Predicting short- and long-term survival of patients with cancer may improve their care. This study investigates whether natural language processing can predict survival of patients with general cancer from a patient’s initial oncologist consultation document. Using data from 47,625 patients at BC Cancer, machine learning models achieved balanced accuracy of 0.856 (AUC 0.928) for 6-month survival prediction, suggesting the approach may predict survival using readily available data without focusing on a single cancer type.
This study demonstrates that natural language processing applied to initial oncology consultation documents can effectively predict patient survival outcomes across multiple cancer types. The models perform comparably with or better than previous prediction models while using readily available clinical data.