Knowledge-Enhanced Electronic Profile Review (KEEPER)

KEEPER is part of HADES.

Introduction

An R package for reviewing patient profiles for phenotype validation.

Features

• Extracts patient level data for a) a random sample of patients in a cohort or b) patients in a user-specified list and formats it according to the KEEPER principles.

• Supports review of patient profiles by humans through an interactive Shiny app.

• Supports review of patient profiles by large language models.

Examples

keeperConceptSets <- generateKeeperConceptSets(
  phenotype = "Gastrointestinal bleeding",
  client = ellmer::chat_openai_compatible(),
  vocabConnectionDetails = connectionDetails,
  vocabDatabaseSchema = "cdm"
)

keeper <- generateKeeper(
  connectionDetails = connectionDetails,
  cdmDatabaseSchema = "cdm",
  cohortDatabaseSchema = "results",
  cohortTable = "cohort",
  cohortDefinitionId = 1234,
  sampleSize = 100,
  removePii = TRUE,
  phenotypeName = "Gastrointestinal bleeding",
  keeperConceptSets = keeperConceptSets
)

keeperTable <- convertKeeperToTable(keeper)

Technology

Keeper is an R package.

Installation

1. See the instructions here for configuring your R environment, including Java.

2. Install Keeper from GitHub:

   install.packages("remotes")
   remotes::install_github("ohdsi/Keeper")

Technology

KEEPER is an R package.

System Requirements

Requires R (version 4.1.0 or higher).

User Documentation

Documentation can be found on the package website.

PDF versions of the documentation are also available:

• Vignette: Generating KEEPER profiles for review
• Vignette: Using Keeper with Large Language Models
• Package manual: Keeper manual

Support

• Developer questions/comments/feedback: OHDSI Forum
• We use the GitHub issue tracker for all bugs/issues/enhancements

Contributing

Read here how you can contribute to this package.

License

Keeper is licensed under Apache License 2.0.

Development

Keeper is being developed in R Studio.

Development status

Ready for testing. Interface may still change in future versions.

Acknowledgements

• None