Larsvanderlaan/npOddsRatio: Semiparametric Targeted Maximum-Likelihood estimation of the conditional odds ratio in a partially-linear logistic-link regression model with post-treatment informative missingness.

Utilizing the framework of Targeted Maximum-Likelihood estimation (TMLE), efficient estimates and inference is given for the conditional odds ratio in a partially-linear logistic-link regression model with post-treatment informative missingness. The user can supply an arbitrary parametric form of the conditional log odds ratio. Estimates and confidence intervals are returned. Nuisance functions can be estimated using machine-learning (specifically the machine-learning pipeline R package sl3) ,thereby avoiding biases due to model misspecification. This package allows for the outcome missingness to informed by pre-treatment variables W and the treatment (See function npOR), or through pre-treatment variables, treatment, and post-treatment variables (see function npORMissing).

install.packages("remotes")
remotes::install_github("Larsvanderlaan/npOddsRatio")