AIVIS (CEO Daehong Lee), a leading company in AI-based pathology analysis solutions, announced on the 15th that it presented its latest research findings using its pathology AI solution, "Qanti® Breast HER2," at the ESMO Asia Congress 2025 (European Society for Medical Oncology Asia Congress), held at the Suntec Singapore Convention & Exhibition Center from December 5–7. The study demonstrated that AI assistance significantly improves inter-observer agreement and diagnostic accuracy among pathologists in HER2 interpretation for breast cancer. Notably, it highlighted increased clinical reliability in identifying patients in the HER2-low and HER2-ultralow categories. As targeted therapies for HER2 (Human Epidermal Growth Factor Receptor 2) prove effective not only for "HER2-positive" patients but also for those with "HER2-low" and "HER2-ultralow" expression, the importance of immunohistochemistry (IHC) interpretation in treatment decisions is rapidly increasing. However, research has shown significant inter-observer variability among pathologists in these emerging categories, leading to a continuous search for methods to ensure more consistent diagnoses. To address this, AIVIS conducted a large-scale Reader Study in collaboration with Samsung Medical Center (SMC) and the Breast Pathology Study Group under the Korean Society of Pathologists to evaluate the impact of AI-assisted HER2 IHC reading on pathologist consistency and accuracy. The study revealed a 41% improvement in inter-observer agreement among pathologists when using AI assistance compared to manual reading (Fleiss' kappa 0.5181 vs. 0.7232; p < 0.0001). Specific improvements in diagnostic accuracy were also statistically significant: • HER2 1+ (Low): Accuracy increased by 12.7% (0.7586 vs. 0.8853; p < 0.0001) • HER2 Ultralow: Accuracy increased by 12.1% (0.7397 vs. 0.8610; p < 0.0001) These results are evaluated as providing a foundation for more precise patient selection, reducing the risks of under-treatment or over-treatment, particularly for "borderline cases" that were previously difficult to categorize, and ensuring that patients who can benefit from HER2-targeted therapies are correctly identified. An AIVIS representative stated, “By providing consistent analysis results through AI-assisted IHC reading, we expect to enhance diagnostic accuracy among pathologists. This will lead to more precise identification of HER2-low and ultralow patients, ultimately increasing the reliability of the patient selection process for appropriate treatment options.” The representative added, “Moving forward, we aim to establish a global standard for HER2 digital diagnostic references through multi-national, multi-center clinical trials and validation across real-world clinical environments.” This research aligns with the strategic MOU signed between AIVIS and AstraZeneca earlier this year. It is expected to act as a catalyst for activating the precision biomarker diagnostic ecosystem, including the HER2-low and ultralow domains. AIVIS emphasized, “Building on this ESMO Asia 2025 presentation and our partnership with AstraZeneca, we will actively expand collaborative research with global pharmaceutical companies based on AI pathology.”