Overview: MRMCaov is an R package that, similar to the stand-alone software OR-DBM 2.51 (which currently can no longer be installed due to Microsoft no longer supporting the needed .NET framework files), has the ability to perform Obuchowski-Rockette (OR) analyses for all three types of sampling analyses: random readers and random cases, random readers and fixed cases, and fixed reader and random cases. For the OR method three types of error covariance estimation can be selected: jackknife, unbiased, or the method of DeLong et al. (1988, Biometrics). Reader performance measures that can be analyzed include the AUC, partial AUC, sensitivity at fixed specificity, specificity at fixed sensitivity, expected utility of ROC curves, likelihood ratio of positive or negative tests, sensitivity, specificity, and user-defined metrics. ROC curve fitting methods include empirical, binormal, and binormal likelihood-ratio (a.k.a. “proper binormal” and bi-chi-square). Other features include parametric and nonparametric estimation and plotting of ROC curves; support for factorial, nested, and partially paired study designs; and conversion of Obuchowski-Rockette to Roe & Metz simulation model parameters and vice versa.
This package was developed by Brian Smith (Dept. of Biostatistics, University of Iowa) with Stephen Hillis (Depts. Of Radiology and Biostatistics, University of Iowa) advising on statistical issues.
This software was funded as part of the “Generalized Obuchowski-Rockette Methodology for Analysis of Radiologic Diagnostic Imaging Studies” grant, funded by the National Institute of Biomedical Imaging and Bioengineering (NIBIB) of the National Institutes of Health under Award Number R01EB025174.
Click on the following link to go to the download page (you will then need to scroll down the page to locate the download link), or alternatively you can download it directly from within R-Studio: Tools --> Install Packages --> Install from: Repository (CRAN).