README.md
In romainkp/LtAtStructuR: Structuring of Complex Longitudinal Data into Long Format

R/`LtAtStructuR`

Structuring of Complex Longitudinal Data into Long Format

Authors: Romain Neugebauer, Noel Pimentel, and Nima Hejazi

What’s `LtAtStructuR`?

The goal of LtAtStructuR is to structure a collection of time-stamped measurements (e.g., electronic health record data) into a standard long format analytic dataset suitable for the evaluation of the effect of multiple time-point interventions in the presence of time-dependent confounding or sources of selection bias.

Download the LtAtStructuR directory and install the package from the command line using

R CMD INSTALL LtAtStructuR

or from within R using the package tarball (.tar.gz) with

install.packages("PATH/TO/LtAtStructuR_version.tar.gz", type = "local")

where the path to the package tarball and version number must be amended. A tarball may be generated by invoking Rscript -e "devtools::build()" from the root directory of this project.

How to use the package? (Description of input/output)

What data coarsening algorithms are implemented by the package?

This is a basic example which shows how to structure longitudinal data with an interval exposure:

library(LtAtStructuR)
library(data.table)
library(lubridate)
library(future) # optional (for parallel processing)
plan(multiprocess) # optional (for parallel processing)

## Define one cohort dataset, one exposure dataset, and one or more covariate
## datasets
cohort <- setCohort(cohortDT, "ID", "IndexDate", "EOFDate", "EOFtype",
                    "AMI", c("ageEntry", "sex", "race", "A1c", "eGFR"),
                    list("ageEntry"=list("categorical"=FALSE,
                                        "impute"=NA,
                                        "impute_default_level"=NA),
                        "sex"=list("categorical"=TRUE,
                                   "impute"=NA,
                                   "impute_default_level"=NA),
                        "race"=list("categorical"=TRUE,
                                    "impute"=NA,
                                    "impute_default_level"=NA)) )
exposure <- setExposure(expDT, "ID", "startA", "endA")
covariate1 <- setCovariate(a1cDT, "sporadic", "ID", "A1cDate", "A1c",
                           categorical = FALSE)
covariate2 <- setCovariate(egfrDT, "sporadic", "ID", "eGFRDate", "eGFR",
                           categorical = TRUE)

## Gather each input dataset into a single object that specifies the content of
## the output dataset to be constructed
LtAt.specification <- cohort + exposure + covariate1 + covariate2

## Construct the output dataset
LtAt.data <- construct(LtAt.specification, time_unit = 15, first_exp_rule = 1,
                       exp_threshold = 0.75)

LtAtStructuR can also also be used through an interactive Shiny web application that automates the process of defining the cohort, exposure, and covariate datasets. The shiny application currently only supports interval exposures.

The shiny application can be executed by running the command:

runShiny()

LtAtStructur Shiny App.

The tabs Cohort, Exposure, Covariates, and Construct are analogous to the setCohort, setExposure, setCovariate, and construct functions, respectively; each tab offers an interactive user-interace to select input criteria pertaining to each of these functions. As a final step of the application, the user can download the output dataset and R template code specific to the construction of that dataset:

LtAtStructur Shiny App -
Construct.