Automated paper – based pathogenicity assessment… Demonstrating 89.6% accuracy, eliminating bottlenecks in clinical variant interpretation.
3billion (CEO Changwon Geum), an AI-based rare genetic disease diagnosis company, has announced the results of a study that innovatively increased the efficiency of genome interpretation using a large language model (LLM).
3billion announced on the 24th that the research titled 'AIVARI (AI VARiant Interpreter): LLM-based system for automatic extraction of literature-based evidence' presented by Dr. Dong-Seok Moon of the Clinical Genome Research Team at the '2025 Korean Society of Medical Genetics Fall Conference' held on the 20th was recognized for its technical value and received an award for excellent presentation.
AIVARI is a solution that automates the process of "literature-based evidence collection," a major bottleneck in rare disease diagnosis, using generative AI. Previously, clinical geneticists and medical professionals had to manually sift through vast, unstructured literature and analyze it in accordance with the American College of Clinical Genetics (ACMG) guidelines, the global standard for genetic variant interpretation. However, AIVARI automates this highly specialized, time-consuming process, significantly improving diagnostic efficiency.
The key lies in precise contextual analysis utilizing the latest LLM. The AI automatically extracts the key evidence necessary for pathogenicity assessment from over 100,000 new medical research papers published annually, and determines whether the five criteria of the ACMG guidelines are met with 89.6% accuracy. This dramatically reduces the time and human error inherent in traditional manual analysis, ensuring both accuracy and consistency in interpretation.
This study is particularly significant because it demonstrates the potential for scalability as a high-throughput model, capable of automatically identifying valid data within the literature without requiring separate mutation input. Based on this, 3billion plans to build a large-scale mutation database and further enhance the technological sophistication of its genetic diagnostic solutions.
Moon Dong-seok, a researcher at the 3billion Clinical Genome Research Institute, explained, “AIVARI is a technology that uses AI to perform the literature curation section, which was the biggest bottleneck in variant interpretation,” and “It will substantially contribute to medical staff making faster and more accurate diagnostic decisions.”
Geum Chang-won, CEO of 3billion, said, “Medical papers are a key resource not only for diagnosing rare diseases but also for discovering new drug targets and elucidating mechanisms.” He added, “Based on AIVARI, 3billion will continue to expand its global technological competitiveness by strengthening its capabilities to secure evidence data in all areas of AI new drug development with its diagnosis and genetic mutation interpretation SaaS ‘GEBRA.’”
You must be logged in to post a comment.