Bayesian Super Imposition by Translation and Rotation Growth Curve Analysis

The Super Imposition by Translation and Rotation (SITAR) model is a shape-invariant nonlinear mixed effect model that fits a natural cubic spline mean curve to the growth data and aligns individual-specific growth curves to the underlying mean curve via a set of random effects (see Cole, 2010 for details). The non-Bayesian version of the SITAR model can be fit by using the already available R package 'sitar'. Unlike the 'sitar' package which allows modelling of a single outcome only, the 'bsitar' package offers great flexibility in fitting models of varying complexities, including joint modelling of multiple outcomes such as height and weight (multivariate model). Additionally, the 'bsitar' package allows for the simultaneous analysis of an outcome separately for subgroups defined by a factor variable such as gender. This is achieved by fitting separate models for each subgroup (for example males and females for gender variable). An advantage of this approach is that posterior draws for each subgroup are part of a single model object, making it possible to compare coefficients across subgroups and test hypotheses. Since the 'bsitar' package is a front-end to the R package 'brms', it offers excellent support for post-processing of posterior draws via various functions that are directly available from the 'brms' package. In addition, the 'bsitar' package includes various customized functions that allow for the visualization of distance (increase in size with age) and velocity (change in growth rate as a function of age), as well as the estimation of growth spurt parameters such as age at peak growth velocity and peak growth velocity.

bsitar

Overview

The bsitar package provides an interface for Bayesian implementation of the Super Imposition by Translation and Rotation (SITAR) growth model. The SITAR is a shape-invariant nonlinear mixed effect model that fits a natural cubic spline mean curve and aligns individual-specific growth curves to the underlying mean curve via a set of random effects: size, timing and intensity. The bsitar package package is a front-end to the R package brms which itself uses the Stan program to performing full Bayesian inference.

Installation

To install the latest release version from CRAN use

install.packages("bsitar")

The current developmental version can be installed from GitHub as:

if (!requireNamespace("remotes")) {
  install.packages("remotes")
}
remotes::install_github("Sandhu-SS/bsitar")

The brms package can be installed from the CRAN

install.packages("brms")

The latest developmental version of brms can be downloaded from GitHub as follows

remotes::install_github("paul-buerkner/brms")

Note that the brms, and hence the bsitar too, are based on Stan, and therefore a C++ compiler is required. The program Rtools (available on https://cran.r-project.org/bin/windows/Rtools/) comes with a C++ compiler for Windows. On Mac, you should install Xcode. For further instructions on how to get the compilers running, see the prerequisites section on https://github.com/stan-dev/rstan/wiki/RStan-Getting-Started.