Approximating counterfactual bounds while fusing observational, biased and randomised data sources. (English) Zbl 1531.68142
Summary: We address the problem of integrating data from multiple, possibly biased, observational and interventional studies, to eventually compute counterfactuals in structural causal models. We start from the case of a single observational dataset affected by a selection bias. We show that the likelihood of the available data has no local maxima. This enables us to use the causal expectation-maximisation scheme to approximate the bounds for partially identifiable counterfactual queries, which are the focus of this paper. We then show how the same approach can address the general case of multiple datasets, no matter whether interventional or observational, biased or unbiased, by remapping it into the former one via graphical transformations. Systematic numerical experiments and a case study on palliative care show the effectiveness of our approach, while hinting at the benefits of fusing heterogeneous data sources to get informative outcomes in case of partial identifiability.
MSC:
| 68T37 | Reasoning under uncertainty in the context of artificial intelligence |
| 62D20 | Causal inference from observational studies |
| 68T09 | Computational aspects of data analysis and big data |
Keywords:
counterfactuals; randomised data; biased data; structural causal models; expectation maximisationSoftware:
CREDICIReferences:
| [1] | Balke, A.; Pearl, J., Counterfactual probabilities: computational methods, bounds and applications, (De Mantaras, R. L.; Poole, D., Proceedings of the Tenth International Conference on Uncertainty in Artificial Intelligence (1994), Morgan Kaufmann Publishers Inc.), 46-54 |
| [2] | Bareinboim, E.; Pearl, J., Controlling selection bias in causal inference, (Lawrence, N. D.; Girolami, M., Proceedings of the Fifteenth International Conference on Artificial Intelligence and Statistics (2012), PMLR), 100-108 |
| [3] | Bareinboim, E.; Pearl, J., Causal inference and the data-fusion problem, Proc. Natl. Acad. Sci., 113, 7345-7352 (2016) |
| [4] | Bareinboim, E.; Tian, J., Recovering causal effects from selection bias, (Koenig, S.; Blai Bonet, B., Proceedings of the Twenty-Ninth AAAI Conference on Artificial Intelligence (2015), AAAI Press) |
| [5] | Cabañas, R.; Antonucci, A.; Huber, D.; Zaffalon, M., CREDICI: a Java library for causal inference by credal networks, (Yaeger, M.; Nielsen, T. D., Proceedings of the Tenth International Conference on Probabilistic Graphical Models. Proceedings of the Tenth International Conference on Probabilistic Graphical Models, PMLR, vol. 138 (2020)), 597-600, JMLR.org |
| [6] | Cooper, G. F., Causal discovery from data in the presence of selection bias, (Fisher, D.; Lenz, H. J., Pre-Proceedings of the Fifth International Workshop on Artificial Intelligence and Statistics (1995), PMLR), 140-150 |
| [7] | Darwiche, A., Causal inference using tractable circuits (2022) |
| [8] | Duarte, G.; Finkelsteinm, N.; Knox, D.; Mummolo, J.; Shpitser, I., An automated approach to causal inference in discrete settings, J. Am. Stat. Assoc. (2023) |
| [9] | Huber, D.; Chen, Y.; Antonucci, A.; Darwiche, A.; Zaffalon, M., Tractable bounding of counterfactual queries by knowledge compilation (2023), Sixth Workshop on Tractable Probabilistic Modeling @ UAI 2023 |
| [10] | Kern, H.; Corani, G.; Huber, D.; Vermes, N.; Zaffalon, M.; Varini, M.; Wenzel, C.; Fringer, A., Impact on place of death in cancer patients: a causal exploration in southern Switzerland, BMC Palliat. Care, 19, 1-10 (2020) |
| [11] | Koller, D.; Friedman, N., Probabilistic Graphical Models: Principles and Techniques (2009), MIT · Zbl 1183.68483 |
| [12] | Mueller, S.; Pearl, J., Personalized decision making-a conceptual introduction, J. Causal Inference, 11, 1-13 (2023) |
| [13] | Pearl, J., Causality (2009), Cambridge University Press · Zbl 1188.68291 |
| [14] | Pearl, J., A solution to a class of selection-bias problems (2012), UCLA Cognitive Systems Laboratory, Technical Report R-407 |
| [15] | Shpitser, I.; Pearl, J., What counterfactuals can be tested, (Parr, R.; van der Gaag, L., Proceedings of the Twenty-Third Conference on Uncertainty in Artificial Intelligence (2007), AUAI Press), 352-359 |
| [16] | Tian, J., Studies in Causal Reasoning and Learning (2002), UCLA, Ph.D. thesis |
| [17] | Tian, J.; Pearl, J., Probabilities of causation: bounds and identification, Ann. Math. Artif. Intell., 28, 287-313 (2000) · Zbl 1048.03502 |
| [18] | Zaffalon, M.; Antonucci, A.; Cabañas, R., Structural causal models are (solvable by) credal networks, (Jaeger, M.; Nielsen, T. D., Proceedings of the Tenth International Conference on Probabilistic Graphical Models. Proceedings of the Tenth International Conference on Probabilistic Graphical Models, PMLR, vol. 138 (2020)), 581-592, JMLR.org |
| [19] | Zaffalon, M.; Antonucci, A.; Cabañas, R., Causal expectation-maximisation (2021), Why-21@NeurIPS |
| [20] | Zaffalon, M.; Antonucci, A.; Cabañas, R.; Huber, D.; Azzimonti, D., Bounding counterfactuals under selection bias, (Salmerón, A.; Rumí, R., Proceedings of the Eleventh International Conference on Probabilistic Graphical Models. Proceedings of the Eleventh International Conference on Probabilistic Graphical Models, PMLR, vol. 186 (2022)), 289-300, JMLR.org |
| [21] | Zaffalon, M.; Antonucci, A.; Cabañas, R.; Huber, D.; Azzimonti, D., On the efficient bounding of counterfactual queries (2023), submitted for publication |
| [22] | Zhang, J.; Tian, J.; Bareinboim, E., Partial counterfactual identification from observational and experimental data, (Chaudhuri, K.; Jegelka, S.; Song, L.; Szepesvári, C.; Niu, G.; Sabato, S., Proceedings of the Thirty-Ninth International Conference on Machine Learning (2022)), 26548-26558, JMLR.org |
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.
