Estimating mortality risk of cardiac surgery using a fuzzy additive model

The purpose of this study is the introduction of a fuzzy additive risk model to estimate mortality risk following cardiac surgery. The model uses seven variables: 3 binary (e.g. male vs. female) type, and four fuzzified (e.g. age and complexity of surgery) type. Each of the 7 variables is multiplied by a weight to produce a fuzzy output. A second fuzzy function maps this output to specify individual patient risk. This simple model predicts cardiac mortality with greater accuracy than standard logistic regression models. Its performance is similar to a more complex and less intuitive probabilistic neural network model.