According to a recent MIT News article, researchers from MIT and Dana-Farber Cancer Institute have developed a machine learning-based computational model, called OncoNPC, that can analyze the sequence of around 400 genes to predict the origin of tumors in the body, particularly for cases where cancer's origin is unclear, which may guide personalized treatment decisions. The model was tested on over 900 tumors of unknown origin, with high-confidence predictions achieved for about 40% of them, showing potential to increase the number of patients eligible for genomically guided, targeted treatments based on the tumor's origin. The researchers intend to expand the model to include other types of data, like pathology and radiology images, to enhance its predictive ability..
