screenr: screenr Package

In sgutreuter/screenr: Construction of Binary Test-Screening Rules

screenr Package

Description

The screenr package enables construction of binary test-screening tools. It is designed to enable those with only a basic familiarity with R to develop, validate and implement screening tools for diagnostic tests. screenr integrates the capabilities of the glm, glmpath and pROC packages for convenience and ease of use.

Consider the situation where a diagnostic test for some condition is relatively expensive, and the condition is rare. In that case, universal testing would not be efficient in terms of the yield of postive results per test performed. Now suppose that responses to a set of simple screening questions may be predictive of the condition. Package screenr enables estimation of thresholds for making decisions about when to test in order to screen in/out individuals based on Receiver Operating Characteristics (ROC) estimated from an initial sample. The choice of a particular screening threshold is left to the user, and should be based on careful consideration of application-specific tradeoffs between sensitivity and specificity. screenr also enables easy construction of screening tools.

A tutorial is available from vignette("screenr_Tutorial", package = "screenr").

The pdf versions of the package manual and the tutorial are available at https://github.com/sgutreuter/screenr.

Details

The high-level functions in the screenr package are:

lasso_screenr

Selection of logistic models based on GLM path regularization

logreg_screenr

Test-screening based maximum-likelhood estimation of logistic models

easy_tool

Easy implementation of test-screening tools

inverse_link

Compute the inverses of logit, cloglog and probit link functions

simple_screenr

(Too) Simple un-optimized test-screening

rescale_to_int

Rescale a strictly positive vector or matrix of real numbers to integers

link{roc_ci}

Compute bootstrap confidence intervals from a roc-class object. See package pROC

sens_spec_plus

Sensitivity, specificity and friends

screenr provides the usual plot, print, summary, predict methods for the objects produced by lasso_screenr, logreg_screenr, gee_screenr, simple_screenr and easy_tool, and also coef and confint methods for lasso_screenr, screenr also provides get_what methods to extract object components, and ntpp methods for computation of the average number of tests required to detect a single positive result and the residual positivity among those screened out of testing.

Note

The canonical source repository for screenr is https://github.com/sgutreuter/screenr

Author(s)

Steve Gutreuter: [e-mail]

References

Teferi W, Gutreuter S, Bekele A et al. Adapting strategies for effective and efficient pediatric HIV case finding: Risk screening tool for testing children presenting at high-risk entry points. BMC Infectious Diseases. 2022; 22:480. http://doi.org/10.1186/s12879-022-07460-w

sgutreuter/screenr documentation built on Oct. 19, 2024, 12:49 p.m.