BaselineRegularization

Introduction

Installation

The package can be installed from github with

devtools::install_github("sverchkov/BaselineRegularization", build_vignettes = TRUE)

Tutorial

The R vignettes include a tutorial that covers most features of the package, these can be accessed with

browseVignettes("BaselineRegularization")

Example

# Load the package
library("BaselineRegularization")

# Connect to Database, e.g. postgres [1]
con <- DBI::dbConnect( RPostgreSQL::PostgreSQL()
                     , host = "localhost"
                     , user = "user"
                     , dbname = "omop_example"
                     , password = rstudioapi::askForPassword("Database Password") )

# Define the Event of interest
event = 4110956 # The concept_id for "Acute myocardial infarction NOS"

# Extract relevant data
br_data <- prepareBRData( con, response_event = event )

# Parametrize task
parameters <- defineBRParameters()

# Fit model
fit <- fitBaselineRegularization( br_data, parameters )

# Show results (beta coefficients)
getCoefficients( fit )

1. See this guide for a more comprehensive overview of database connection. We use dplyr under the hood.

Technology

BaselineRegularization is an R package.

System Requirements

Requires R (version 3.0.0 or greater).

Dependencies

Always required:

• rlang (>= 0.2)
• Matrix (>= 1.2)
• dplyr (>= 0.7)
• tidyr (>= 0.8)
• glmnet (>= 2.0)
• futile.logger (>= 1.4)

Required for database access:

• DBI (>= 1.0.0),
• dbplyr (>= 1.2),
• Database drivers, depending on the database being accessed, e.g.
  • RPostgreSQL
  • RSQLite

Required for building the documentation:

• knitr
• rmarkdown

Required for testing:

• testthat (>= 2.0.0)

License

BaselineRegularization is licensed under Apache License 2.0

Development

BaselineRegularization is being developed in R Studio.

We use the GitHub issue tracker for bugs and feature requests

References

Original papers describing the underlying algorithms:

• Kuang, Z. et al. (2016). Computational drug repositioning using continuous self-controlled case series. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. ACM, pp. 491-500.
• Kuang, Z. et al. (2017). Pharmacovigilance via Baseline Regularization with Large-Scale Longitudinal Observational Data. In: Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. ACM, pp. 1537-1546.

OHDSI Symposium 2018 poster