Swarm learning for decentralized artificial intelligence in cancer histopathology

TLDR

AI can predict molecular alterations from routine histopathology slides, but robust training requires large datasets that are hindered by practical, ethical, and legal obstacles. We demonstrate that swarm learning enables training of AI models on large, multicentric gigapixel histopathology datasets from over 5,000 patients, and propose its future use to train distributed models without data transfer. Swarm learning allows partners to jointly train AI models while avoiding data transfer and monopolistic governance; we applied it to three cohorts from Northern Ireland, Germany, and the US, validating performance on two independent UK datasets. SL‑trained AI models accurately predict BRAF mutation and microsatellite instability from H&E colorectal cancer slides, outperform most locally trained models, match performance of merged‑dataset models, and are data efficient.

Abstract

Abstract Artificial intelligence (AI) can predict the presence of molecular alterations directly from routine histopathology slides. However, training robust AI systems requires large datasets for which data collection faces practical, ethical and legal obstacles. These obstacles could be overcome with swarm learning (SL), in which partners jointly train AI models while avoiding data transfer and monopolistic data governance. Here, we demonstrate the successful use of SL in large, multicentric datasets of gigapixel histopathology images from over 5,000 patients. We show that AI models trained using SL can predict BRAF mutational status and microsatellite instability directly from hematoxylin and eosin (H&E)-stained pathology slides of colorectal cancer. We trained AI models on three patient cohorts from Northern Ireland, Germany and the United States, and validated the prediction performance in two independent datasets from the United Kingdom. Our data show that SL-trained AI models outperform most locally trained models, and perform on par with models that are trained on the merged datasets. In addition, we show that SL-based AI models are data efficient. In the future, SL can be used to train distributed AI models for any histopathology image analysis task, eliminating the need for data transfer.