Swarm Learning for decentralized and confidential clinical machine learning

Warnat-Herresthal, Stefanie and Schultze, Hartmut and Shastry, Krishnaprasad Lingadahalli and Manamohan, Sathyanarayanan and Mukherjee, Saikat and Garg, Vishesh and Sarveswara, Ravi and Händler, Kristian and Pickkers, Peter and Aziz, N. Ahmad and Ktena, Sofia and Tran, Florian and Bitzer, Michael and Ossowski, Stephan and Casadei, Nicolas and Herr, Christian and Petersheim, Daniel and Behrends, Uta and Kern, Fabian and Fehlmann, Tobias and Schommers, Philipp and Lehmann, Clara and Augustin, Max and Rybniker, Jan and Altmüller, Janine and Mishra, Neha and Bernardes, Joana P. and Krämer, Benjamin and Bonaguro, Lorenzo and Schulte-Schrepping, Jonas and De Domenico, Elena and Siever, Christian and Kraut, Michael and Desai, Milind and Monnet, Bruno and Saridaki, Maria and Siegel, Charles Martin and Drews, Anna and Nuesch-Germano, Melanie and Theis, Heidi and Heyckendorf, Jan and Schreiber, Stefan and Kim-Hellmuth, Sarah and Nattermann, Jacob and Skowasch, Dirk and Kurth, Ingo and Keller, Andreas and Bals, Robert and Nürnberg, Peter and Rieß, Olaf and Rosenstiel, Philip and Netea, Mihai G. and Theis, Fabian and Mukherjee, Sach and Backes, Michael and Aschenbrenner, Anna C. and Ulas, Thomas and Breteler, Monique M. B. and Giamarellos-Bourboulis, Evangelos J. and Kox, Matthijs and Becker, Matthias and Cheran, Sorin and Woodacre, Michael S. and Goh, Eng Lim and Schultze, Joachim L.
(2021) Swarm Learning for decentralized and confidential clinical machine learning.
Nature.

[img] Text
s41586-021-03583-3.pdf

Download (12MB)

Abstract

Fast and reliable detection of patients with severe and heterogeneous illnesses is a major goal of precision medicine. Patients with leukaemia can be identified using machine learning on the basis of their blood transcriptomes. However, there is an increasing divide between what is technically possible and what is allowed, because of privacy legislation. Here, to facilitate the integration of any medical data from any data owner worldwide without violating privacy laws, we introduce Swarm Learning—a decentralized machine-learning approach that unites edge computing, blockchain-based peer-to-peer networking and coordination while maintaining confidentiality without the need for a central coordinator, thereby going beyond federated learning. To illustrate the feasibility of using Swarm Learning to develop disease classifiers using distributed data, we chose four use cases of heterogeneous diseases (COVID-19, tuberculosis, leukaemia and lung pathologies). With more than 16,400 blood transcriptomes derived from 127 clinical studies with non-uniform distributions of cases and controls and substantial study biases, as well as more than 95,000 chest X-ray images, we show that Swarm Learning classifiers outperform those developed at individual sites. In addition, Swarm Learning completely fulfils local confidentiality regulations by design. We believe that this approach will notably accelerate the introduction of precision medicine.

Actions

Actions (login required)

View Item View Item