods (logistic regression, loess or splines) and performs meta-analysis across clusters to generate aggregated calibration curves with confidence and prediction intervals.

Usage

MAC2(
  data = NULL,
  p,
  y,
  cluster,
  grid,
  cl.level = 0.95,
  alpha.lr = 0.05/3,
  plot = TRUE,
  cluster_curves = FALSE,
  knots = 3,
  transf = "logit",
  method_choice = c("splines", "log", "loess"),
  method.tau = "REML",
  prediction = TRUE,
  random = TRUE,
  sm = "PLOGIT",
  hakn = FALSE,
  linewidth = 1,
  method.predict = "HTS",
  verbose = FALSE
)

Value

A list containing:

cluster_data: Data frame with linear predictors and standard errors for each method per cluster
plot_data: Data frame with meta-analysis results including predictions and intervals
plot: A ggplot2 object if plot = TRUE, otherwise NULL

Arguments

data: optional data frame containing the variables p, y, and cluster. If supplied, variable names should be given without quotation marks.
p: predicted probabilities (numeric vector) or name of the column in data.
y: binary outcome variable or the name of the column in data.
cluster: Cluster identifier (factor, character, or integer) or name of the column in data.
grid: the grid for the calibration curve evaluation
cl.level: the confidence level for the calculation of the confidence interval. Default is 0.95.
alpha.lr: the alpha-level used for the likelihood ratio test, selecting the number of knots for the restricted cubic splines
plot: logical, indicates whether to plot the calibration curves. Default is TRUE.
cluster_curves: logical, whether to include cluster-specific curves in the plot. Default is FALSE.
knots: integer, number of knots for splines. Default is 3.
transf: character, transformation for predictions: "logit" or "identity". Default is "logit".
method_choice: character, which method to use for meta-analysis. Options are: "log", "loess" or "splines". Default is "splines".
method.tau: character, method for between-study heterogeneity estimation. Default is "REML". This argument is passed to the metagen function.
prediction: logical, whether to compute prediction intervals. Default is TRUE. This argument is passed to the prediction argument of the metagen function.
random: logical, whether to use random-effects model. Default is TRUE. This argument is passed to the random argument of the metagen function.
sm: character, summary measure for meta-analysis. Default is "PLOGIT". This argument is passed to the sm argument of the metagen function.
hakn: logical, whether to use Hartung-Knapp adjustment. Default is FALSE. This argument is passed to the method.random.ci argument of the metagen function.
linewidth: numeric, line width for the meta-curve. Default is 1.
method.predict: character, method for prediction intervals. Default is "HTS". This argument is passed to the method.predict argument of the metagen function.
verbose: logical, indicates whether progress has to be printed in the console.

Details

This function estimates the center-specific calibration curves using logistic regression, loess or splines. Hereafter, it aggregates the calibration curves using meta-analytical techniques. The meta-analysis is performed using the function metagen from the meta package. The method_choice argument determines which method is for the meta-analytical aggregation.