Causal Modeling of Policy Interventions From Treatment-Outcome Sequences

Çağlar Hızlı, S. T. John, Anne Tuulikki Juuti, Tuure Tapani Saarinen, Kirsi Hannele Pietiläinen, Pekka Marttinen
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:13050-13084, 2023.

Abstract

A treatment policy defines when and what treatments are applied to affect some outcome of interest. Data-driven decision-making requires the ability to predict what happens if a policy is changed. Existing methods that predict how the outcome evolves under different scenarios assume that the tentative sequences of future treatments are fixed in advance, while in practice the treatments are determined stochastically by a policy and may depend, for example, on the efficiency of previous treatments. Therefore, the current methods are not applicable if the treatment policy is unknown or a counterfactual analysis is needed. To handle these limitations, we model the treatments and outcomes jointly in continuous time, by combining Gaussian processes and point processes. Our model enables the estimation of a treatment policy from observational sequences of treatments and outcomes, and it can predict the interventional and counterfactual progression of the outcome after an intervention on the treatment policy (in contrast with the causal effect of a single treatment). We show with real-world and semi-synthetic data on blood glucose progression that our method can answer causal queries more accurately than existing alternatives.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-hizli23a, title = {Causal Modeling of Policy Interventions From Treatment-Outcome Sequences}, author = {H{\i}zl{\i}, \c{C}a\u{g}lar and John, S. T. and Juuti, Anne Tuulikki and Saarinen, Tuure Tapani and Pietil\"{a}inen, Kirsi Hannele and Marttinen, Pekka}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {13050--13084}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/hizli23a/hizli23a.pdf}, url = {https://proceedings.mlr.press/v202/hizli23a.html}, abstract = {A treatment policy defines when and what treatments are applied to affect some outcome of interest. Data-driven decision-making requires the ability to predict what happens if a policy is changed. Existing methods that predict how the outcome evolves under different scenarios assume that the tentative sequences of future treatments are fixed in advance, while in practice the treatments are determined stochastically by a policy and may depend, for example, on the efficiency of previous treatments. Therefore, the current methods are not applicable if the treatment policy is unknown or a counterfactual analysis is needed. To handle these limitations, we model the treatments and outcomes jointly in continuous time, by combining Gaussian processes and point processes. Our model enables the estimation of a treatment policy from observational sequences of treatments and outcomes, and it can predict the interventional and counterfactual progression of the outcome after an intervention on the treatment policy (in contrast with the causal effect of a single treatment). We show with real-world and semi-synthetic data on blood glucose progression that our method can answer causal queries more accurately than existing alternatives.} }
Endnote
%0 Conference Paper %T Causal Modeling of Policy Interventions From Treatment-Outcome Sequences %A Çağlar Hızlı %A S. T. John %A Anne Tuulikki Juuti %A Tuure Tapani Saarinen %A Kirsi Hannele Pietiläinen %A Pekka Marttinen %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-hizli23a %I PMLR %P 13050--13084 %U https://proceedings.mlr.press/v202/hizli23a.html %V 202 %X A treatment policy defines when and what treatments are applied to affect some outcome of interest. Data-driven decision-making requires the ability to predict what happens if a policy is changed. Existing methods that predict how the outcome evolves under different scenarios assume that the tentative sequences of future treatments are fixed in advance, while in practice the treatments are determined stochastically by a policy and may depend, for example, on the efficiency of previous treatments. Therefore, the current methods are not applicable if the treatment policy is unknown or a counterfactual analysis is needed. To handle these limitations, we model the treatments and outcomes jointly in continuous time, by combining Gaussian processes and point processes. Our model enables the estimation of a treatment policy from observational sequences of treatments and outcomes, and it can predict the interventional and counterfactual progression of the outcome after an intervention on the treatment policy (in contrast with the causal effect of a single treatment). We show with real-world and semi-synthetic data on blood glucose progression that our method can answer causal queries more accurately than existing alternatives.
APA
Hızlı, Ç., John, S.T., Juuti, A.T., Saarinen, T.T., Pietiläinen, K.H. & Marttinen, P.. (2023). Causal Modeling of Policy Interventions From Treatment-Outcome Sequences. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:13050-13084 Available from https://proceedings.mlr.press/v202/hizli23a.html.

Related Material