Ερευνητικό υλικό ΕΚΠΑ

Swarm Learning for decentralized and confidential clinical machine learning

Αγγλικά

Fast and reliable detection of patients with severe and heterogeneous illnesses is a major goal of precision medicine1,2. Patients with leukaemia can be identified using machine learning on the basis of their blood transcriptomes3. However, there is an increasing divide between what is technically possible and what is allowed, because of privacy legislation4,5. 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. © 2021, The Author(s).

2021

Warnat-Herresthal, S.
Schultze, H.
Shastry, K.L.
Manamohan, S.
Mukherjee, S.
Garg, V.
Sarveswara, R.
Händler, K.
Pickkers, P.
Aziz, N.A.
Ktena, S.
Tran, F.
Bitzer, M.
Ossowski, S.
Casadei, N.
Herr, C.
Petersheim, D.
Behrends, U.
Kern, F.
Fehlmann, T.
Schommers, P.
Lehmann, C.
Augustin, M.
Rybniker, J.
Altmüller, J.
Mishra, N.
Bernardes, J.P.
Krämer, B.
Bonaguro, L.
Schulte-Schrepping, J.
De Domenico, E.
Siever, C.
Kraut, M.
Desai, M.
Monnet, B.
Saridaki, M.
Siegel, C.M.
Drews, A.
Nuesch-Germano, M.
Theis, H.
Heyckendorf, J.
Schreiber, S.
Kim-Hellmuth, S.
Balfanz, P.
Eggermann, T.
Boor, P.
Hausmann, R.
Kuhn, H.
Isfort, S.
Stingl, J.C.
Schmalzing, G.
Kuhl, C.K.
Röhrig, R.
Marx, G.
Uhlig, S.
Dahl, E.
Müller-Wieland, D.
Dreher, M.
Marx, N.
Nattermann, J.
Skowasch, D.
Kurth, I.
Keller, A.
Bals, R.
Nürnberg, P.
Rieß, O.
Rosenstiel, P.
Netea, M.G.
Theis, F.
Mukherjee, S.
Backes, M.
Aschenbrenner, A.C.
Ulas, T.
Angelov, A.
Bartholomäus, A.
Becker, A.
Bezdan, D.
Blumert, C.
Bonifacio, E.
Bork, P.
Boyke, B.
Blum, H.
Clavel, T.
Colome-Tatche, M.
Cornberg, M.
De La Rosa Velázquez, I.A.
Diefenbach, A.
Dilthey, A.
Fischer, N.
Förstner, K.
Franzenburg, S.
Frick, J.-S.
Gabernet, G.
Gagneur, J.
Ganzenmueller, T.
Gauder, M.
Geißert, J.
Goesmann, A.
Göpel, S.
Grundhoff, A.
Grundmann, H.
Hain, T.
Hanses, F.
Hehr, U.
Heimbach, A.
Hoeper, M.
Horn, F.
Hübschmann, D.
Hummel, M.
Iftner, T.
Iftner, A.
Illig, T.
Janssen, S.
Kalinowski, J.
Kallies, R.
Kehr, B.
Keppler, O.T.
Klein, C.
Knop, M.
Kohlbacher, O.
Köhrer, K.
Korbel, J.
Kremsner, P.G.
Kühnert, D.
Landthaler, M.
Li, Y.
Ludwig, K.U.
Makarewicz, O.
Marz, M.
McHardy, A.C.
Mertes, C.
Münchhoff, M.
Nahnsen, S.
Nöthen, M.
Ntoumi, F.
Overmann, J.
Peter, S.
Pfeffer, K.
Pink, I.
Poetsch, A.R.
Protzer, U.
Pühler, A.
Rajewsky, N.
Ralser, M.
Reiche, K.
Ripke, S.
da Rocha, U.N.
Saliba, A.-E.
Sander, L.E.
Sawitzki, B.
Scheithauer, S.
Schiffer, P.
Schmid-Burgk, J.
Schneider, W.
Schulte, E.-C.
Sczyrba, A.
Sharaf, M.L.
Singh, Y.
Sonnabend, M.
Stegle, O.
Stoye, J.
Vehreschild, J.
Velavan, T.P.
Vogel, J.
Volland, S.
von Kleist, M.
Walker, A.
Walter, J.
Wieczorek, D.
Winkler, S.
Ziebuhr, J.
Breteler, M.M.B.
Giamarellos-Bourboulis, E.J.
Kox, M.
Becker, M.
Cheran, S.
Woodacre, M.S.
Goh, E.L.
Schultze, J.L.
COVID-19 Aachen Study (COVAS)
Deutsche COVID-19 Omics Initiative (DeCOI)

Nature

Institute of Geographic Sciences and Natural Resources Research

594

7862

265-270

transcriptome, blood; computer simulation; decentralization; detection method; feasibility study; heterogeneity; machine learning; precision, Article; blockchain based peer to peer coordination; blockchain based peer to peer networking; classifier; confidential clinical machine learning; coronavirus disease 2019; decentralized machine learning; edge computing; feasibility study; human; information processing; learning; leukemia; lung disease; machine learning; personalized medicine; prediction; Swarm Learning; thorax radiography; tuberculosis; clinical decision making; confidentiality; epidemic; epidemiology; female; information processing; leukemia; leukocyte; lung disease; male; pathology; procedures; software, Blockchain; Clinical Decision-Making; Confidentiality; COVID-19; Datasets as Topic; Disease Outbreaks; Female; Humans; Leukemia; Leukocytes; Lung Diseases; Machine Learning; Male; Precision Medicine; Software; Tuberculosis

https://doi.org/10.1038/s41586-021-03583-3

10.1038/s41586-021-03583-3

https://pergamos.lib.uoa.gr/uoa/dl/object/3056276