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.
@article{arxiv.2303.13567,
title = {AI Models Close to your Chest: Robust Federated Learning Strategies for Multi-site CT},
author = {Edward H. Lee and Brendan Kelly and Emre Altinmakas and Hakan Dogan and Maryam Mohammadzadeh and Errol Colak and Steve Fu and Olivia Choudhury and Ujjwal Ratan and Felipe Kitamura and Hernan Chaves and Jimmy Zheng and Mourad Said and Eduardo Reis and Jaekwang Lim and Patricia Yokoo and Courtney Mitchell and Golnaz Houshmand and Marzyeh Ghassemi and Ronan Killeen and Wendy Qiu and Joel Hayden and Farnaz Rafiee and Chad Klochko and Nicholas Bevins and Faeze Sazgara and S. Simon Wong and Michael Moseley and Safwan Halabi and Kristen W. Yeom},
journal= {arXiv preprint arXiv:2303.13567},
year = {2023}
}