Analysis of Competing Risks Data Using Neural Network Models

Modeling and analysis of lifetime data using neural network models is a topic of recent interest in survival studies. In the present study, we use multilayer perceptron neural network models for the analysis of competing risks data. Multilayer perceptron neural network with a modification is used for the estimation of survivor function in presence of covariates. The estimates are compared to the smoothed estimates proposed by Wells[1], by extending the procedure into the competing risks set up. Neural network models are also used for classification of failure types in competing risks data. We illustrate the methods using real data sets.