An information theoretic method for classifying patients according to the risk of adverse hospital incidents.

The measure of Mutual Information is used to select, out of a large number of possibly interrelated qualitative patient variables, those that are most related to the probability of occurrence of adverse incidents. The subpopulations defined by the variables selected are grouped according to risk levels while minimizing the loss of information. Results of the application of the method to predict the occurrence of three types of incidents in four hospitals are reported.