Elix and LINC Launch World's First Federated Learning AI Drug Discovery Platform, Trained on Data from 16 Pharma Giants

Elix, Inc., an AI drug discovery company based in Tokyo, and the Life Intelligence Consortium (LINC) have announced the commercialization of Elix Discovery™, the world's first AI drug discovery platform incorporating federated learning-based AI models. This groundbreaking platform is the result of a collaborative effort involving 16 pharmaceutical companies and multiple research institutions under the Project Promoting Support for Drug Discovery, led by the Japan Agency for Medical Research and Development (AMED). The core challenge in AI drug discovery is the need for high-quality and diverse datasets, which pharmaceutical companies typically struggle to obtain due to limited access to proprietary and public data. To address this issue, Elix partnered with the Department of Biomedical Data Intelligence at Kyoto University to develop kMoL, a federated learning library. Federated learning allows multiple companies to contribute data for model training while keeping their confidential information secure, thus overcoming the data scarcity problem. Initially launched in fiscal 2020, the AMED project, titled "Development of a Next-generation Drug Discovery AI through Industry-academia Collaboration" (DAIIA), aimed to establish an AI-driven drug discovery infrastructure. Over the course of the project, 17 pharmaceutical companies, including those from LINC, and about 10 IT companies with AI expertise worked together to develop models trained on over 1 million compounds and 10 million data points. These models are now integrated into Elix Discovery™, making the platform available for immediate use by participating companies. The significance of this commercialization is multifaceted. According to Yasushi Okuno, Ph.D., Representative Director of LINC and a professor at Kyoto University, this project stands out because it ensures the practical application of government-funded research beyond the initial funding period. The ongoing data sharing among pharmaceutical companies, facilitated by federated learning, is a notable achievement in an industry often driven by individual corporate interests. Okuno emphasizes that this collaboration is crucial for advancing drug discovery and benefiting patients globally. Teruki Honma, Ph.D., Team Director at RIKEN and an R&D principal investigator for the AMED DAIIA project, highlights the unprecedented scale of the training data used. The models can predict on/off-target effects, ADMET properties (absorption, distribution, metabolism, excretion, and toxicity), and generate novel molecular structures. Honma notes that continuous updates are essential for maintaining the platform's effectiveness and expects Elix Discovery™ to significantly accelerate drug discovery research through ongoing enhancements. Shinya Yuki, Ph.D., Co-Founder and CEO of Elix, emphasizes the importance of addressing data scarcity in AI drug discovery. He points out that kMoL, the federated learning system, has enabled 16 pharmaceutical companies to contribute data without compromising confidentiality, leading to the creation of highly sophisticated AI models. Yuki views this commercialization as a major milestone, setting a new standard in AI drug discovery and potentially making Elix Discovery™ the go-to platform in Japan and beyond. Elix plans to encourage more companies to participate, further enriching the dataset and improving the platform's capabilities. Elix Discovery™ is expected to gain widespread adoption and become the de-facto standard for AI drug discovery in Japan. Initial users are the 16 pharmaceutical companies involved in DAIIA, but the platform is open to additional participants. The combination of federated learning and Elix's expertise in AI engineering and medicinal chemistry positions the company to revolutionize the drug discovery process, reducing costs and timelines while increasing the likelihood of successful outcomes. Industry insiders and academic experts agree that this initiative marks a significant step forward in AI drug discovery. The collaboration between pharmaceutical companies, academia, and technology firms sets a new benchmark for data-driven research and could have far-reaching implications for the pharmaceutical industry. Elix and LINC's efforts showcase Japan's commitment to leveraging AI to enhance medical research and ultimately improve patient care.