In osorensen/metagam:

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

metagam

metagam is an R-package for flexible meta-analysis of generalized additive (mixed) models (GAMs/GAMMs).

The package is under development, so changes to the interface can be expected. Suggestions for improvements and bug reports are warmly welcome, either by filing an Issue or opening a Pull Request.

Installation

Install the current development version of metagam from GitHub with:

# install.packages("devtools")
devtools::install_github("osorensen/metagam")

Example

metagam has two main functions: prepare_meta and metagam. To illustrate them, we start by simulating three datasets. We need the to load the mgcv package in order to fit GAMs.

library(metagam)
library(mgcv)
## simulate three datasets
set.seed(123)
datasets <- lapply(1:3, function(x) gamSim(scale = 3, verbose = FALSE))

First we fit a separate GAM to each dataset. In the typical application of metagam, this will be done separately in the location of each dataset. The function prepare_meta removes all subject-specific data from the model object, so that it can be shared to a common location while protecting privacy.

## fit a generalized additive model to each dataset separately
fits <- lapply(datasets, function(dat){
  ## Full fit using mgcv
  gamfit <- gam(y ~ s(x0) + s(x1) + s(x2), data = dat)
  ## Extract the necessary components for performing a meta-analysis
  ## This removes all subject-specific data
  prepare_meta(gamfit)
})

Finally we meta-analyze the fits. To this end, we need to define a grid over which to make predictions. In this case, we are interested in inference for the term s(x2).

grid <- data.frame(x0 = 0, x1 = 0, x2 = seq(from = 0, to = 1, by = .01))
meta_analysis <- metagam(fits, grid = grid, terms = "s(x2)", intercept = FALSE)

Methods for visualizing and summarizing metagam objects will be added during further package development. At the moment, scripts for plotting have to be supplied by the user. The code chunk below plots the meta-analytic fit as a solid line and the separate fits as dashed lines.

plot_data <- meta_analysis$prediction

plot(plot_data$x2, plot_data$fit, type = "l", xlab = "x2", ylab = "fit", ylim = c(-6, 6))
for(fit in fits){
  lines(plot_data$x2, predict(fit, newdata = grid, type = "iterms", terms = "s(x2)"), lty = 2)  
}