The FeatureCloud Platform for Federated Learning in Biomedicine: Unified Approach

Julian Matschinske; Julian Späth; Mohammad Bakhtiari; Niklas Probul; Mohammad Mahdi Kazemi Majdabadi; Reza Nasirigerdeh; Reihaneh Torkzadehmahani; Anne Hartebrodt; Balazs Attila Orban; Sándor József Fejér; Olga Zolotareva; Supratim Das; Linda Baumbach; Josch K. Pauling; Olivera Tomašević; Béla Bihari; Marcus Bloice; Nina C. Donner; Walid Fdhila; Tobias Frisch; Anne Christin Hauschild; Dominik Heider; Andreas Holzinger; Walter Hötzendorfer; Jan Hospes; Tim Kacprowski; Markus Kastelitz; Markus List; Rudolf Mayer; Mónika Moga; Heimo Müller; Anastasia Pustozerova; Richard Röttger; Christina C. Saak; Anna Saranti; Harald H.H.W. Schmidt; Christof Tschohl; Nina K. Wenke; Jan Baumbach

doi:10.2196/42621

The FeatureCloud Platform for Federated Learning in Biomedicine: Unified Approach

Julian Matschinske^*
, Julian Späth
, Mohammad Bakhtiari
, Niklas Probul
, Mohammad Mahdi Kazemi Majdabadi
, Reza Nasirigerdeh
, Reihaneh Torkzadehmahani
, Anne Hartebrodt
, Balazs Attila Orban
, Sándor József Fejér
, Olga Zolotareva
, Supratim Das
, Linda Baumbach
, Josch K. Pauling
, Olivera Tomašević
, Béla Bihari
, Marcus Bloice
, Nina C. Donner
, Walid Fdhila
, Tobias Frisch

Anne Christin Hauschild, Dominik Heider, Andreas Holzinger, Walter Hötzendorfer, Jan Hospes, Tim Kacprowski, Markus Kastelitz, Markus List, Rudolf Mayer, Mónika Moga, Heimo Müller, Anastasia Pustozerova, Richard Röttger, Christina C. Saak, Anna Saranti, Harald H.H.W. Schmidt, Christof Tschohl, Nina K. Wenke, Jan Baumbach

^*Corresponding author for this work

University of Hamburg
Technical University of Munich
Gnome Design SRL
University of Novi Sad
Medical University of Graz
Concentris Research Management GGmbH
SBA Research gGmbH
Philipps University of Marburg
Research Institute AG & Co KG
Braunschweig University of Technology
Maastricht University
University Medical Center Hamburg-Eppendorf
Technical University Munich
University Medical Center Göttingen

Research output: Contribution to journal › Journal article › Research › peer-review

63 Downloads (Pure)

Abstract

Background: Machine learning and artificial intelligence have shown promising results in many areas and are driven by the increasing amount of available data. However, these data are often distributed across different institutions and cannot be easily shared owing to strict privacy regulations. Federated learning (FL) allows the training of distributed machine learning models without sharing sensitive data. In addition, the implementation is time-consuming and requires advanced programming skills and complex technical infrastructures. Objective: Various tools and frameworks have been developed to simplify the development of FL algorithms and provide the necessary technical infrastructure. Although there are many high-quality frameworks, most focus only on a single application case or method. To our knowledge, there are no generic frameworks, meaning that the existing solutions are restricted to a particular type of algorithm or application field. Furthermore, most of these frameworks provide an application programming interface that needs programming knowledge. There is no collection of ready-to-use FL algorithms that are extendable and allow users (eg, researchers) without programming knowledge to apply FL. A central FL platform for both FL algorithm developers and users does not exist. This study aimed to address this gap and make FL available to everyone by developing FeatureCloud, an all-in-one platform for FL in biomedicine and beyond. Methods: The FeatureCloud platform consists of 3 main components: a global frontend, a global backend, and a local controller. Our platform uses a Docker to separate the local acting components of the platform from the sensitive data systems. We evaluated our platform using 4 different algorithms on 5 data sets for both accuracy and runtime. Results: FeatureCloud removes the complexity of distributed systems for developers and end users by providing a comprehensive platform for executing multi-institutional FL analyses and implementing FL algorithms. Through its integrated artificial intelligence store, federated algorithms can easily be published and reused by the community. To secure sensitive raw data, FeatureCloud supports privacy-enhancing technologies to secure the shared local models and assures high standards in data privacy to comply with the strict General Data Protection Regulation. Our evaluation shows that applications developed in FeatureCloud can produce highly similar results compared with centralized approaches and scale well for an increasing number of participating sites. Conclusions: FeatureCloud provides a ready-to-use platform that integrates the development and execution of FL algorithms while reducing the complexity to a minimum and removing the hurdles of federated infrastructure. Thus, we believe that it has the potential to greatly increase the accessibility of privacy-preserving and distributed data analyses in biomedicine and beyond.

