Collaborative-controlled LASSO for constructing propensity score-based estimators in high-dimensional data

Abstract

Propensity score-based estimators are increasingly used for causal inference in observational studies. However, model selection for propensity score estimation in high-dimensional data has received little attention. In these settings, propensity score models have traditionally been selected based on the goodness-of-fit for the treatment mechanism itself, without consideration of the causal parameter of interest. Collaborative minimum loss-based estimation is a novel methodology for causal inference that takes into account information on the causal parameter of interest when selecting a propensity score model. This ‘‘collaborative learning’’ considers variable associations with both treatment and outcome when selecting a propensity score model in order to minimize a bias-variance tradeoff in the estimated treatment effect. In this study, we introduce a novel approach for collaborative model selection when using the LASSO estimator for propensity score estimation in high-dimensional covariate settings. To demonstrate the importance of selecting the propensity score model collaboratively, we designed quasi-experiments based on a real electronic healthcare database, where only the potential outcomes were manually generated, and the treatment and baseline covariates remained unchanged. Results showed that the collaborative minimum loss-based estimation algorithm outperformed other competing estimators for both point estimation and confidence interval coverage. In addition, the propensity score model selected by collaborative minimum loss-based estimation could be applied to other propensity score-based estimators, which also resulted in substantive improvement for both point estimation and confidence interval coverage. We illustrate the discussed concepts through an empirical example comparing the effects of non-selective nonsteroidal anti-inflammatory drugs with selective COX-2 inhibitors on gastrointestinal complications in a population of Medicare beneficiaries.

Confounder selection strategies targeting stable treatment effect estimators

“…However, such an approach merely conditions on the (selected) covariates, and does not account for the uncertainty induced by the (data-driven) covariate selection procedure. In general, existing postselection inference procedures (see, eg, Berk et al, 55 Chernozhukov et al, 56 and Ju et al, 43 among others) typically consider regularized regression methods, and share the limitation that they lack finite-sample guarantees and attain their desired theoretical properties only for large sample sizes. Various aspects of our proposal (ie, using double selection to prioritize covariates, focusing on the stability of the marginal effect estimator, and employing randomization inference) were chosen so that standard inference would deliver reasonable approximations in finite samples.…”

“…M6 A Wald test using CTMLE with LASSO by fitting a penalized logistic regression model to estimate the PS model. 43 Instead of a forward selection approach, the sequence of candidate PS models is now indexed by different values of the LASSO regularization penalty. As in the discrete CTMLE approach above, the PS model that minimizes the (cross-validated) loss function of the treatment effect estimator is chosen.…”

On adaptive propensity score truncation in causal inference

“…Recently, developed C-TMLE algorithms for continues tuning parameter, with the general theorem of the asymptotic normality of the resulting C-TMLE estimators. Based on this work, further proposed LASSO-C-TMLE, where the PS is estimated by LASSO controlled by C-TMLE, and [Ju et al, 2017e] demonstrated its performance on high-dimensional electronic health dataset. We simply consider the truncation quantile γ as a tuning parameter, and extend the C-TMLE algorithm to select the optimal γ for the estimation of the causal parameter.…”

Identification of a ferroptosis-related gene signature predictive model in colon cancer

“…cbioportal.org) [18]. Least absolute shrinkage and selection operator (LASSO) regression analysis, a common method of performing regression analysis with high dimensional factors, was used to construct a prognostic model to minimize the level of overfitting [19,20]. Three-fold cross-validation was conducted to reduce the potential instability of the results, and the optimal tuning parameter λ was identified based on a 1-SE (standard error) standard.…”