Collaborative Heterogeneous Causal Inference Beyond Meta-analysis

Tianyu Guo, Sai Praneeth Karimireddy, Michael Jordan
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:16849-16868, 2024.

Abstract

Collaboration between different data centers is often challenged by heterogeneity across sites. To account for the heterogeneity, the state-of-the-art method is to re-weight the covariate distributions in each site to match the distribution of the target population. Nevertheless, this method still relies on the concept of traditional meta-analysis after adjusting for the distribution shift. This work proposes a collaborative inverse propensity score weighting estimator for causal inference with heterogeneous data. Instead of adjusting the distribution shift separately, we use weighted propensity score models to collaboratively adjust for the distribution shift. Our method shows significant improvements over the methods based on meta-analysis when heterogeneity increases. By incorporating outcome regression models, we prove the asymptotic normality when the covariates have dimension $d<8$. Our methods preserve privacy at individual sites by implementing federated learning protocols.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-guo24c, title = {Collaborative Heterogeneous Causal Inference Beyond Meta-analysis}, author = {Guo, Tianyu and Karimireddy, Sai Praneeth and Jordan, Michael}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {16849--16868}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/guo24c/guo24c.pdf}, url = {https://proceedings.mlr.press/v235/guo24c.html}, abstract = {Collaboration between different data centers is often challenged by heterogeneity across sites. To account for the heterogeneity, the state-of-the-art method is to re-weight the covariate distributions in each site to match the distribution of the target population. Nevertheless, this method still relies on the concept of traditional meta-analysis after adjusting for the distribution shift. This work proposes a collaborative inverse propensity score weighting estimator for causal inference with heterogeneous data. Instead of adjusting the distribution shift separately, we use weighted propensity score models to collaboratively adjust for the distribution shift. Our method shows significant improvements over the methods based on meta-analysis when heterogeneity increases. By incorporating outcome regression models, we prove the asymptotic normality when the covariates have dimension $d<8$. Our methods preserve privacy at individual sites by implementing federated learning protocols.} }
Endnote
%0 Conference Paper %T Collaborative Heterogeneous Causal Inference Beyond Meta-analysis %A Tianyu Guo %A Sai Praneeth Karimireddy %A Michael Jordan %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-guo24c %I PMLR %P 16849--16868 %U https://proceedings.mlr.press/v235/guo24c.html %V 235 %X Collaboration between different data centers is often challenged by heterogeneity across sites. To account for the heterogeneity, the state-of-the-art method is to re-weight the covariate distributions in each site to match the distribution of the target population. Nevertheless, this method still relies on the concept of traditional meta-analysis after adjusting for the distribution shift. This work proposes a collaborative inverse propensity score weighting estimator for causal inference with heterogeneous data. Instead of adjusting the distribution shift separately, we use weighted propensity score models to collaboratively adjust for the distribution shift. Our method shows significant improvements over the methods based on meta-analysis when heterogeneity increases. By incorporating outcome regression models, we prove the asymptotic normality when the covariates have dimension $d<8$. Our methods preserve privacy at individual sites by implementing federated learning protocols.
APA
Guo, T., Karimireddy, S.P. & Jordan, M.. (2024). Collaborative Heterogeneous Causal Inference Beyond Meta-analysis. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:16849-16868 Available from https://proceedings.mlr.press/v235/guo24c.html.

Related Material