clone-censor-ipcw approach "1. Create an augmented dataset by replicating individuals; 2. Censor replicates at tx strategy violation; 3. Fit censoring model for ipcw, 4. outcome modeling" https://t.co/59du5niUHV https://t.co/VrIb4Ch0gX
Statistical Methods for Dynamic Treatment Regimes https://t.co/ZqM88rSyJO The outcome model has to condition on baseline covariates even though there is no baseline confounding after cloning everyone to every arm because baseline covariates are in the IPCW
https://t.co/TzNyezawH9 "Comparing multiple dynamic treatment regimes in a MSM framework requires first the realization that an individual study participant’s observed treatment history may be compatible with several treatment regimes, at least for some pa
Bibhas Chakraborty, Erica E.M. Moodie - Google Books https://t.co/J1qVIQAC8n "Note that unlike Q-learning or G-estimation, which are performed recursively in time (backwards induction methods), G-computation models and then simulates data forward in time."