This repository contains R code implementing the simulation experiments from “Estimating Optimal Dynamic Treatment Strategies Under Resource Constraints Using Dynamic Marginal Structural Models”. Relevant generated objects and their interpretations are listed below.
• opt_x_rc2 is the estimated optimal strategy under resource constraints • opt_y_rc2 is the estimated expected outcome under the constrained optimal strategy
• opt_x2 is the estimated unconstrained optimal strategy • opt_y2 is the estimated expected outcome under the unconstrained optimal strategy
• actual_opt_y_x_hat is the actual (computed via Monte Carlo simulation) counterfactual mean outcome under the estimated unconstrained optimal strategy.
• actual_opt_y_x_hat_rc is the actual (computed via Monte Carlo simulation) counterfactual mean outcome under the estimated resource constrained optimal strategy
• actual_naive is the actual (computed via Monte Carlo simulation) counterfactual mean under the strategy: “follow the optimal unconstrained strategy until resources run out, then administer no further treatments”.