Predictive averaging and Rubin’s rule-based model pooling to predict survival risk with imputations in the presence of missing patient data: methodology and verification using two case studies and simulations
Abstract
Background: Practical estimation of patient risk is often complicated by predictor values being unobserved or missing. In this paper we study the calibration and application of models for patient risk prediction when both the past patient data to which the prediction models are fit contains such missing values, while the future patient records may also be partially observed. Typical applications are often encountered in prognostic observational research or with registry data. Methods: This paper presents a detailed study on the use of multiple imputation in prediction, to deal with missing values in both past (calibration data) as well as future patient predictor records. We present the two distinct approaches which may be considered to use imputations to account for the presence of missing values. One method directly averages survival predictions obtained from separate models which are fitted on distinct imputations of the data. The other method first pools the intermediate effect and baseline estimates of these models, before calculation of the predictions from the pooled model. Such pooled estimates could be obtained from Rubin's rules, for example. Methods are introduced and demonstrated based on two motivating datasets. The application for Cox regression survival modeling is studied in detail. Method performance is verified through cross-validation, a separate validation set and an extensive simulation study. Results: All methods are comparable with respect to bias and Brier score (accuracy) assessment. Major differences are however found between both methods when comparing predicted per-patient risks between repetitions of the fitting procedure with a different set of imputations and when the same number of imputations is used. Predictive averaging is preferable, because the difference between such replicate predictions can be reduced to zero by increasing the number of imputations, which is not the case with the pooled model. Conclusions: Predictive averaging should be preferred to use of a single pooled model, when using multiple imputations for missing predictor data. Single imputation should not be used in prediction on missing data.
Related
- Integration of a smooth mesh-based contact pressure model into tracking and predictive simulations
- A Machine Learning Approach to Predict Listeria monocytogenes Growth by Integrating FSSP and PMP Simulations: A Focus on t₁₀₀ as a Possible Endpoint Parameters for Predictive Models.
- Dimensionality-induced critical phase transition in stochastic Lotka-Volterra equation: From statistical averaging to systemic tipping point
- Averaging in gene perturbation benchmarks hides perturbation-strength variation: Consequences for model evaluation
- Taylor's Law, Smith's Law, and Diversity Power Laws: A Novel Triple Power Law Methodology for Scaling Diversity and Heterogeneity
BibTeX
@article{mertens2026predictive,
title = {{Predictive averaging and Rubin’s rule-based model pooling to predict survival risk with imputations in the presence of missing patient data: methodology and verification using two case studies and simulations}},
author = {Bart J.A. Mertens and Lars L. J. van der Burg and Erwan Donal and Nina Ajmone Marsan and Liesbeth C. de Wreede},
journal = {BMC Medical Research Methodology},
year = {2026},
month = oct,
doi = {10.1186/s12874-026-02888-z},
url = {https://hal.science/hal-05830308v1},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - Predictive averaging and Rubin’s rule-based model pooling to predict survival risk with imputations in the presence of missing patient data: methodology and verification using two case studies and simulations
AU - Bart J.A. Mertens
AU - Lars L. J. van der Burg
AU - Erwan Donal
AU - Nina Ajmone Marsan
AU - Liesbeth C. de Wreede
PY - 2026
DA - 2026/10/21
JO - BMC Medical Research Methodology
DO - 10.1186/s12874-026-02888-z
UR - https://hal.science/hal-05830308v1
AB - Background: Practical estimation of patient risk is often complicated by predictor values being unobserved or missing. In this paper we study the calibration and application of models for patient risk prediction when both the past patient data to which the prediction models are fit contains such missing values, while the future patient records may also be partially observed. Typical applications are often encountered in prognostic observational research or with registry data. Methods: This paper presents a detailed study on the use of multiple imputation in prediction, to deal with missing values in both past (calibration data) as well as future patient predictor records. We present the two distinct approaches which may be considered to use imputations to account for the presence of missing values. One method directly averages survival predictions obtained from separate models which are fitted on distinct imputations of the data. The other method first pools the intermediate effect and baseline estimates of these models, before calculation of the predictions from the pooled model. Such pooled estimates could be obtained from Rubin's rules, for example. Methods are introduced and demonstrated based on two motivating datasets. The application for Cox regression survival modeling is studied in detail. Method performance is verified through cross-validation, a separate validation set and an extensive simulation study. Results: All methods are comparable with respect to bias and Brier score (accuracy) assessment. Major differences are however found between both methods when comparing predicted per-patient risks between repetitions of the fitting procedure with a different set of imputations and when the same number of imputations is used. Predictive averaging is preferable, because the difference between such replicate predictions can be reduced to zero by increasing the number of imputations, which is not the case with the pooled model. Conclusions: Predictive averaging should be preferred to use of a single pooled model, when using multiple imputations for missing predictor data. Single imputation should not be used in prediction on missing data.
SN - 1471-2288
ER -