Predicting the impact of treatments over time with uncertainty aware neural differential equations.

Edward De Brouwer, Javier Gonzalez, Stephanie Hyland
Proceedings of The 25th International Conference on Artificial Intelligence and Statistics, PMLR 151:4705-4722, 2022.

Abstract

Predicting the impact of treatments from ob- servational data only still represents a major challenge despite recent significant advances in time series modeling. Treatment assignments are usually correlated with the predictors of the response, resulting in a lack of data support for counterfactual predictions and therefore in poor quality estimates. Developments in causal inference have lead to methods addressing this confounding by requiring a minimum level of overlap. However, overlap is difficult to assess and usually not satisfied in practice. In this work, we propose Counterfactual ODE (CF-ODE), a novel method to predict the impact of treatments continuously over time using Neural Ordinary Differential Equations equipped with uncertainty estimates. This allows to specifically assess which treatment outcomes can be reliably predicted. We demonstrate over several longitudinal datasets that CF-ODE provides more accurate predictions and more reliable uncertainty estimates than previously available methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v151-de-brouwer22a, title = { Predicting the impact of treatments over time with uncertainty aware neural differential equations. }, author = {De Brouwer, Edward and Gonzalez, Javier and Hyland, Stephanie}, booktitle = {Proceedings of The 25th International Conference on Artificial Intelligence and Statistics}, pages = {4705--4722}, year = {2022}, editor = {Camps-Valls, Gustau and Ruiz, Francisco J. R. and Valera, Isabel}, volume = {151}, series = {Proceedings of Machine Learning Research}, month = {28--30 Mar}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v151/de-brouwer22a/de-brouwer22a.pdf}, url = {https://proceedings.mlr.press/v151/de-brouwer22a.html}, abstract = { Predicting the impact of treatments from ob- servational data only still represents a major challenge despite recent significant advances in time series modeling. Treatment assignments are usually correlated with the predictors of the response, resulting in a lack of data support for counterfactual predictions and therefore in poor quality estimates. Developments in causal inference have lead to methods addressing this confounding by requiring a minimum level of overlap. However, overlap is difficult to assess and usually not satisfied in practice. In this work, we propose Counterfactual ODE (CF-ODE), a novel method to predict the impact of treatments continuously over time using Neural Ordinary Differential Equations equipped with uncertainty estimates. This allows to specifically assess which treatment outcomes can be reliably predicted. We demonstrate over several longitudinal datasets that CF-ODE provides more accurate predictions and more reliable uncertainty estimates than previously available methods. } }
Endnote
%0 Conference Paper %T Predicting the impact of treatments over time with uncertainty aware neural differential equations. %A Edward De Brouwer %A Javier Gonzalez %A Stephanie Hyland %B Proceedings of The 25th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2022 %E Gustau Camps-Valls %E Francisco J. R. Ruiz %E Isabel Valera %F pmlr-v151-de-brouwer22a %I PMLR %P 4705--4722 %U https://proceedings.mlr.press/v151/de-brouwer22a.html %V 151 %X Predicting the impact of treatments from ob- servational data only still represents a major challenge despite recent significant advances in time series modeling. Treatment assignments are usually correlated with the predictors of the response, resulting in a lack of data support for counterfactual predictions and therefore in poor quality estimates. Developments in causal inference have lead to methods addressing this confounding by requiring a minimum level of overlap. However, overlap is difficult to assess and usually not satisfied in practice. In this work, we propose Counterfactual ODE (CF-ODE), a novel method to predict the impact of treatments continuously over time using Neural Ordinary Differential Equations equipped with uncertainty estimates. This allows to specifically assess which treatment outcomes can be reliably predicted. We demonstrate over several longitudinal datasets that CF-ODE provides more accurate predictions and more reliable uncertainty estimates than previously available methods.
APA
De Brouwer, E., Gonzalez, J. & Hyland, S.. (2022). Predicting the impact of treatments over time with uncertainty aware neural differential equations. . Proceedings of The 25th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 151:4705-4722 Available from https://proceedings.mlr.press/v151/de-brouwer22a.html.

Related Material