Hierarcffical neural networks for censored data

Loading...
Thumbnail Image
Date
2007-06
Authors
Ai Wern, Ooi
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Although neural networks have been applied to survival analysis in recent years, their use in survival analysis has been limited, especially when in the presence of censored data. In this dissertation, a method for handling this problem is suggested. In particular, a hierarchical neural network is proposed. To train the effectiveness of this network. a data set consisting of 86 larynx cancer patients (Klein and Moeschberger, 1997) is used in the hierarchical neural networks and the nonhierarchical neural networks. Some error measurements are utilized to compare the effectiveness among the two models. The results show that the hierarchical model performs better than the nonhierarchical model.
Description
Keywords
Neural networks , Censored data
Citation