Roundtable: Accelerating Drug Discovery by Competitive Cooperation
Historically, pharmaceutical companies have kept their machine learning models and data strictly confidential. The EU/EFPIA IMI2 Joint Undertaking funded “MELLODDY” (Machine Learning Ledger Orchestration for Drug Discovery, Grant n° 831472) initiative, a 3 year 18.4M EUR effort, brings together 10 leading pharma companies and 7 tech partners to build a new collaborative ML platform that boosts drug discovery model development without compromising security and privacy (including commercial confidentiality of data and models). This new and unprecedented co-opetitive consortium connects over 10 million small molecules and 1 billion molecular assays into the world’s largest chemoinformatics database. Enabled by federated learning, participating pharma companies can benefit from the joint representation of their combined data while keeping the mapping onto their assays strictly confidential. Enabling Federated Learning for the MELLODDY project, Owkin Connect helped to secure the first Federated Learning model for drug discovery to perform at scale. This federate learning platform has been audited to ensure the highest level of confidentiality and has been deployed in the 10 participating pharmaceutical companies.
3 Key Takeaways:
*In June 2019, 10 major pharmaceutical companies — Amgen, Astellas, AstraZeneca, Bayer, Boehringer Ingelheim, GSK, Institut De Recherches Servier, Janssen, Merck, and Novartis — inked an agreement to build a shared platform called MELLODDY (Machine Learning Ledger Orchestration for Drug Discovery). In partnership with Owkin, NVIDIA, and others, the group sought to leverage techniques like federated learning to collectively train AI on datasets without having to share any proprietary data.
*Although R&D competition remains as fierce as ever, many pharmaceutical companies engage in various types of cooperation and partnership ventures to help drive innovation. The goal of coopetition — when companies cooperate on certain projects while maintaining a competitive stance — is to make drug development more efficient and effective.
*In September 2020, the MELLODDY Project met its year one objective with the successful deployment and running of the world’s first secure platform for multi-task federated learning in drug discovery among 10 Pharma companies.
As Chief Technology Officer of Owkin, Camille currently leads the development of a collaborative machine learning platform designed for data scientists and medical experts. Prior to joining Owkin, Camille was an academic researcher and worked in startups focused on reproducible and traceable AI. She graduated from Mines ParisTech and completed her Ph.D. in applied data science at the Université Pierre et Marie Curie.