Original language	English
Article number	e42621
Journal	Journal of Medical Internet Research
Volume	25
Number of pages	16
ISSN	1438-8871
DOIs	https://doi.org/10.2196/42621
Publication status	Published - 2023

Funding

The FeatureCloud project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement 826078. This publication reflects only the authors’ view, and the European Commission is not responsible for any use that may be made of the information it contains. This work was further supported by the German Federal Ministry of Education and Research (BMBF)–funded German Network for Bioinformatics Infrastructure (de.NBI) Cloud within the de.NBI (031A532B, 031A533A, 031A533B, 031A534A, 031A535A, 031A537A, 031A537B, 031A537C, 031A537D, and 031A538A). This work was further supported by the German BMBF within the framework of “CLINSPECT-M” (grant FKZ161L0214A). This work was partly developed as part of the Center for Female Artificial Intelligence project funded by the German BMBF under grant 01IS21079. This work was further funded by the German BMBF under grant 16DTM100A. JKP was funded by the Bavarian State Ministry of Education and the Arts in the framework of the Bavarian Research Institute for Digital Transformation (bidt, grant LipiTUM). This study uses data from the Survey of Health, Aging and Retirement in Europe (SHARE) Wave 8 (refer to the study by Börsch-Supan et al [38] for methodological details). The SHARE data collection has been funded by the European Commission, Directorate-General for Research and Innovation (DG RTD) through FP5 (QLK6-CT-2001-00360), FP6 (SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, and SHARELIFE: CIT4-CT-2006-028812), FP7 (SHARE-PREP: GA N°211909, SHARE-LEAP: GA N°227822, SHARE M4: GA N°261982, and Data Service Infrastructure for the Social Sciences and Humanities (DASISH): GA N°283646), and Horizon 2020 (SHARE-DEV3: GA N°676536, SHARE-COHESION: GA N°870628, SERISS: GA N°654221, SSHOC: GA N°823782, and SHARE-COVID19: GA N°101015924); and by DG Employment, Social Affairs & Inclusion through VS 2015/0195, VS 2016/0135, VS 2018/0285, VS 2019/0332, and VS 2020/0313. Additional funding from the German Ministry of Education and Research, the Max Planck Society for the Advancement of Science, the US National Institute on Aging (U01_AG09740-13S2, P01_AG005842, P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11, OGHA_04-064, HHSN271201300071C, and RAG052527A), and various national funding sources is gratefully acknowledged [46].

Keywords

additive secret sharing
AI store
artificial intelligence
federated learning
interactive platform
privacy-enhancing technologies
privacy-preserving machine learning

Documents & Links

10.2196/42621Licence: CC BY

Open Access VersionFinal published version, 866 KBLicence: CC BY

SDU Link

Check access to the material in SDU Library's search engine

Cite this

Matschinske, J., Späth, J., Bakhtiari, M., Probul, N., Majdabadi, M. M. K., Nasirigerdeh, R., Torkzadehmahani, R., Hartebrodt, A., Orban, B. A., Fejér, S. J., Zolotareva, O., Das, S., Baumbach, L., Pauling, J. K., Tomašević, O., Bihari, B., Bloice, M., Donner, N. C., Fdhila, W., ... Baumbach, J. (2023). The FeatureCloud Platform for Federated Learning in Biomedicine: Unified Approach. Journal of Medical Internet Research, 25, Article e42621. https://doi.org/10.2196/42621

