[Edaily Reporter KIM SAE-MI ] Lunit Inc.(328130)The company held its first hackathon utilizing a healthcare-specific artificial intelligence (AI) foundation model currently under development as part of a national project and set out to identify potential application services.

Lunit Inc. announced on the 26th that it held the “Medical-Specific Foundation Model Hackathon,” sponsored by the Ministry of Science and ICT and the National IT Industry Promotion Agency (NIPA), at its headquarters in Gangnam-gu, Seoul, on the 21st and 22nd.

This event was organized to validate the potential applications of “L2-preview”—developed by Lunit Inc. through the “AI-Specialized Foundation Model” project, a national strategic initiative of the Ministry of Science and ICT—and to identify use cases in the medical field. A total of about 70 participants, comprising 15 teams made up of AI developers, undergraduate and graduate students, clinicians, and engineers, took part.

Participating teams were assigned tasks such as developing a medical question-and-answer system and a health management chatbot using L2-preview, as well as conducting benchmark tests. In the benchmarking task, all participating teams answered the same medical and health management-related questions, and their performance was compared by scoring the quality of their responses.

The health management chatbot task focused on the practical applicability of Q&A services for the general public. Clinicians and foundation model developers served as judges, evaluating benchmark performance, medical accuracy, and the potential for real-world service implementation.

Throughout the event, benchmark results were published in a leaderboard to enable participating teams to continuously refine their submissions. OpenAI provided ChatGPT Pro subscriptions and API usage credits to support the competition’s operation and award ceremony.

The Benchmark Award went to the AIM team, which achieved high performance on major medical science benchmarks.

The PROTIA Award, which evaluated applicability in clinical settings, was awarded to the “Nalyeophan Wigobi” team. This team developed a service using a foundation model trained on medical-specific data and a knowledge base search system.

Lunit Inc. plans to connect the outstanding outcomes from this hackathon to real-world validation in clinical settings. By leveraging its network of partners, including hospitals, the company will provide opportunities for clinical validation and technical support, and plans to link these efforts to pilot projects in actual hospital environments and follow-up research and development (R&D).

Yoo Dong-geun, Lunit Inc.’s Chief AI Officer (CAIO), stated, “In the medical AI ecosystem, it’s not enough to simply release models; we need a space where developers and clinicians can come together to directly validate and co-create solutions.” He added, “We will provide support to ensure that the ideas and use cases identified at this hackathon can be implemented in real-world medical settings.”