A neural network model for survival data

• DOI: 10.1002/sim.4780140108
• Publication Date: 2007-04-29T09:18:19Z
• Authors: David Faraggi; Richard Simon
• Keywords: Akaike information criterion

Abstract Neural networks have received considerable attention recently, mostly by non‐statisticians. They are considered many to be very promising tools for classification and prediction. In this paper we present an approach modelling censored survival data using the input—output relationship associated with a simple feed‐forward neural network as basis non‐linear proportional hazards model. This can extended other models used data. The parameters estimated method of maximum likelihood. These likelihood based compared, readily available techniques such ratio test Akaike criterion. illustrated on men prostatic carcinoma. A interpreting predictions factorial contrasts is presented.