@article{74ffd45881a14a5abda366102c86b11c,

title = "The FeatureCloud Platform for Federated Learning in Biomedicine: Unified Approach",

abstract = "Background: Machine learning and artificial intelligence have shown promising results in many areas and are driven by the increasing amount of available data. However, these data are often distributed across different institutions and cannot be easily shared owing to strict privacy regulations. Federated learning (FL) allows the training of distributed machine learning models without sharing sensitive data. In addition, the implementation is time-consuming and requires advanced programming skills and complex technical infrastructures. Objective: Various tools and frameworks have been developed to simplify the development of FL algorithms and provide the necessary technical infrastructure. Although there are many high-quality frameworks, most focus only on a single application case or method. To our knowledge, there are no generic frameworks, meaning that the existing solutions are restricted to a particular type of algorithm or application field. Furthermore, most of these frameworks provide an application programming interface that needs programming knowledge. There is no collection of ready-to-use FL algorithms that are extendable and allow users (eg, researchers) without programming knowledge to apply FL. A central FL platform for both FL algorithm developers and users does not exist. This study aimed to address this gap and make FL available to everyone by developing FeatureCloud, an all-in-one platform for FL in biomedicine and beyond. Methods: The FeatureCloud platform consists of 3 main components: a global frontend, a global backend, and a local controller. Our platform uses a Docker to separate the local acting components of the platform from the sensitive data systems. We evaluated our platform using 4 different algorithms on 5 data sets for both accuracy and runtime. Results: FeatureCloud removes the complexity of distributed systems for developers and end users by providing a comprehensive platform for executing multi-institutional FL analyses and implementing FL algorithms. Through its integrated artificial intelligence store, federated algorithms can easily be published and reused by the community. To secure sensitive raw data, FeatureCloud supports privacy-enhancing technologies to secure the shared local models and assures high standards in data privacy to comply with the strict General Data Protection Regulation. Our evaluation shows that applications developed in FeatureCloud can produce highly similar results compared with centralized approaches and scale well for an increasing number of participating sites. Conclusions: FeatureCloud provides a ready-to-use platform that integrates the development and execution of FL algorithms while reducing the complexity to a minimum and removing the hurdles of federated infrastructure. Thus, we believe that it has the potential to greatly increase the accessibility of privacy-preserving and distributed data analyses in biomedicine and beyond.",

keywords = "additive secret sharing, AI store, artificial intelligence, federated learning, interactive platform, privacy-enhancing technologies, privacy-preserving machine learning",

author = "Julian Matschinske and Julian Sp{\"a}th and Mohammad Bakhtiari and Niklas Probul and Majdabadi, \{Mohammad Mahdi Kazemi\} and Reza Nasirigerdeh and Reihaneh Torkzadehmahani and Anne Hartebrodt and Orban, \{Balazs Attila\} and Fej{\'e}r, \{S{\'a}ndor J{\'o}zsef\} and Olga Zolotareva and Supratim Das and Linda Baumbach and Pauling, \{Josch K.\} and Olivera Toma{\v s}evi{\'c} and B{\'e}la Bihari and Marcus Bloice and Donner, \{Nina C.\} and Walid Fdhila and Tobias Frisch and Hauschild, \{Anne Christin\} and Dominik Heider and Andreas Holzinger and Walter H{\"o}tzendorfer and Jan Hospes and Tim Kacprowski and Markus Kastelitz and Markus List and Rudolf Mayer and M{\'o}nika Moga and Heimo M{\"u}ller and Anastasia Pustozerova and Richard R{\"o}ttger and Saak, \{Christina C.\} and Anna Saranti and Schmidt, \{Harald H.H.W.\} and Christof Tschohl and Wenke, \{Nina K.\} and Jan Baumbach",

