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

We study the problem of inferring heterogeneous treatment effects fromtime-to-event data. While both the related problems of (i) estimating treatmenteffects for binary or continuous outcomes and (ii) predicting survival outcomeshave been well studied in the recent machine learning literature, theircombination -- albeit of high practical relevance -- has received considerablyless attention. With the ultimate goal of reliably estimating the effects oftreatments on instantaneous risk and survival probabilities, we focus on theproblem of learning (discrete-time) treatment-specific conditional hazardfunctions. We find that unique challenges arise in this context due to avariety of covariate shift issues that go beyond a mere combination ofwell-studied confounding and censoring biases. We theoretically analyse theireffects by adapting recent generalization bounds from domain adaptation andtreatment effect estimation to our setting and discuss implications for modeldesign. We use the resulting insights to propose a novel deep learning methodfor treatment-specific hazard estimation based on balancing representations. Weinvestigate performance across a range of experimental settings and empiricallyconfirm that our method outperforms baselines by addressing covariate shiftsfrom various sources.

SurvITE: Learning Heterogeneous Treatment Effects from Time-to-Event Data