Reviews: Forecasting Treatment Responses Over Time Using Recurrent Marginal Structural Networks

The authors rebuttal was very helpful and could clarify many questions. Especially, the novel contribution of learning a model for treatment effects over time made as improving our ratings. Treatment response forecasting always suffers from time-dependent confounders, making the exploration of causality very difficult. The standard approach to deal with this problem are Marginal Structural Models (MSM). However, these methods strongly rely on the accuracy of manual estimation steps.