Synthetic continuous glucose monitoring (CGM) data and corresponding R code to fit multi-level functional Beta model that supports the following paper:

Gaynanova I, Punjabi N and Crainiceanu, C (2020+). Modeling continuous glucose monitoring (CGM) data during sleep. Biostatistics, accepted.

Description of files:

• BetaEstimatingFunctions.R - functions for fitting multi-level functional Beta model

• CGMdataSynthetic.Rdata - synthetic CGM data. Contains list of n=124 subjects with 14 night trajectories per subject over 5 min grid in 7 hour sleep period. Also contains minimal and maximal glucose values for each subject. The data have been generated based on real CGM data described in the paper following SimulateCGMdata.R script

• fPCAStability.R - code for visual display of functional PCA results for mean and standard deviation processes, as well as for evaluating stability of corresponding eigenfunctions via subsampling

• Code_for_figures.R - code for creating figures of subject-specific night trajectories, corresponding quantile bands, and minimal/maximal glucose values

• Code_for_A1C_analyses.R - code for the analysis of association between fPCA scores and A1C as described in the paper together with code for creating corresponding figures

• ComparePercentiles.R - code for comparing percentiles of functional Normal and Beta models based on simulated data

• QQplots.R - code for creating QQplots at a global, time-specific and subject-specific levels

• NaiveBands.R - code for estimation of quantiles based on separate subject/time approaches (parametric Beta and nonparametric kernel density estimators)

• Beta_model_fit_perturb.R - comparison of estimated quantiles between Beta model with estimated values of min/max, and Beta model with perturbed values of min/max

• Example_mfPCA_issues.R - code for fitting multi-level fPCA model on CGM data with code for figure that highlight issues with level 1 fits

• FPCA_methods_comparison.R - code for comparing different FPCA methods in terms of estimated mean and eigenfunctions for subjects' means and standard deviations