I wrote an article for Applied Health Economics and Health Policy a while back on using decision trees in healthcare. The models are not as sophisticated as the models we were using internally at the time however it gives a short demonstration of the power of decision trees. I thoroughly believe that even these simple models are easier to understand and put into practice than typical statistical models. Take for example the following tree for pneumonia:
It is easy to see that patients who have many diagnoses (specifically greater than 9), had a cardiovascular operation, and have several chronic conditions are more likely to be readmitted. Furthermore, this can be easily operationalized with as little as a checklist.