Get Started
Home
Topics
Search
Library
Research questionHow does parameter-efficient adapter choice affect pathology accuracy and subgroup disparities when adapting a frozen chest X-ray encoder?Different adapters can change pathology discrimination and protected-attribute representation in a frozen chest X-ray encoder. Overall accuracy or attribute encoding may not reliably indicate performance gaps across subgroups, which can vary by pathology.
AI
Computer Vision
Evaluation & Benchmarks
Health
Machine Learning
Research Paper
Latest papersRecent research connected to this question, newest first.Subgroup performance analysis of adaptation strategies for chest X-ray foundation modelsThe evidence covers three parameter-efficient adapters applied to the frozen Rad-DINO encoder, eight MIMIC-CXR pathologies, and race, sex, and imaging-view subgroups evaluated on a prevalence-preserving, demographically balanced test set. It also examines protected-attribute encoding and attention-pooling layer combinations; the findings do not establish a universal relationship between representation strength, accuracy, and fairness.research paper · Sep 3, 2026
Related questions
How can pathology foundation models be compressed to reduce inference cost without sacrificing classification accuracy or reliability?How can pediatric chest X-ray classifiers preserve usable decisions across countries when calibration and source thresholds shift?How can retinal fundus models remain interpretable and accurate when pathology alters vessel appearance?How can computational pathology distinguish disease signal from acquisition shortcuts in reactive brain tissue?