Key points are not available for this paper at this time.
L1-regularization incorporates all covariates simultaneously into the propensity score model and offers propensity score performance superior to the hdPS marginal screen.
Tian et al. (Thu,) studied this question.