'Accelerating AI Drug Development with Federated Learning' K-Melody Kickoff Meeting Held

A gathering will be held where companies and research institutes leading the 'K-Melody' project, an AI drug development acceleration project based on federated learning, come together to share detailed task contents and directions.

[Photo by Korea Pharmaceutical and Bio-Pharma Manufacturers Association]

The Korea Pharmaceutical and Bio-Pharma Manufacturers Association K-Melody Project Group announced on the 14th that it will hold the 'Federated Learning-based Drug Development Acceleration Project Kick-off Meeting' on the afternoon of the 20th at the 4th-floor auditorium of the Pharmaceutical Hall in Seocho-gu, Seoul.

The K-Melody project is an R&D initiative aiming to develop the 'Federated ADMET Model (FAM),' a federated learning-based prediction model for drug Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET). The Ministry of Health and Welfare and the Ministry of Science and ICT jointly promote the project, with operational support from the Korea Health Industry Development Institute and the National Research Foundation of Korea, while the Pharmaceutical and Bio-Pharma Manufacturers Association and the K-Melody Project Group oversee it.

Recently, through public recruitment and evaluation, the project group selected 26 detailed tasks across three areas: ▲building a federated learning online system (platform), ▲utilization and quality management of drug development data, and ▲AI solution development, along with the lead research institutions for each task.

At the kick-off meeting on the 20th, Kim Hwajong, the head of the K-Melody Project Group, will begin by explaining reference points for project execution, followed by lead researchers presenting introductions of their institutions and research contents for each task.

First, EvidNet, responsible for building the federated learning-based FAM operation platform, will share information on federated learning framework development and advancement, ensuring the safety of drug development data, and plans for developing incentive algorithms.

Additionally, 20 institutions in charge of drug development data utilization and quality management will present their organizations and discuss supplying existing or newly produced data to the platform and participating in federated learning to carry out tasks. These 20 institutions include ▲eight pharmaceutical companies: Daewoong Pharmaceutical, Dongwha Pharm, Samjin Pharmaceutical, Yuhan Corporation, Jeil Pharmaceutical, Hanmi Pharmaceutical, Huons, JW Pharmaceutical; ▲six universities and hospitals: Gachon University, Catholic University, Kyungpook National University, Korea University (Sejong), Seoul National University, Seoul National University Hospital; ▲four research institutes and foundations: Daegu-Gyeongbuk Medical Innovation Foundation, Korea Research Institute of Bioscience and Biotechnology, Institut Pasteur Korea, Korea Research Institute of Chemical Technology; and ▲two companies: Simplex and A-Face.

Five institutions engaged in AI solution development?Gwangju Institute of Science and Technology, Mokam Life Science Research Institute, Izen Science Jeonbuk National University Industry-Academic Cooperation Foundation, and Korea Advanced Institute of Science and Technology (KAIST)?will also introduce their organizations and present plans to develop federated learning-based ADMET prediction solutions.

Following the presentations by each institution, networking and Q&A sessions will take place. Through this kick-off meeting, it is expected that collaboration among the government, pharmaceutical companies, research institutes, and universities will accelerate the creation of an AI-utilized drug development ecosystem.