Academia Sinica Logo

Pancreatic cancer remains the "king of cancers" due to its late-stage diagnosis and a dismal five-year survival rate of only 13%. To address this challenge, a cross-disciplinary team from Genomics Research Center and Institute of Chemistry, Academia Sinica, and National Taiwan University Hospital integrated two decades of clinical expertise with advanced metabolic profiling to develop PanMETAI, an AI-driven predictive model.

Operating on a highly standardized Nuclear Magnetic Resonance (NMR) platform, PanMETAI systematically extracts key molecular features from over 260,000 data points per individual. This technology captures the holistic metabolic signature of pancreatic cancer from pre-cancerous stages to early lesions, overcoming the limitations of traditional single biomarkers.
Validation through independent blind tests across cohorts from Taiwan and Lithuania demonstrated exceptional performance, with predictive accuracy (AUC) reaching 93–99%. This framework not only offers a powerful screening tool for high-risk populations but also provides a scalable "multi-cancer early prediction platform" for the future of precision medicine.

The study, authored by Dr. Dan-Ni Wu and corresponding authors Prof. Yu-Ting Chang, Prof. Chao-Ping Hsu, and Prof. Chun-Mei Hu, was published in Nature Communications on February 13, 2026.