Neural network and regression predictions of 5-year survival after colon carcinoma treatment.

摘要: BACKGROUND. The Commission on Cancer data from the National Data Base (NCDB) for patients with colon carcinoma was used to develop several artificial neural network and regression-based models. These models were designed predict likelihood of 5-year survival after primary treatment carcinoma. METHODS. Two modeling methods in study. Artificial networks select more important variables NCDB database model survival. A standard parametric logistic regression also two compared a prospective set not development. RESULTS. yielded receiver operating characteristic (ROC) area 87.6%. At sensitivity mortality 95% specificity 41%. ROC 82% at gave 27%. CONCLUSIONS. found strong pattern predictive status. produced somewhat less accurate, but good, results.