Objective This study constructed the generalized boosting model combined with propensity score overlap weighting(GBM-OW).Methods Within the situations that there are complex relationships between confounders and treatment factors,and different sample size and different propensity score overlap,we explored the performance of GBM-OW model in balance confounders and estimate effect.And compared with multivariate adjusted model and other three propensity score weighting models.Results and Conclusion From the simulation results we concluded that when the relationship between variables is complex,the sample size is large,and the propensity score value overlap is small,the GBM-OW model has a good performance in all aspects and can be used in observational studies.