Incremental Causal Effect for Time to Treatment Initialization
Andrew Ying, Zhichen Zhao, Ronghui Xu

TL;DR
This paper introduces a method to estimate the causal effect of the timing of treatment initiation in various health and industry contexts, without relying on the positivity assumption, using inverse probability weighting.
Contribution
It develops an identification strategy for incremental causal effects of treatment timing that relaxes the positivity assumption and provides an estimation framework.
Findings
Successfully applied to simulate data demonstrating the method's validity.
Applied to rheumatoid arthritis data to assess methotrexate timing effects.
Shows the approach can handle real-world treatment timing scenarios.
Abstract
We consider time to treatment initialization. This can commonly occur in preventive medicine, such as disease screening and vaccination; it can also occur with non-fatal health conditions such as HIV infection without the onset of AIDS; or in tech industry where items wait to be reviewed manually as abusive or not, etc. While traditional causal inference focused on `when to treat' and its effects, including their possible dependence on subject characteristics, we consider the incremental causal effect when the intensity of time to treatment initialization is intervened upon. We provide identification of the incremental causal effect without the commonly required positivity assumption, as well as an estimation framework using inverse probability weighting. We illustrate our approach via simulation, and apply it to a rheumatoid arthritis study to evaluate the incremental effect of time to…
Peer Reviews
Decision·ICLR 2025 Poster
While there are some clarity issues with the narrative of the paper, the individual sections and descriptions in the paper are clear and easily readable. The literature review on incremental causal effects and time to treatment is very thorough, situating the paper nicely in the literature. The examples chosen, especially the rheumatoid arthritis example, are strong, and the analysis of the rheumatoid arthritis experiment (the reasoning about doubling the hazard decreasing joint pain) is espec
The biggest weakness of this paper in my eyes is that the problem being solved is not clearly defined. Some of this seems to be due to language issues (while the occasional grammatical issue or awkward phrase don't generally impede understanding, there are a few parts where the intended meaning isn't clear), and some of this is due to the lack of a clear motivating example". Specifically: - In the introduction, the authors define an incremental intervention as an intervention "that is not pre
1- The paper addresses an important problem and proposes an algorithm to solve it. 2- The proposed approach has been analyzed both theoretically and empirically.
1- Some definitions in Subsection 3.1, such as the hazard function and related concepts, are not clear to the reader. It would be beneficial to provide more detail, as there is still enough space available.
1. This paper extends incremental causal effects, which do not rely on the traditional positivity assumption, to a new setting. This advancement allows for new approaches to studying time-to-treatment problems in fields such as public health and policy-making. 2. Theoretical guarantee is provided. 3. The presentation and flow of this paper are clear.
1. Additional experiments could provide deeper insights into the behavior and robustness of the proposed approach under different scenarios. For example exploring different shift interventions, hazard functions, and comparative analysis against any alternative estimators. Minor comments: 2. The clarity of Theorems 2 and 3 could be improved by stating all conditions and notations explicitly. 2. In the simulation, it can be made more clear to state the true effect and whether the outcome is cens
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsHealth Systems, Economic Evaluations, Quality of Life