note = "Funding Information: The FeatureCloud project has received funding from the European Union{\textquoteright}s Horizon 2020 research and innovation program under grant agreement 826078. This publication reflects only the authors{\textquoteright} view, and the European Commission is not responsible for any use that may be made of the information it contains. This work was further supported by the German Federal Ministry of Education and Research (BMBF){\textendash}funded German Network for Bioinformatics Infrastructure (de.NBI) Cloud within the de.NBI (031A532B, 031A533A, 031A533B, 031A534A, 031A535A, 031A537A, 031A537B, 031A537C, 031A537D, and 031A538A). This work was further supported by the German BMBF within the framework of {\textquotedblleft}CLINSPECT-M{\textquotedblright} (grant FKZ161L0214A). This work was partly developed as part of the Center for Female Artificial Intelligence project funded by the German BMBF under grant 01IS21079. This work was further funded by the German BMBF under grant 16DTM100A. JKP was funded by the Bavarian State Ministry of Education and the Arts in the framework of the Bavarian Research Institute for Digital Transformation (bidt, grant LipiTUM). This study uses data from the Survey of Health, Aging and Retirement in Europe (SHARE) Wave 8 (refer to the study by B{\"o}rsch-Supan et al [38] for methodological details). The SHARE data collection has been funded by the European Commission, Directorate-General for Research and Innovation (DG RTD) through FP5 (QLK6-CT-2001-00360), FP6 (SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, and SHARELIFE: CIT4-CT-2006-028812), FP7 (SHARE-PREP: GA N°211909, SHARE-LEAP: GA N°227822, SHARE M4: GA N°261982, and Data Service Infrastructure for the Social Sciences and Humanities (DASISH): GA N°283646), and Horizon 2020 (SHARE-DEV3: GA N°676536, SHARE-COHESION: GA N°870628, SERISS: GA N°654221, SSHOC: GA N°823782, and SHARE-COVID19: GA N°101015924); and by DG Employment, Social Affairs \& Inclusion through VS 2015/0195, VS 2016/0135, VS 2018/0285, VS 2019/0332, and VS 2020/0313. Additional funding from the German Ministry of Education and Research, the Max Planck Society for the Advancement of Science, the US National Institute on Aging (U01\_AG09740-13S2, P01\_AG005842, P01\_AG08291, P30\_AG12815, R21\_AG025169, Y1-AG-4553-01, IAG\_BSR06-11, OGHA\_04-064, HHSN271201300071C, and RAG052527A), and various national funding sources is gratefully acknowledged [46].",

year = "2023",

doi = "10.2196/42621",

language = "English",

volume = "25",

journal = "Journal of Medical Internet Research",

issn = "1438-8871",

publisher = "JMIR Publications",

}

Matschinske, J, Späth, J, Bakhtiari, M, Probul, N, Majdabadi, MMK, Nasirigerdeh, R, Torkzadehmahani, R, Hartebrodt, A, Orban, BA, Fejér, SJ, Zolotareva, O, Das, S, Baumbach, L, Pauling, JK, Tomašević, O, Bihari, B, Bloice, M, Donner, NC, Fdhila, W, Frisch, T, Hauschild, AC, Heider, D, Holzinger, A, Hötzendorfer, W, Hospes, J, Kacprowski, T, Kastelitz, M, List, M, Mayer, R, Moga, M, Müller, H, Pustozerova, A, Röttger, R, Saak, CC, Saranti, A, Schmidt, HHHW, Tschohl, C, Wenke, NK & Baumbach, J 2023, 'The FeatureCloud Platform for Federated Learning in Biomedicine: Unified Approach', Journal of Medical Internet Research, vol. 25, e42621. https://doi.org/10.2196/42621

TY - JOUR

T1 - The FeatureCloud Platform for Federated Learning in Biomedicine

T2 - Unified Approach

AU - Matschinske, Julian

AU - Späth, Julian

AU - Bakhtiari, Mohammad

AU - Probul, Niklas

AU - Majdabadi, Mohammad Mahdi Kazemi

AU - Nasirigerdeh, Reza

AU - Torkzadehmahani, Reihaneh

AU - Hartebrodt, Anne

AU - Orban, Balazs Attila

AU - Fejér, Sándor József

AU - Zolotareva, Olga

AU - Das, Supratim

AU - Baumbach, Linda

AU - Pauling, Josch K.

