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

Predictive models for clinical outcomes that are accurate on average in apatient population may underperform drastically for some subpopulations,potentially introducing or reinforcing inequities in care access and quality.Model training approaches that aim to maximize worst-case model performanceacross subpopulations, such as distributionally robust optimization (DRO),attempt to address this problem without introducing additional harms. Weconduct a large-scale empirical study of DRO and several variations of standardlearning procedures to identify approaches for model development and selectionthat consistently improve disaggregated and worst-case performance oversubpopulations compared to standard approaches for learning predictive modelsfrom electronic health records data. In the course of our evaluation, weintroduce an extension to DRO approaches that allows for specification of themetric used to assess worst-case performance. We conduct the analysis formodels that predict in-hospital mortality, prolonged length of stay, and 30-dayreadmission for inpatient admissions, and predict in-hospital mortality usingintensive care data. We find that, with relatively few exceptions, no approachperforms better, for each patient subpopulation examined, than standardlearning procedures using the entire training dataset. These results imply thatwhen it is of interest to improve model performance for patient subpopulationsbeyond what can be achieved with standard practices, it may be necessary to doso via data collection techniques that increase the effective sample size orreduce the level of noise in the prediction problem.

A comparison of approaches to improve worst-case predictive model performance over patient subpopulations