Most examples and solutions will be concentrated on Missing Data topics. Posts may also concern Regression modeling, Clinical Prediction models and Spline regression modeling. Most examples originate from FAQs asked during statistical consultations or during courses. Examples are provided for applied researchers and frequenly in RStudio. RStudio is an open source free package with a lot of possibilities, making it easy to share code so that you can use the code example and adjust it to your own needs.

Continue reading

External validation means that the performance of a prediction model is studied in a new (external) patient dataset that is not used to develop the model. It is important to use the correct predicted probabilities to evaluate if the external validation of the model is successful!

Continue reading

psfmi package update

With the psfmi_mm function pooling and selection of (generalized) linear mixed models, i.e linear and logistic mixed models, is possible. With the psfmi_stab function the stability of models after using psfmi_lr, psfmi_coxr and psfmi_mm can be evaluated. The latter function uses (single) bootstrapping for the psfmi_lr and psfmi_coxr functions and cluster bootstrapping for the psfmi_mm function. With the function the bootstrap inclusion frequency of predictors and models can be estimated.

Continue reading

Package psfmi

More information about the package can be found on the Cran website. You can easily install the package by running install.packages("psfmi") in the Console window in Rstudio or R. The development version can be installed from Github by using: install.packages("devtools") library(devtools) devtools::install_github("mwheymans/psfmi") library(psfmi) If you have questions about the psfmi package send an email to [email protected] or leave a message below. Enjoy using the package!

Continue reading

Author's picture

Martijn W Heymans

Assistant Professor of Applied Biostatistics | Department of Epidemiology and Biostatistics | Courses: Missing Data and Prediction Models | Publications | email: [email protected]