AU - Tomašević, Olivera

AU - Bihari, Béla

AU - Bloice, Marcus

AU - Donner, Nina C.

AU - Fdhila, Walid

AU - Frisch, Tobias

AU - Hauschild, Anne Christin

AU - Heider, Dominik

AU - Holzinger, Andreas

AU - Hötzendorfer, Walter

AU - Hospes, Jan

AU - Kacprowski, Tim

AU - Kastelitz, Markus

AU - List, Markus

AU - Mayer, Rudolf

AU - Moga, Mónika

AU - Müller, Heimo

AU - Pustozerova, Anastasia

AU - Röttger, Richard

AU - Saak, Christina C.

AU - Saranti, Anna

AU - Schmidt, Harald H.H.W.

AU - Tschohl, Christof

AU - Wenke, Nina K.

AU - Baumbach, Jan

N1 - Funding Information: The FeatureCloud project has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement 826078. This publication reflects only the authors’ view, and the European Commission is not responsible for any use that may be made of the information it contains. This work was further supported by the German Federal Ministry of Education and Research (BMBF)–funded German Network for Bioinformatics Infrastructure (de.NBI) Cloud within the de.NBI (031A532B, 031A533A, 031A533B, 031A534A, 031A535A, 031A537A, 031A537B, 031A537C, 031A537D, and 031A538A). This work was further supported by the German BMBF within the framework of “CLINSPECT-M” (grant FKZ161L0214A). This work was partly developed as part of the Center for Female Artificial Intelligence project funded by the German BMBF under grant 01IS21079. This work was further funded by the German BMBF under grant 16DTM100A. JKP was funded by the Bavarian State Ministry of Education and the Arts in the framework of the Bavarian Research Institute for Digital Transformation (bidt, grant LipiTUM). This study uses data from the Survey of Health, Aging and Retirement in Europe (SHARE) Wave 8 (refer to the study by Börsch-Supan et al [38] for methodological details). The SHARE data collection has been funded by the European Commission, Directorate-General for Research and Innovation (DG RTD) through FP5 (QLK6-CT-2001-00360), FP6 (SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, and SHARELIFE: CIT4-CT-2006-028812), FP7 (SHARE-PREP: GA N°211909, SHARE-LEAP: GA N°227822, SHARE M4: GA N°261982, and Data Service Infrastructure for the Social Sciences and Humanities (DASISH): GA N°283646), and Horizon 2020 (SHARE-DEV3: GA N°676536, SHARE-COHESION: GA N°870628, SERISS: GA N°654221, SSHOC: GA N°823782, and SHARE-COVID19: GA N°101015924); and by DG Employment, Social Affairs & Inclusion through VS 2015/0195, VS 2016/0135, VS 2018/0285, VS 2019/0332, and VS 2020/0313. Additional funding from the German Ministry of Education and Research, the Max Planck Society for the Advancement of Science, the US National Institute on Aging (U01_AG09740-13S2, P01_AG005842, P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11, OGHA_04-064, HHSN271201300071C, and RAG052527A), and various national funding sources is gratefully acknowledged [46].

