OrthoGen: A Generative Orthogonal Learner for Time-Varying Treatments
Estimating conditional distributional potential outcomes (CDPOs) over time is important in medicine (e.g., to estimate patient-specific risks under different treatment sequences).
Key points
- In this paper, we aim to learn CDPOs under time-varying treatments using flexible generative models.
- (1) We introduce a tailored adjustment strategy for our setting, namely, generative recursive g-computation.
- (2) We thus introduce OrthoGen, a Neyman-orthogonal and doubly robust generative learner.
- Across experiments with synthetic, semi-synthetic and real-world datasets, we find that OrthoGen is highly effective.
Sources (1)
- [1]OrthoGen: A Generative Orthogonal Learner for Time-Varying TreatmentsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 03:07 PM
Estimating conditional distributional potential outcomes (CDPOs) over time is important in medicine (e.g., to estimate patient-specific risks under different treatment sequences).
In this paper, we aim to learn CDPOs under time-varying treatments using flexible generative models.
Extractive summary: sentences quoted from the sources.