LLM-Powered Variant Interpretation: 3billion’s AIVARI Wins Best Presentation at KSMGG 2025

3billion announced that its LLM-based variant interpretation technology AIVARI received the Best Oral Presentation Award at the 2025 Fall Conference of the Korean Society of Medical Genetics and Genomics (KSMGG).

AIVARI addresses one of the biggest bottlenecks in genomic diagnostics: the manual collection of literature-based evidence for variant interpretation. Using a state-of-the-art large language model, AIVARI automatically scans newly published medical literature, extracts key evidence required for pathogenicity assessment, and evaluates ACMG guideline criteria with 89.6% precision.

This dramatically reduces the time and expert effort needed for variant interpretation, improving both consistency and diagnostic efficiency. The system also functions as a high-throughput model, identifying relevant evidence even without explicitly provided variant information—an important capability for building large-scale variant databases and advancing automated genomic analysis.

Dr. Dongseok Moon of 3billion’s Clinical Genomics Research Team said,

“AIVARI eliminates the bottleneck of literature curation and helps clinicians reach diagnostic decisions faster and more confidently.”

3billion CEO Changwon Keum added,

“AIVARI strengthens our evidence-generation capabilities across diagnostics, our variant interpretation SaaS GEBRA™, and AI-driven drug discovery, reinforcing our global competitiveness.”