image: The bregr package provides a streamlined, modular workflow for batch regression modeling. The process begins with installation and initialization, followed by core modeling steps such as setting dependent and independent variables, selecting regression types (e.g., linear, logistic, Cox), and fitting models using functions like br_set_y(), br_set_x(), and br_run(). The package supports batch processing of multiple models with parallel computing and robust error handling. Results can be extracted and visualized using tidy outputs, forest plots, risk network plots, and subgroup analyses. The entire workflow is modular, reproducible, and designed for scalability, making bregr a powerful tool for large-scale regression modeling in biomedical datasets.
Credit: Shixiang Wang , Yun Peng , Chenyang Shu, Chunyang Wang , Yuxi Yang , Yankun Zhao , Yanru Cui , Dehua Hu , Jian‐Guo Zhou
We introduce bregr, a novel open-source R package designed to streamline batch processing and visualization of biomedical regression models. Addressing the inefficiency and reproducibility challenges of manually constructing multiple univariate and multivariate models, bregr provides a cohesive, tidyverse-style workflow. Built on the S7 object-oriented framework for extensibility, it supports diverse models—including generalized linear, Cox proportional hazards, and mixed-effects models—leveraging native R pipes, parallel computing, and robust error handling. Key features include automated model fitting for variable combinations, tidy result extraction, integrated publication-quality visualizations (e.g., forest plots), and a one-step pipeline function. Validated on TCGA cohorts, bregr enhances efficiency, scalability, and reproducibility for large-scale biomedical data analysis. The package is available on CRAN and GitHub.
Journal
Med Research
Method of Research
Data/statistical analysis
Subject of Research
Human tissue samples
Article Title
bregr: An R Package for Streamlined Batch Processing and Visualization of Biomedical Regression Models
Article Publication Date
17-Sep-2025
COI Statement
The authors declare no conflicts of interest.