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

Deep Learning of neural networks has progressively become more prominent inhealthcare with models reaching, or even surpassing, expert accuracy levels.However, these success stories are tainted by concerning reports on the lack ofmodel transparency and bias against some medical conditions or patients'sub-groups. Explainable methods are considered the gateway to alleviate many ofthese concerns. In this study we demonstrate that the generated explanationsare volatile to changes in model training that are perpendicular to theclassification task and model structure. This raises further questions abouttrust in deep learning models for healthcare. Mainly, whether the modelscapture underlying causal links in the data or just rely on spuriouscorrelations that are made visible via explanation methods. We demonstrate thatthe output of explainability methods on deep neural networks can varysignificantly by changes of hyper-parameters, such as the random seed or howthe training set is shuffled. We introduce a measure of explanation consistencywhich we use to highlight the identified problems on the MIMIC-CXR dataset. Wefind explanations of identical models but with different training setups have alow consistency: $\approx$ 33% on average. On the contrary, kernel methods arerobust against any orthogonal changes, with explanation consistency at 94%. Weconclude that current trends in model explanation are not sufficient tomitigate the risks of deploying models in real life healthcare applications.

Agree to Disagree: When Deep Learning Models With Identical Architectures Produce Distinct Explanations