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 medical image diagnosis, fairness has become increasingly crucial. Withoutbias mitigation, deploying unfair AI would harm the interests of theunderprivileged population and potentially tear society apart. Recent researchaddresses prediction biases in deep learning models concerning demographicgroups (e.g., gender, age, and race) by utilizing demographic (sensitiveattribute) information during training. However, many sensitive attributesnaturally exist in dermatological disease images. If the trained model onlytargets fairness for a specific attribute, it remains unfair for otherattributes. Moreover, training a model that can accommodate multiple sensitiveattributes is impractical due to privacy concerns. To overcome this, we proposea method enabling fair predictions for sensitive attributes during the testingphase without using such information during training. Inspired by prior workhighlighting the impact of feature entanglement on fairness, we enhance themodel features by capturing the features related to the sensitive and targetattributes and regularizing the feature entanglement between correspondingclasses. This ensures that the model can only classify based on the featuresrelated to the target attribute without relying on features associated withsensitive attributes, thereby improving fairness and accuracy. Additionally, weuse disease masks from the Segment Anything Model (SAM) to enhance the qualityof the learned feature. Experimental results demonstrate that the proposedmethod can improve fairness in classification compared to state-of-the-artmethods in two dermatological disease datasets.

Achieve Fairness without Demographics for Dermatological Disease Diagnosis