AI Paper Atlas

Research map / Probabilistic and scientific modeling / Treatment effects

Efficient augmentation and relaxation learning for individualized treatment rules using observational data

Ying-Qi Zhao, Eric B. Laber, Yang Ning, Sumona Saha, Bruce E. Sands

JMLR 2019 · Probabilistic and scientific modeling › Treatment effects · approach: Benchmarking treatment effect models · 81 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add outcome regression augmentation (EARL) instead of IPWE-only outcome weighted learningEARL · Simulation (Scenarios 1-3)value of estimated ruleImproved
use convex relaxation (EARL) instead of direct AIPWE maximization via genetic algorithmAIPWE value search · Simulationbias, variance and runtime: approximately 200 times slowerImproved
correctly specify the Q-function model in Q-learning (ablation)Q-learning · Simulationvalue of estimated ruleMixed
misspecify the Q-function model rather than the propensity model (ablation)EARL and AIPWE · Simulationvalue of estimated ruleHurt
bound propensity scores away from zero (Scenario 2 versus Scenario 1) (ablation)EARL · Simulationvalue of estimated ruleImproved
set propensity to constant 0.025 with correct propensity but wrong outcome model (CI) versus the reverse (IC) (ablation)EARL · Simulation Scenario 3value of estimated rule: CI outperformed IC, especially at small sample sizeImproved
use nonparametric working models for propensity score instead of parametric (ablation)EARL · Simulation (Web Appendix F)value of estimated ruleImproved
use hinge loss instead of smooth surrogate losses when the regression model is incorrect (ablation)EARL · Simulation (Web Appendix F)value of estimated ruleImproved
use EARL instead of observed treatment allocationEARL logistic · OSCCAR IBD registryestimated disease activity score: 2.24 -> 1.75Improved
use EARL logistic instead of Q-learningEARL logistic · OSCCAR IBD registryestimated disease activity score: 1.80 -> 1.75Improved
use EARL logistic instead of OWL logisticEARL logistic · OSCCAR IBD registryestimated disease activity score and patients on top-down: 1.75 vs 1.75; 161 vs 225 patients recommended top-downNo clear effect
switch EARL surrogate loss among logistic, hinge, squared hinge, exponential (ablation)EARL · OSCCAR IBD registryrule agreement: agreement 0.996 between logit and hinge/sqhinge; 0.588 with exponentialNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy