Prediction of Probability of Survival in Critically Ill Patients Optimizing the Area under the ROC Curve

: This article presents the method of Support Vectors Machines (SVM) for predicting probability of survival in critically ill patients by using Platt’s method to fit a sigmoid1 . Moreover, the comparison of two interesting exploitations is provided: the maximization of the AUC and the minimization of the error rate in binary classification tasks.

• Type: Conference
• Year: 2007
• Where: IJCAI
• Authors: Oscar Luaces, José Ramón Quevedo, Francisco Taboada, Guillermo M. Albaiceta, Antonio Bahamonde
• Keywords: Artificial Intelligence, Binary Classification Tasks, IJCAI 2007, Platt’s Method, Support Vectors Machines