LG AI Research Unveils Next-Generation Cancer Diagnosis AI Model 'EXAONE Path 2.0'

LG AI Research is leading the era of precision medicine by releasing its next-generation cancer diagnosis AI model, EXAONE Path 2.0, as open source for research use. This model applies technology that integratively learns pathology images from the patch level up to the whole-slide level.

This technology solves the 'Feature Collapse' problem that undermines the accuracy of AI analysis, raising genetic mutation prediction accuracy to a world-leading level. Through this, it precisely captures relationships between tissues and provides high reliability and speed in actual clinical settings.

Key achievements are as follows:

• Reduced testing time: Shortened genetic testing time, which previously took more than 2 weeks, to within just seconds.
• Multimodal learning: Trained on large-scale data of over 10,000 images matching RNA genetic information with pathology images.
• Realizing precision medicine: Enables gene expression prediction and targeted therapy recommendations using only tissue images, without the need for expensive RNA testing.

Through this release, LG AI Research plans to spread the public value of AI technology and continue expanding innovation beyond cancer into diagnosis, prognosis prediction, and treatment response analysis for various diseases.