Multiple IMputation In CLInical Prediction modelling (MIMICLIP)

• Applicant: Dr. Manuel Feißt
• Subject Area: Epidemiology and Medical Biometry/Statistics
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 547069732

The aim of this work is to investigate and compare two different main strategies combining multiple imputation and k-fold cross-validation in the development process of predictive models based on an incomplete dataset in a clinical context. The first strategy is to perform multiple imputation before model validation. The second strategy is to perform the multiple imputation in the corresponding step of the model validation. Both strategies will be compared for a variety of clinically relevant scenarios to evaluate their advantages and disadvantages in the development and validation of a clinical prediction model. The focus is on evaluating the performance of the strategies and their possible extensions or variations as well as developing recommendations on which strategy should be preferred in which scenario.

• DFG Programme: Research Grants