“Koroyawa,” a fracture-prevention mat, raised 1.28 billion yen. But can a mat really soften only during a fall?

The provided sources detail two significant breakthroughs in medical AI aimed at enhancing diagnostic capabilities without disrupting existing hospital workflows. First, the Mayo Clinic developed the REDMOD AI, which utilizes human-engineered radiomics to detect pancreatic cancer signs on standard CT scans up to three years earlier than clinical diagnosis. Second, the startup Noetik introduced TARIO-2, a foundation model that predicts roughly 19,000 gene expression patterns directly from inexpensive, routine H&E tissue stains. While the Mayo Clinic's model relies on specific mathematical filters to highlight invisible textures, Noetik's system uses self-supervised learning to uncover complex biological relationships beyond human perception. Together, these technologies suggest a future where AI co-pilots act as a double-check for specialists, significantly improving early detection and personalized treatment insights. By integrating seamlessly with standard imaging and pathology, these tools offer a scalable path toward high-precision medicine.