Identification of Risk Groups in Pharmacovigilance Using Penalized Regression and Machine Learning (RGP)

RGP is an R package for analyzing healthcare claims data and simulated data using penalized regression and machine learning methods. This package contains function wrappers to create a simulated cohort, group predictors based on functional targets (from KEGG and TTD) and conventional groups (ATC/ICD systems) and analyze the data using various types of penalized regression (LASSO) and machine learning methods (random forests and block forests).

Functionalities

1. cohort simulation (R/sim_create_cohort.R)

2. functional target-based grouping - KEGG (R/ftarget_db_manager.R)

3. functional target-based grouping - TTD (R/ftarget_db_manager.R)

4. penalized regression, group-based analysis and results assessment (R/rgp_grpl.R)

5. classification measures (R/rgp_classification_measures.R)

Structure

• data:
  • ttd_clean_data: data curated from TTD (drug-target, disease-target.. etc.) as binary matrices
  • Paper13_events.csv: file containing ICD list of the ADEs for the project
• R: functions and packages used in the analysis
  • algorithms-task13.R: analysis methods wrappers functions
  • helpers.R: helper functions
  • ftsim: Functional Target Simulation; a full independent package to download, clean and manage TTD and KEGG data. It also use them to create a simple simulated cohort with grouped covariates.
• README.md: We are here. We explain the project contents, used packages, and of course the directory structure.
• analysis:
  • analysis scripts from 2-8. Run one by one on slave nodes. Steps to obtain healthcare claims data from health insurance databases are omitted for data protection reasons.
  • code to generate descriptive statistics and read performance metrics
  • script to define study variables (inclusion criteria that applies to GePaRD)

See the documentation ? and ? for more info.

Installation

devtools::install_github("bips-hb/rgp")

Usage

Acknowledgements

We gratefully acknowledge the financial support from the innovation fund (“Innovationsfonds”) of the Federal Joint Committee in Germany (grant number: 01VSF16020).

Contact

Mariam R. Rizkallah
Leibniz Institute for Prevention Research & Epidemiology - BIPS GmbH E-mail: rizkallah-issak [at] leibniz-bips [dot] de