R/Medicine 2021, the premier conference for the use of R in clinical applications is less than two weeks away! This conference reflects the increasing importance of data science, computational statistics and machine learning to clinical applications, and emphasizes the effectiveness of the R language as a vehicle for making data driven medicine accessible to clinicians with diverse backgrounds. The two conference keynote talks Bringing Machine Learning Models to the Bedside by Karandeep Singh and Dissecting Algorithmic Bias by Ziad Obermeyer directly address important technical and ethical issues confronting modern data driven medicine.
R/Medicine will offer six short courses spread out over the two days of August 24th and August 25th. These courses are included in the registration price. The Tuesday short courses will be:
- Secure Medical Data Collection: Best Practices with Excel, and Leveling Up to REDCap and collaboratoR taught by Peter Higgins, Will Beasley, Kenneth McLean
- Intro to R for Medical Data taught by Ted Laderas, Daniel Chen, Mara Alexeev
- An Introductory R Guide for Targeted Maximum Likelihood Estimation in Medical Research taught by Ehsan Karim
The Wednesday short courses will be:
- Mapping Spatial Health Data taught by Marynia Kolak
- R Markdown for Reproducible Research taught by Alison Hill and Stephan Kadauke
- From SAS to R taught by Joe Korszun
The conference will proper will run from 11:00 to 19:30 EDT on Thursday, August 26th, and from 10:50 to 18:00 EDT on Friday, August 27th. The full schedule is here. Priced at only $50 full fare, $25 for academics and $10 for students and trainees, this is an affordable, important conference that you will not want to miss.
You can register here.
To get an idea of the international scope of the conference, and a feel for what the virtual conference experience might be like, have a look at the R Journal article written by the organizing team about last year’s conference: R Medicine 2020: The Power of Going Virtual.
And finally, to get an idea of the various ways that R is contributing to getting everyday work done in clinical practice: the following are the packages from the Medicine category that made it to my lists of “Top 40” new CRAN packages in posts over the past twelve months.
“Top 40” Picks for new CRAN Packages for Medicine
afdx v1.1.1: Provides functions to estimate diagnosis performance (Sensitivity, Specificity, Positive predictive value, Negative predicted value) of a diagnostic test when there is no golden standard by estimating the attributable fraction using either a logitexponential model or a latent class model.
aldvmm v0.8.4: Fits health state utility adjusted limited dependent variable mixture models, i.e. finite mixtures of normal distributions with an accumulation of density mass at the limits, and a gap between 100% quality of life and the next smaller utility value. See Alava and Wailoo (2015) for background and the vignette for an example.
babsim.hospital v11.5.14: Implements a discrete-event simulation model for a hospital resource planning. Motivated by the challenges faced by health care institutions in the current COVID-19 pandemic, it can be used by health departments to forecast demand for intensive care beds, ventilators, and staff resources. See Ucar, Smeets & Azcorra (2019), Lawton & McCooe (2019) and the website for background, and the vignette to get started.
beats v0.1.1: Provides functions to import data from UFI devices and process electrocardiogram (ECG) data. It also includes a Shiny app for finding and exporting heart beats. See README to get started.
bhmbasket v0.9.1: Provides functions to evaluate basket trial designs with binary endpoints using Bayesian hierarchical models and Bayesian decision rules. See Berry et al. (2013), Neuenschwander et al. (2016) and Fisch et al. (2015) for background and the vignette for an example.
card v0.1.0: Provides tools to help assess the autonomic regulation of cardiovascular physiology with respect to electrocardiography, circadian rhythms, and the clinical risk of autonomic dysfunction on cardiovascular health through the perspective of epidemiology and causality. For background on the analysis of circadian rhythms through cosinor analysis see Cornelissen (2014) and Refinetti et al. (2014). There are two vignettes: circadian and cosinor.
causalCmprisk v1.0.0: Provides functions to estimate average treatment effects of two static treatment regimes on time-to-event outcomes with competing events. The method uses propensity scores weighting for emulation of baseline randomization. See the vignette.
clinDataReview v1.1.0: Provides functions to create interactive tables, listings and figures and associated reports for exploratory analysis in a clinical trial setting. There are vignettes on Prerocessing, Visualization, and Creating Reports.
cmprskcoxmsm v0.2.0: Provides functions to estimate treatment effect a under marginal structure model for the cause-specific hazard of competing risk events. Functions also estimate the risk of the potential outcomes, risk difference and risk ratio. See Hernan et al. (2001) for the theory and the vignette for examples.
coder v0.13.5: Provides functions to classify individuals or items based on external code data identified by regular expressions. A typical use case considers patients with medically coded data, such as codes from the International Classification of Diseases. There is an overview and vignettes on class codes, interpreting regular expressions, and example data.
covidcast v0.4.2: Provides an interface to Delphi’s COVIDcast Epidata including tools for data access, maps and time series plotting, and basic signal processing, and a collection of numerous indicators relevant to the COVID-19 pandemic in the United States. There is a Getting Started Guide, and vignettes on Computing Signal Correlations, Combining Data Sources, Manipulating Multiple Signals, and Plotting and Mapping Signals.
epigraphdb v0.2.1: Provides access to the EpiGraphDB platform. There is an overview, vignettes on the API, Platform Functionality, Meta Functions and three case studies on SNP protein associations, Drug Targets and Causal Evidence
EpiNow2 v1.2.1: Provides functions to estimate the time-varying reproduction number, rate of spread, and doubling time using a range of open-source tools Abbott et al. (2020) for background, Gostic et al. (2020) for current best practices, and README for examples.
