psfmi
Prediction Model Pooling, Selection and Performance Evaluation Across Multiply Imputed Datasets
Last update: 17 June 2023
Software version identifier: 1.1.0, 0.1.0, 0.2.0, 0.5.0, 0.7.1, 1.0.0, 1.3.0, 1.4.0
Copyright license: GNU General Public License, version 3.0, GNU General Public License, version 2.0
Pooling, backward and forward selection of linear, logistic and Cox regression models in multiply imputed datasets. Backward and forward selection can be done from the pooled model using Rubin's Rules (RR), the D1, D2, D3, D4 and the median p-values method. This is also possible for Mixed models. The models can contain continuous, dichotomous, categorical and restricted cubic spline predictors and interaction terms betweenall these type of predictors. The stability of the modelscan be evaluated using (cluster) bootstrapping. The package further contains functions to pool model performance measures as ROC/AUC, Reclassification, R-squared, scaled Brier score, H&L test and calibrationplots for logistic regression models. Internal validation can be done across multiply imputed datasets with cross-validation or bootstrapping. The adjusted intercept after shrinkage of pooled regression coefficients can be obtained. Backward and forward selection as part of internal validation is possible. A function to externally validate logistic prediction models in multiple imputed datasets is available and a function to compare models. For Cox models a strata variablecan be included.Eekhout (2017) <doi:10.1186/s12874-017-0404-7>.Wiel (2009) <doi:10.1093/biostatistics/kxp011>.Marshall (2009) <doi:10.1186/1471-2288-9-57>.
- Methods for significance testing of categorical covariates in logistic regression models after multiple imputation: power and applicability analysis
- Testing the prediction error difference between 2 predictors
- Combining estimates of interest in prognostic modelling studies after multiple imputation: current practice and guidelines