Automatic Urticaria Activity Score: Deep Learning-Based Automatic Hive Counting for Urticaria Severity Assessment

Authors

Taig Mac Carthy, Ignacio Hernández Montilla, Andy Aguilar, Rubén García Castro, Ana María González Pérez, Alejandro Vilas Sueiro, Laura Vergara de la Campa, Fernando Alfageme Roldán, Alfonso Medela.

Abstract

The Urticaria Activity Score is used in clinical care and trials, but manual hive counting is time-consuming and dependent on the observer. AUAS uses a deep-learning lesion-detection model, Legit.Health-UAS-HiveNet, to automate hive counting and support a more reproducible assessment of chronic urticaria severity.

The proof-of-concept study used 313 urticaria images annotated by four specialists. Its knowledge-unification method combined those annotations before the model was trained and evaluated with cross-validation.

The authors reported model performance similar to human performance on this dataset. They also identified the limited dataset and small annotation group as constraints. The study did not evaluate use in clinical workflow or effects on patient outcomes.

Publication

Published in JID Innovations, volume 4, issue 1, article 100218. First published online on 12 July 2023.

Available at https://doi.org/10.1016/j.xjidi.2023.100218.