The van der Schaar Lab’s ninth Inspiration Exchange session took place virtually on May 28, 2021, and was attended by students and professionals from the AI and machine learning community.
This was the second in a pair of sessions focusing on individualized treatment effect inference (ITE inference). Building on our discussions the previous month, this session examined a range of ITE inference problems and approaches specifically within the time series setting.
The former half of the session featured 4 short presentations, followed by a participative discussion in which members of the lab and members of the audience discussed several key topics relevant to ITE inference in the time series setting, including counterfactuals, assumptions, missing data, and interpretability.
Introduction – 0:00
Resources: overview on ITE inference – 1:44
Resources: ITE inference video tutorial series – 2:19
Mihaela’s introduction to ITE inference in the time series setting – 3:00
Presentation 1 [SyncTwin: treatment effect estimation with longitudinal EHR data // Zhaozhi Qian] – 5:48
Presentation 2 [Counterfactual recurrent network: from longitudinal observational data to ITE using causal inference // Ioana Bica] – 12:29
Presentation 3 [Time series deconfounder: estimating ITE over time in the presence of hidden confounders // Ioana Bica] – 22:20
Presentation 4 [Policy analysis using synthetic controls in continuous time // Alexis Bellot] – 34:06
Participative discussion: topic 1 [complexities and challenges for ITE in the time series setting] – 49:12
Participative discussion: topic 2 [handling assumptions of overlap and hidden confounding in the time series setting] – 57:07
Participative discussion: topic 3 [missingness in ITE over time] – 1:00:51
Participative discussion: topic 4 [making ITE over time interpretable] – 1:06:12
Intro to next sessions – 1:08:48
A full list of the lab’s publications can be found here.