Skip to content

OHDSI/RiskStratifiedEstimation

main
Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
R
 
 
 
 
 
 
man
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

RiskStratifiedEstimation

Introduction

RiskStratifiedEstimation is an R package for implementing risk stratified analyses for the assessment of treatment effect heterogeneity in an observational database in the OMOP Common Data Model. The package combines functionality of PatientLevelPrediction and CohortMethod R packages.

The figure below illustrates the overall process of performing a risk-based analysis of treatment effect heterogeneity using the RiskStratifiedEstimation package. (A) Starting from a treatment (top), a comparator (bottom) and an outcome (middle) cohort we estimate the propensity scores on the entire target population. (B) We match patients on the propensity scores and estimate the prediction model. Since we match patients we develop the prediction model on smaller subset of the initial population and, therefore, the number of patients is smaller in B compared to A. (C) We apply the prediction model on the entire population (green: lower 25% of the risk distribution; yellow: patients with risk between 25% and 50% of the risk distribution; orange: patients with risk between 50% and 75% of the risk distribution; red: patients at risk higher than 75% of the risk distribution). (D) We separate in risk subgroups, here quarters. Within risk quarters propensity scores are estimated again and relative and absolute treatment effects are estimated.

CohortMethod propensity score plot

Features

  • Extracts the necessary data from a database in OMOP Common Data Model format.
  • Uses a large set of covariates for the prediction, the propensity and the outcome model, including for example all drugs, diagnoses, procedures, as well as age, comorbidity indexes, etc.
  • Imports functionality from PatientLevelPrediction package for the development of prediction models for risk stratification.
  • Large scale regularized regression to fit the propensity and outcome models.
  • Includes diagnostic functions for the prediction process, including calibration plots and ROC curves demostrating the performance of the developed prediction models in sub-populations of interest.
  • Includes diagnostic functions for the estimation process, including propensity score distribution plots and plots showing covariate balance before and after performing a propensity score-based analysis.

Screenshots

CohortMethod propensity score plot CohortMethod covariate balance plot
Comparison plot across risk strataCovariate balance plot

Technology

RiskStratifiedEstimation is being developed in R.

System Requirements

Requires R (version 3.1.0 or higher). Installation on Windows requires RTools. Libraries used in RiskStratifiedEstimation require Java.

Dependencies

  • Cyclops
  • DatabaseConnector
  • SqlRender
  • FeatureExtraction
  • ParallelLogger
  • CohortMethod
  • PatientLevelPrediction

Getting Started

  1. On Windows, make sure RTools is installed.
  2. The DatabaseConnector and SqlRender packages require Java. Java can be downloaded from http://www.java.com. Once Java is installed, ensure that Java is being pathed correctly. Under environment variables in the control panel, ensure that the jvm.dll file is added correctly to the path.
  3. In R, use the following commands to download and install RiskStratifiedEstimation:
install.packages("devtools")
devtools::install_github("OHDSI/RiskStratifiedEstimation")

Getting Involved

License

RiskStratifiedEstimation is licensed under Apache License 2.0

Development

RiskStratifiedEstimation is being developed in R Studio.

RiskStratifiedEstimation is ready for use.

About

An R package for performing effect estimation in risk strata to assess treatment effect heterogeneity

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages