Privacy-Preserving Fundus Disease Diagnosis Using Federated Learning

Overview

In recent years, advances in the diagnosis and prediction of diseases using machine learning (ML) has been growing exponentially. However, due to the many privacy regulations regarding personal data, it's often infeasible to share data from multiple sources and store them in a single (centralized) location for traditional ML model training. Federated Learning (FL), a collaborative learning paradigm, is able to sidestep this major pitfall by allowing the creation of a global ML model that is trained by aggregating model weights from individual models that are separately trained on their own data silos, therefore avoiding any data privacy concerns. This study aims to address the centralized data issue by applying a novel FL approach to fundus disease diagnosis from ophthalmic images. This study proposes a novel method for aggregating model weights by comparing the size of each model's data and taking the weighted average of all the models' weights. Experimental results showed only a slight 2.04% decrease in accuracy when training using FL (84.88%) compared to a centralized ML model (86.63%). This study shows that FL is able to achieve maximum privacy for ML in fundus disease diagnosis while only compromising a miniscule amount of accuracy.

Competition history

• AJAS 2025 Category not listed

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Browse more like this

• All Biomedical and Health Sciences projects
• AJAS Biomedical and Health Sciences projects
• Every AJAS 2025 project