AI Models Close to your Chest: Robust Federated Learning Strategies for Multi-site CT
This work addresses the challenge of developing robust AI models for medical imaging across diverse populations while preserving data privacy, though it is incremental in improving federated learning methods.
The study tackled the problem of training AI models on diverse medical data without sharing patient data by evaluating federated learning strategies on a large, multi-site COVID-19 chest CT dataset, achieving results that highlight performance disparities due to data heterogeneity. It proposed a strategy using synthetic data to address class and size imbalances.
While it is well known that population differences from genetics, sex, race, and environmental factors contribute to disease, AI studies in medicine have largely focused on locoregional patient cohorts with less diverse data sources. Such limitation stems from barriers to large-scale data share and ethical concerns over data privacy. Federated learning (FL) is one potential pathway for AI development that enables learning across hospitals without data share. In this study, we show the results of various FL strategies on one of the largest and most diverse COVID-19 chest CT datasets: 21 participating hospitals across five continents that comprise >10,000 patients with >1 million images. We also propose an FL strategy that leverages synthetically generated data to overcome class and size imbalances. We also describe the sources of data heterogeneity in the context of FL, and show how even among the correctly labeled populations, disparities can arise due to these biases.