Bayesian Federated Inference | SaddlepointScienceBV

Bayesian Federated Inference (BFI)

Harness the combined statistical power of multiple datasets without sharing patient data

Bayesian Federated Inference (BFI) enables organisations to combine statistical information from separate datasets without pooling or sharing the underlying patient-level data.

BFI can recover what would have been learned from a pooled analysis, while keeping sensitive data at the organisation where they were collected.

Why Bayesian Federated Inference?

Analyse distributed datasets
- Learn from data held across multiple hospitals, research centres or organisations without requiring the underlying patient data to be centralised.
Preserve data locality
- Patient-level data remains within the organisation where it was collected, reducing the need for direct data sharing.
Unlock statistical power
- Combine information from multiple datasets to gain insights that may not be possible from individual datasets alone.
Designed for high-dimensional biomedical data
- BFI addresses the growing challenge of analysing datasets containing many measurements per patient.
One-shot method in contrast to conventional federated approaches
- More data safe
- More accurate
- Easier to add new research institutions

What can BFI help you achieve?

Collaborative clinical research
- Combine statistical information from multiple clinical centres without pooling patient-level datasets.
Multi-centre studies
- Harness information across participating centres while keeping local data within each centre.
Rare disease research
- Increase the effective sample size by combining information across organisations when individual datasets are necessarily small.
Privacy-conscious data analysis
- Perform collaborative statistical analyses without requiring direct exchange of sensitive patient-level data.
High-dimensional biomedical research
- Combine information across datasets containing large numbers of patient characteristics.
One-shot approach
- Avoid risks and excessive CPU demands of federated learning approaches

Publications

Jonker, M. A., Pazira, H., & Coolen, A. C. C. (2024). Bayesian federated inference for estimating statistical models based on non-shared multicenter data sets. Statistics in Medicine, 43, 2421–2438. https://doi.org/10.1002/sim.10072
Jonker, M. A., Pazira, H., & Coolen, A. C. C. (2025). Bayesian Federated Inference for regression models based on non-shared medical center data. Research Synthesis Methods, 16, 383–423. https://doi.org/10.1017/rsm.2025.23
Pazira, H., Massa, E. M., Weijers, J. A., Coolen, A. C. C., & Jonker, M. A. (2025). Bayesian federated inference for survival models. Journal of Applied Statistics. https://doi.org/10.1080/02664763.2025.2511932

Are you ready to explore what your data can reveal?

Do you have a challenging biomedical dataset or research question?

Whether you are working with a small dataset, high-dimensional measurements, heterogeneous patient populations or distributed clinical data, our team can help explore whether our approach is suitable for your research question.