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

In health related machine learning applications, the training data oftencorresponds to a non-representative sample from the target populations wherethe learners will be deployed. In anticausal prediction tasks, selection biasesoften make the associations between confounders and the outcome variableunstable across different target environments. As a consequence, thepredictions from confounded learners are often unstable, and might fail togeneralize in shifted test environments. Stable prediction approaches aim tosolve this problem by producing predictions that are stable across unknown testenvironments. These approaches, however, are sometimes applied to the trainingdata alone with the hope that training an unconfounded model will be enough togenerate stable predictions in shifted test sets. Here, we show that this isinsufficient, and that improved stability can be achieved by deconfounding thetest set features as well. We illustrate these observations using bothsynthetic data and real world data from a mobile health study.

Stable predictions for health related anticausal prediction tasks affected by selection biases: the need to deconfound the test set features