3billion to Unveil Four Research Abstracts at ASHG 2025, Highlighting AI’s Role in Rare Disease Diagnosis

BOSTON, US – 3billion (CEO Geum Chang-won), a company specializing in AI-based rare disease diagnosis, announced on the 16th that it will present four research abstracts related to its AI genetic variant interpretation technology at the ‘2025 American Society of Human Genetics (ASHG 2025)’ Annual Meeting. The event will take place in Boston, USA, from October 14 to 18 (local time).

ASHG is the world’s most prestigious genetics conference, attended by over 8,000 genetics and genomics experts globally. At this year’s meeting, 3billion focused on introducing the clinical utility and innovative research achievements of its AI-powered diagnostic technology.

3billion is presenting a total of four abstracts, including:

1. ‘3ASC 2.0,’ an AI-integrated variant interpretation model.
2. ‘AIVARI (AI VARiant Interpreter),’ an automated ACMG classification system utilizing Large Language Models (LLM).
3. ‘GEBRA,’ an AI genome analysis platform.
4. Clinical research results on whole-genome sequencing-based rare disease diagnosis, jointly conducted with Seoul National University Hospital.

Among these, the research on ‘3ASC 2.0’ was selected as one of the top 60 out of over 3,000 submitted posters, earning the prestigious opportunity for an oral presentation in the ‘Lightning Talk’ session. The Lightning Talk is a newly established “top-tier poster presentation” format reserved for studies demonstrating high scientific innovation and clinical contribution, marking ‘3ASC 2.0’ as a major research highlight of the conference.

‘3ASC 2.0’ is a Transformer-based deep learning model incorporating the latest technology to simultaneously evaluate various variant types by integrating Whole-Genome Sequencing (WGS) and Whole-Exome Sequencing (WES) data from rare disease patients. Previously, different variants—such as Single Nucleotide Variants (SNV), Insertions and Deletions (INDEL), Copy Number Variants (CNV), and Short Tandem Repeats (STR)—had to be interpreted separately. Now, 3ASC 2.0 is engineered to perform this integrated analysis within a single AI system.

To further enhance diagnostic accuracy and address the issue of imbalanced variant data, 3billion developed and applied its own ‘MIL-GFE (Multiple Instance Learning–Global Feature Extractor)’ structure, which learns both the patient’s clinical and genetic information. Furthermore, Large Language Model (LLM) technology was integrated into the process, enabling the system to self-learn the association between patient symptoms and diseases for more refined interpretation.

Actual research results show that 3ASC 2.0 demonstrates overwhelming performance compared to competing models. It detected the true causative variant within the top 10 candidate variants with a 98% probability (Top-10 recall), and the overall model accuracy (AUROC) reached 99.6%. Moreover, reviewing just the top 10% of variants flagged as important by the AI identified 99% of the causative variants, significantly enhancing the analysis efficiency for clinical geneticists.

3billion has participated in ASHG for four consecutive years since 2022, continuously building up its unique technological prowess and research achievements in AI-based rare disease diagnosis. Innovative AI models like 3ASC 2.0, which was presented as an oral talk, are acting as a key competitive advantage by drastically shortening the diagnosis turnaround time, thus expanding its utilization in global clinical trials and collaborative research.

“Presenting research results at a globally renowned conference like ASHG for the fourth consecutive year and receiving recognition with an oral presentation underscores the core competencies of 3billion,” said Dr. Kyoungyeul Lee, Chief Scientific Officer of 3billion. “Having proven that our AI-based genome interpretation and rare disease diagnosis capabilities have reached a world-class level, we will continue to drive technological innovation to contribute to the diagnosis and treatment of rare disease patients globally.”