English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Selecting causal inference models for estimating individualized treatmenteffects (ITE) from observational data presents a unique challenge since thecounterfactual outcomes are never observed. The problem is challenged furtherin the unsupervised domain adaptation (UDA) setting where we only have accessto labeled samples in the source domain, but desire selecting a model thatachieves good performance on a target domain for which only unlabeled samplesare available. Existing techniques for UDA model selection are designed for thepredictive setting. These methods examine discriminative density ratios betweenthe input covariates in the source and target domain and do not factor in themodel's predictions in the target domain. Because of this, two models withidentical performance on the source domain would receive the same risk score byexisting methods, but in reality, have significantly different performance inthe test domain. We leverage the invariance of causal structures across domainsto propose a novel model selection metric specifically designed for ITE methodsunder the UDA setting. In particular, we propose selecting models whosepredictions of interventions' effects satisfy known causal structures in thetarget domain. Experimentally, our method selects ITE models that are morerobust to covariate shifts on several healthcare datasets, including estimatingthe effect of ventilation in COVID-19 patients from different geographiclocations.

Selecting Treatment Effects Models for Domain Adaptation Using Causal Knowledge