The aim of this research is to investigate estimators of the average treatment effect
conditional on both prognostic scores (Full Prognostic Score) on control and treated (e.g. regression functions) ψ0(x) and ψ1(x) in the presence of heterogeneous treatment effects. Additionally, we contribute to clarification on the definition of effect modifier. Moreover, the performance of the proposed full prognostic score is reviewed with different estimators such as regression imputation, regression forest, AIPW estimator, and TMLE estimator.