Novel ensemble feature selection techniques applied to high-grade gastroenteropancreatic neuroendocrine neoplasms for the prediction of survival
Jenul, Anna; Stokmo, Henning Langen; Schrunner, Stefan; Hjortland, Geir Olav; Rootwelt-Revheim, Mona-Elisabeth; Tomic, Oliver
Peer reviewed, Journal article
Published version
Date
2023Metadata
Show full item recordCollections
Original version
Computer Methods and Programs in Biomedicine. 2023, 244 . 10.1016/j.cmpb.2023.107934Abstract
• We model and predict the overall survival in patients with high-grade gastroenteropancreatic neuroendocrine neoplasms. • Novel ensemble feature selectors RENT and UBayFS are assessed for predictive performance, stability, and interpretability. • Data- as well as expert-driven feature selection (with and without prior knowledge) are evaluated. • Our results show that both methods allow accurate predictions and have a stabilizing effect on the model parameters. • RENT and UBayFS identify influencing factors for overall survival and support clinical experts in decision-making. Novel ensemble feature selection techniques applied to high-grade gastroenteropancreatic neuroendocrine neoplasms for the prediction of survival