escalation v0.1.2: Implements methods for working with dose-finding clinical trials and includes a common interface to various dose-finding methodologies such as the continual reassessment method (CRM) by O’Quigley et al. (1990), the Bayesian optimal interval design (BOIN) by Liu & Yuan (2015), and the 3+3 described by Korn et al. (1994). There are vignettes on Working with dose-paths, Working with dose selectors, and Simulating dose-escalation trials.
eventglm v1.0.2 Implements methods for doing event history regression for marginal estimands, including cumulative incidence the restricted mean survival, as described in the methodology reviewed in Andersen & Perme (2010). See the vignette for examples.
goldilocks v0.3.0: Implements the Goldilocks adaptive trial design for a time to event outcome using a piecewise exponential model and conjugate Gamma prior distributions as described in Broglio et al. (2014). See the vignette for an example.
healthyR v0.1.1: Implements hospital data analysis workflow tools including modeling tools, and tools to review common administrative hospital data such as average length of stay, readmission rates, average net pay amounts by service lines, and more. See the vignette.
inTextSummaryTable v3.0.1: Provides functions to create tables of summary statistics or counts for clinical data for TLFs. These tables can be exported as in-text table for a Clinical Study Report in MS Word format or a presentation MS PowerPoint format, or as interactive table. There is an Introduction and six more vignettes including Aesthetics and Visualization.
IPDfromKM v0.1.10: Implements a method to reconstruct individual patient data from Kaplan-Meier (KM) survival curves, visualize and assess the accuracy of the reconstruction, and perform secondary analysis on the reconstructed data. The package also implements iterative KM estimation algorithm proposed in Guyot (2012).
NHSDataDictionaRy v1.2.1: Provides a common set of simplified web scraping tools for working with the NHS Data Dictionary.This package was commissioned by the NHS-R community to provide this consistency of lookups. See the vignette to get started.
NMADiagT v0.1.2: Implements the hierarchical summary receiver operating characteristic model developed by Ma et al. (2018) and the hierarchical model developed by Lian et al. (2019) for performing meta-analysis. It is able to simultaneously compare one to five diagnostic tests within a missing data framework.
packDAMipd v0.1.2: Provides functions to construct both time-homogenous and time-dependent Markov models for cost-effectiveness analyses, perform decision analyses, and conduct deterministic and probabilistic sensitivity analyses. There are vignettes on deterministic and probabilistic sensitivity analyses, simple “sick-sicker” models, age-dependent “sick-sicker” models, and cycle dependent models.
patientProfilesVis v2.0.1: Provides functions to create patient specific profile visualizations for exploration, diagnostic or monitoring purposes during a clinical trial which display the evolution of parameters such as laboratory measurements, ECG data, vital signs, adverse events and more. There is template for creating patient profiles from CDISC SDTM datasets, and an Introduction to the package.
psrwe v1.2: Provides tools to incorporate real-world evidence (RWE) into regulatory and health care decision making and includes functions which implement the PS-integrated RWE analysis methods proposed in Wang et al. (2019), Wang et al. (2020), and Chen et al. (2020). There is a vignette on propensity score integration.
raveio v0.0.3: implements an interface to the RAVE (R analysis and visualization of human intracranial electroencephalography data) project which aims at analyzing brain recordings from patients with electrodes placed on the cortical surface or inserted into the brain. See Mafnotti et al. (2020) for background.
reconstructKM v0.3.0: Provides functions for reconstructing individual-level data (time, status, arm) from Kaplan-MEIER curves published in academic journals. See Sun et al. (2018) for background and the vignette for the reconstruction procedure.
RevieweR v2.3.6: Implements a portable
Shiny tool to explore patient-level electronic health record data and perform chart review in a single integrated framework. This tool supports the OMOP common data model as well as the MIMIC-III data model, and chart review through a REDCap API. See the RevieweR Website for more information. There are several vignettes including Local, Docker, BigQuery and Shiny Server deployment and performing a Chart Review.
RHRT v1.0.1: Provides methods to scan RR interval data for Premature Ventricular Complexes and parameterise and plot the resulting Heart Rate Turbulence. See Schmidt et al. (1999) and Blesius et al. (2020) and the vignette for examples.
SAMBA v0.9.0: Implements several methods, as proposed in Beesley & Mukherjee (2020) for obtaining bias-corrected point estimates along with valid standard errors using electronic health records data with misclassifird EHR-derived disease status. See the vignette for details.
SteppedPower v0.1.0: Provides tools for power and sample size calculations and design diagnostics for longitudinal mixed models with a focus on stepped wedge designs using methods introduced in Hussey and Hughes (2007) and extensions discussed in Li et al. (2020). See the vignette to get started.
tboot v0.2.0: Provides functions to simulate clinical trial data with realistic correlation structures and assumed efficacy levels by using a tilted bootstrap resampling approach. There is a tutorial on The Tilted Bootstrap and another on Bayesian Marginal Reconstruction.
Tplyr v0.1.3: Implement a tool to simplify table creation and the data manipulation necessary to create clinical reports. There is a Getting Started Guide, and vignettes on Layers, Options, and Tables.
visR v0.2.0: Provides functions to generate clinical graphs and tables with sensible defaults based on graphical principles as described in: Vandemeulebroecke et al. (2018), Vandemeulebroecke et al. (2019), and Morris et al. (2019). Vignettes include Survival Analysis using CDISC ADaM standard, Creating Consort Flow Diagram, Styling Survival Plots and Survival Analysis.