Prediction using Multiple Imputation

Calibration of generalized linear models and Cox regression models for prediction using multiple imputation to account for missing values in the predictors as described in the paper by "Mertens, Banzato and de Wreede" (2018) . The methodology and calculations described in this paper are fully implemented in this package. The vignette describes all data analytic steps which allow users to replicate results using the package functions on the data analyzed in the paper or on their own data. Imputations are generated using the package 'mice' without using the outcomes of observations for which the predictions are generated. Two options are provided to generate predictions. The first is prediction-averaging of predictions calibrated from single models fitted on single imputed datasets within a set of multiple imputations. The second is application of the Rubin's rules pooled model. For both implementations, unobserved values in the predictor data of new observations for which the predictions are derived are automatically imputed. The package contains two basic functions. The first, mipred() generates predictions of outcome on new observations. The second, generates cross-validated predictions with the methodology on existing data for which outcomes have already been observed. The present version is still in development and should support continuous, binary and counting outcomes, but we have only thoroughly checked performance for binary outcome logistic regression modeling. We will include the Cox regression extension later.


Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.