Learning plug-in surrogate endpoints for randomized experiments

Alessandro-Umberto Margueritte, Ahmet Zahid Balcıoğlu, Jesse H. Krijthe, Dave Zachariah, Fredrik D. Johansson
Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence, PMLR 337:4272-4298, 2026.

Abstract

Surrogate endpoints are used in place of long-term outcomes in randomized experiments when observing the real outcome for a large enough cohort is prohibitively expensive or impractical. A short-term surrogate is good if the result of an experiment using the surrogate is predictive of the result of a hypothetical study using the real outcome. Much attention has been paid to formalizing this property in causal terms, but most criteria are unidentifiable and cannot be turned into practical algorithms for learning surrogate endpoints from data. To address this, we study plug-in composite surrogates, functions of post-treatment variables that may be substituted directly for the primary outcome in a randomized experiment. We propose two methods for learning plug-in surrogates that maximize effect predictiveness, and characterize the possibility of finding endpoints that yield unbiased effect estimates in representative scenarios. Finally, in both synthetic experiments with known effects and in data from a real-world experiment, we find that our method, based on directly modeling the surrogate effect, returns plug-in endpoints more predictive of the primary effect than established methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v337-margueritte26a, title = {Learning plug-in surrogate endpoints for randomized experiments}, author = {Margueritte, Alessandro-Umberto and Balc{\i}o{\u{g}}lu, Ahmet Zahid and Krijthe, Jesse H. and Zachariah, Dave and Johansson, Fredrik D.}, booktitle = {Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence}, pages = {4272--4298}, year = {2026}, editor = {Perković, Emilija and Malinsky, Daniel}, volume = {337}, series = {Proceedings of Machine Learning Research}, month = {17--21 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v337/main/assets/margueritte26a/margueritte26a.pdf}, url = {https://proceedings.mlr.press/v337/margueritte26a.html}, abstract = {Surrogate endpoints are used in place of long-term outcomes in randomized experiments when observing the real outcome for a large enough cohort is prohibitively expensive or impractical. A short-term surrogate is good if the result of an experiment using the surrogate is predictive of the result of a hypothetical study using the real outcome. Much attention has been paid to formalizing this property in causal terms, but most criteria are unidentifiable and cannot be turned into practical algorithms for learning surrogate endpoints from data. To address this, we study plug-in composite surrogates, functions of post-treatment variables that may be substituted directly for the primary outcome in a randomized experiment. We propose two methods for learning plug-in surrogates that maximize effect predictiveness, and characterize the possibility of finding endpoints that yield unbiased effect estimates in representative scenarios. Finally, in both synthetic experiments with known effects and in data from a real-world experiment, we find that our method, based on directly modeling the surrogate effect, returns plug-in endpoints more predictive of the primary effect than established methods.} }
Endnote
%0 Conference Paper %T Learning plug-in surrogate endpoints for randomized experiments %A Alessandro-Umberto Margueritte %A Ahmet Zahid Balcıoğlu %A Jesse H. Krijthe %A Dave Zachariah %A Fredrik D. Johansson %B Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2026 %E Emilija Perković %E Daniel Malinsky %F pmlr-v337-margueritte26a %I PMLR %P 4272--4298 %U https://proceedings.mlr.press/v337/margueritte26a.html %V 337 %X Surrogate endpoints are used in place of long-term outcomes in randomized experiments when observing the real outcome for a large enough cohort is prohibitively expensive or impractical. A short-term surrogate is good if the result of an experiment using the surrogate is predictive of the result of a hypothetical study using the real outcome. Much attention has been paid to formalizing this property in causal terms, but most criteria are unidentifiable and cannot be turned into practical algorithms for learning surrogate endpoints from data. To address this, we study plug-in composite surrogates, functions of post-treatment variables that may be substituted directly for the primary outcome in a randomized experiment. We propose two methods for learning plug-in surrogates that maximize effect predictiveness, and characterize the possibility of finding endpoints that yield unbiased effect estimates in representative scenarios. Finally, in both synthetic experiments with known effects and in data from a real-world experiment, we find that our method, based on directly modeling the surrogate effect, returns plug-in endpoints more predictive of the primary effect than established methods.
APA
Margueritte, A., Balcıoğlu, A.Z., Krijthe, J.H., Zachariah, D. & Johansson, F.D.. (2026). Learning plug-in surrogate endpoints for randomized experiments. Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 337:4272-4298 Available from https://proceedings.mlr.press/v337/margueritte26a.html.

Related Material