<p>Estimation of treatment effects constitutes a critical aspect of research, especially in safety-critical fields such as medicine, economics and health-care in which the uncertainty and estimation is vital for decision-making. By quantifying the impact of interventions, researchers are able to assess the causal relationships between treatments and outcomes, enabling more informed decision-making. In this research, we propose a new tree-based model is proposed, named C-XGBoost (treg), for estimating the potential outcomes as well as the propensity score. A remarkable advanced of the proposed models is that it is trained with a new advanced loss function, which is equipped with targeted regularization procedure that is based on non-parametric estimation theory and aims at reducing bias and improving estimation accuracy. The experimental analysis on three challenging collections of causal inference benchmarks provide strong empirical evidence about the efficiency of the proposed model while the statistical analysis secures the effectiveness and robustness of the proposed approach.</p>