This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| rms (R package) | |
|---|---|
| Name | rms |
| Author | Frank E. Harrell Jr. |
| Developer | Vanderbilt University |
| Latest release | 2024 |
| License | GPL |
rms (R package)
rms is a comprehensive R package for regression modeling, validation, calibration, and prediction, widely used in clinical research, epidemiology, and biostatistics. It provides tools for model specification, assessment, and graphical presentation, integrating techniques for survival analysis, logistic regression, and multivariable modeling with a focus on robust inference and reproducible research.
rms supports regression modeling strategies that emphasize transparent model building and validation, aligning with practices promulgated by Frank E. Harrell Jr., Vanderbilt University, Journal of the American Medical Association, New England Journal of Medicine, and regulatory guidance from U.S. Food and Drug Administration. It implements functions for design matrices, penalized splines, model validation, and nomogram creation used in studies published in The Lancet, BMJ, Circulation, Annals of Internal Medicine, and Lancet Oncology. The package is part of the broader R ecosystem alongside packages like survival (R package), ggplot2, nlme, lme4, and caret (R package) used by researchers at institutions including Harvard University, Johns Hopkins University, University of Oxford, and Stanford University.
Development of rms traces to methodological work by Frank E. Harrell Jr. at Vanderbilt University Medical Center and builds on earlier software practices from projects like S-PLUS, R Project for Statistical Computing, and the statistical traditions of Harvard School of Public Health and Mayo Clinic. The package evolved through collaborations involving contributors from Duke University, University of Pennsylvania, University of California, San Francisco, and Massachusetts General Hospital. Major releases were informed by benchmarks and reviews in venues such as Biostatistics (journal), Statistics in Medicine, and presentations at conferences like the Joint Statistical Meetings and UseR! Conference.
rms provides facilities for specifying and fitting regression models with functions that interoperate with R's modeling infrastructure and other packages developed at Vanderbilt University. Key features include creation of design matrices, handling of continuous predictors with restricted cubic splines, and generation of clinical prediction tools such as nomograms and calibration plots used in publications from European Society of Cardiology, American Heart Association, and Society for Clinical Trials. The package offers validation techniques like bootstrap and cross-validation used in reports in The BMJ and JAMA Oncology, and visualization utilities compatible with gridGraphics, lattice, and ggplot2 for high-quality figures in journals like Nature Medicine.
rms implements multivariable regression approaches including linear regression, logistic regression, and Cox proportional hazards models derived from methodology in textbooks by Frank E. Harrell Jr. and contemporary statistical theory from authors affiliated with University of Washington and University of Chicago. It supports penalized regression, spline-based modelling informed by the work of Frank Harrell, Trevor Hastie, Robert Tibshirani, and Jerome H. Friedman, and model selection and shrinkage informed by literature from Bradley Efron and David Cox. The package integrates discrimination measures, calibration statistics, net reclassification improvement used in cardiovascular research at Brigham and Women's Hospital and oncology studies at Memorial Sloan Kettering Cancer Center.
Typical workflows with rms begin with data setup using design functions influenced by practices from S-PLUS and R Project for Statistical Computing, proceed through model development and validation echoing recommendations from CONSORT and TRIPOD statements, and culminate in deployment of prediction tools used in clinical guidance from World Health Organization and European Medicines Agency. Users often combine rms with data manipulation packages from RStudio ecosystems such as dplyr, tidyr, and visualization stacks used at The R Consortium and taught in courses at Johns Hopkins Bloomberg School of Public Health.
rms is compatible with core R packages such as stats (R package), survival (R package), and graphics systems including ggplot2 and lattice. It interfaces with modeling and reporting tools used at institutions like Vanderbilt University, Yale University, and Columbia University. Extensions and complementary packages used alongside rms include Hmisc, rmsb, pec, and modeling workflows supported by knitr, rmarkdown, and environments like RStudio for reproducible reports and reproducible research endorsed by National Institutes of Health and Wellcome Trust funded projects.
Case studies using rms include cardiovascular risk prediction models developed by investigators at Framingham Heart Study, oncologic prognostic models from Memorial Sloan Kettering Cancer Center and Mayo Clinic, and survival analyses in epidemiologic research from Centers for Disease Control and Prevention collaborations. Clinical prediction rules and nomograms produced with rms appear in guidelines and consensus documents from American College of Cardiology, European Society for Medical Oncology, National Institute for Health and Care Excellence, and specialty journals such as Journal of Clinical Oncology and European Heart Journal. Researchers at Stanford University School of Medicine and University of Cambridge have published applied tutorials demonstrating rms workflows integrated with machine learning pipelines influenced by work at Google Research and Microsoft Research.
Category:R (programming language) packages