PY - 2023

Y1 - 2023

N2 - Background: Machine learning and artificial intelligence have shown promising results in many areas and are driven by the increasing amount of available data. However, these data are often distributed across different institutions and cannot be easily shared owing to strict privacy regulations. Federated learning (FL) allows the training of distributed machine learning models without sharing sensitive data. In addition, the implementation is time-consuming and requires advanced programming skills and complex technical infrastructures. Objective: Various tools and frameworks have been developed to simplify the development of FL algorithms and provide the necessary technical infrastructure. Although there are many high-quality frameworks, most focus only on a single application case or method. To our knowledge, there are no generic frameworks, meaning that the existing solutions are restricted to a particular type of algorithm or application field. Furthermore, most of these frameworks provide an application programming interface that needs programming knowledge. There is no collection of ready-to-use FL algorithms that are extendable and allow users (eg, researchers) without programming knowledge to apply FL. A central FL platform for both FL algorithm developers and users does not exist. This study aimed to address this gap and make FL available to everyone by developing FeatureCloud, an all-in-one platform for FL in biomedicine and beyond. Methods: The FeatureCloud platform consists of 3 main components: a global frontend, a global backend, and a local controller. Our platform uses a Docker to separate the local acting components of the platform from the sensitive data systems. We evaluated our platform using 4 different algorithms on 5 data sets for both accuracy and runtime. Results: FeatureCloud removes the complexity of distributed systems for developers and end users by providing a comprehensive platform for executing multi-institutional FL analyses and implementing FL algorithms. Through its integrated artificial intelligence store, federated algorithms can easily be published and reused by the community. To secure sensitive raw data, FeatureCloud supports privacy-enhancing technologies to secure the shared local models and assures high standards in data privacy to comply with the strict General Data Protection Regulation. Our evaluation shows that applications developed in FeatureCloud can produce highly similar results compared with centralized approaches and scale well for an increasing number of participating sites. Conclusions: FeatureCloud provides a ready-to-use platform that integrates the development and execution of FL algorithms while reducing the complexity to a minimum and removing the hurdles of federated infrastructure. Thus, we believe that it has the potential to greatly increase the accessibility of privacy-preserving and distributed data analyses in biomedicine and beyond.

AB - Background: Machine learning and artificial intelligence have shown promising results in many areas and are driven by the increasing amount of available data. However, these data are often distributed across different institutions and cannot be easily shared owing to strict privacy regulations. Federated learning (FL) allows the training of distributed machine learning models without sharing sensitive data. In addition, the implementation is time-consuming and requires advanced programming skills and complex technical infrastructures. Objective: Various tools and frameworks have been developed to simplify the development of FL algorithms and provide the necessary technical infrastructure. Although there are many high-quality frameworks, most focus only on a single application case or method. To our knowledge, there are no generic frameworks, meaning that the existing solutions are restricted to a particular type of algorithm or application field. Furthermore, most of these frameworks provide an application programming interface that needs programming knowledge. There is no collection of ready-to-use FL algorithms that are extendable and allow users (eg, researchers) without programming knowledge to apply FL. A central FL platform for both FL algorithm developers and users does not exist. This study aimed to address this gap and make FL available to everyone by developing FeatureCloud, an all-in-one platform for FL in biomedicine and beyond. Methods: The FeatureCloud platform consists of 3 main components: a global frontend, a global backend, and a local controller. Our platform uses a Docker to separate the local acting components of the platform from the sensitive data systems. We evaluated our platform using 4 different algorithms on 5 data sets for both accuracy and runtime. Results: FeatureCloud removes the complexity of distributed systems for developers and end users by providing a comprehensive platform for executing multi-institutional FL analyses and implementing FL algorithms. Through its integrated artificial intelligence store, federated algorithms can easily be published and reused by the community. To secure sensitive raw data, FeatureCloud supports privacy-enhancing technologies to secure the shared local models and assures high standards in data privacy to comply with the strict General Data Protection Regulation. Our evaluation shows that applications developed in FeatureCloud can produce highly similar results compared with centralized approaches and scale well for an increasing number of participating sites. Conclusions: FeatureCloud provides a ready-to-use platform that integrates the development and execution of FL algorithms while reducing the complexity to a minimum and removing the hurdles of federated infrastructure. Thus, we believe that it has the potential to greatly increase the accessibility of privacy-preserving and distributed data analyses in biomedicine and beyond.

KW - additive secret sharing

KW - AI store

KW - artificial intelligence

KW - federated learning

KW - interactive platform

KW - privacy-enhancing technologies

KW - privacy-preserving machine learning

U2 - 10.2196/42621

DO - 10.2196/42621

M3 - Journal article

C2 - 37436815

AN - SCOPUS:85164625982

SN - 1438-8871

VL - 25

JO - Journal of Medical Internet Research

JF - Journal of Medical Internet Research

M1 - e42621

ER -

The FeatureCloud Platform for Federated Learning in Biomedicine: Unified Approach

Abstract

Funding

Keywords

Documents & Links

SDU Link

Fingerprint

Cite this