The iDash competition and workshop drew participants from around the world.
The Luddy School of Informatics, Computing, and Engineering recently hosted the iDASH Secure Genome Analysis competition as part of the iDASH Privacy and Security Workshop at Luddy Hall, an event that brought together an international field of teams to evaluate state-of-the-art methods that ensure rigorous data confidentiality during data analyses in a cloud environment.
The iDash Competition, which brings together the biomedical and privacy communities from around the world, was co-founded by Professor of Informatics and Computing Haixu Tang and James H. Rudy Professor of Computer Science XiaoFeng Wang, and colleagues at the University of California at San Diego and the University of Texas Health Science Center in Houston. The 2019 event saw 105 teams compete, including groups from the Luddy School, Microsoft Research, IBM Research, French Alternative Energies, the Atomic Energy Commission, the Massachusetts Institute of Technology, and more. Teams were asked to come up with secure solutions for problems in one of four chosen tracks.
“Bringing together the privacy and biomedical communities is important since privacy protection becomes a main hurdle for wide dissemination and analysis of biomedical data, which is critical for the discovery of more effective treatment of hard-to-cure diseases and enhancement of the well-being of the human society,” Wang said. “In particular, the power of artificial intelligence and its impacts on biomedical science can only be fully unleashed with effective protection of patient data. Providing such protection is a grand challenge we are facing today.”
The workshop and competition began in 2014 and has become a benchmark for the progress of privacy techniques for protecting biomedical data analysis and sharing. Future events will focus on combining different confidential computing technologies, ranging from homomorphic encryption to trusted execution environments, together with blockchain techniques to better understand how the scientific community and the industry can work together to move forward on biomedical data protection and how the advance of privacy technologies support biomedical research and development.
“The results, as reported by academia, industry, and government research agencies, showcase the progress that has been made on confidential, verifiable biomedical data sharing and computing,” Wang said. “Particularly, we found that we are closer to training machine learning models for disease prediction across multiple parities, without disclosing data content to each other, and operating a disease inference model in a trusted execution environment without exposing data even to the server running the model. The findings are really exciting.”
The event is sponsored by the National Institute of Health.
“The iDash Workshop has quickly become the premier forum for biomedical privacy research, and we were thrilled to host the event,” said Raj Acharya, dean of the Luddy School who also spoke at the workshop. “I congratulate the winners of the event, and it’s a testament to the vision of our researchers who have worked with world-wide communities in different disciplines to bring about the kinds of innovations that will shape tomorrow.”
Winners of the competition included:
Track I
Distributed Gene-Drug Interaction Data Sharing based on Blockchain and Smart Contracts
First place: Emory Team
Second place: Sandia National Laboratories
Third place: Yale University
Honorable Mention: BGI Group (China)
Track II
Secure Genotype Imputation using Homomorphic Encryption
First place: CEA (France), KU Leuven (Belgium), Inpher (Switzerland)
Second place: Seoul National University, (South Korea)
Third place: EPFL (Switzerland)
Track III
Privacy-preserving Machine Learning as a Service on SGX
First place: Zhejiang U/City U of Hong Kong
Second place: Indiana University and NCI
Third place (tie): UT Dallas; Ant Finance Service Group; Baidu X-Lab
Track IV
Secure Collaborative Training of Machine Learning Model
First place (tie): Alibaba Gemini Lab; Ant Finance Service Group; University of Washington Tacoma
Honorable mention: MIT