Using novel machine learning tools developed at Stanford Medicine, researchers have mapped three distinct cellular configurations that correspond to clinical outcomes for patients with a rare, difficult-to-treat cancer called soft tissue sarcoma.