ucsf

Probabilistic framework improves uncertainty estimation in oncology diagnostics

Researchers have developed a new diagnostic framework that enables neural networks to quantify uncertainty, reducing the risk of overconfident errors in cancer subtyping.