Predicting the Risk of Complications in Coronary Artery Bypass Operations using Neural Networks
Richard P. Lippmann, Linda Kukolich, David M. Shahian · Neural Information Processing Systems · 1994
Experiments demonstrated that sigmoid multilayer perceptron (MLP) networks provide slightly better risk prediction than conventional logistic regression when used to predict the risk of death, stroke, and renal failure on 1257 patients who underwent coronary artery bypass operations at the Lahey Clinic. networks with no hidden layer and networks with one hidden layer were trained using stochastic gradient descent with early stopping. networks and logistic regression used the same input features and were evaluated using bootstrap sampling with 50 replications. ROC areas for predicting mortality using preoperative input features were 70.5% for logistic regression and 76.0% for networks. Regularization provided by early stopping was an important component of improved performance. A simplified approach to generating intervals for risk predictions using an auxiliary confidence MLP was developed. The is trained to reproduce intervals that were generated during training using the outputs of 50 networks trained with different bootstrap samples.