The Application Of Decision Tree In Studying The Risk Factors Of Depression In Somatic Disease Patients

Objective: Apply CART model and CHAID model, two different types of decision tree, and logistic regression model to discuss the risk factors of depression in somatic disease patients. In addition, to compare the differences and similarities between these three models and explore the useful information for logistic regression model from decision tree.Methods: With the dependent variable of patients' HAD score, establish logistic regression model and decision tree model, and compare the accuracy of different models by comparing the areas under the ROC curve. Then, with the information from the tree models, establish a new logistic model and discuss the improvement and differences between the two logistic models.Results: The potential risk factors of depression in somatic disease patients include digestive disease, poor health condition, deterioration of the state of illness, bad marriage, female, large family, undesirable life events, the interaction of poor health condition and large family, and the interaction of poor health condition and undesirable life events, etc. On the other hand, areas under the ROC curve for logistic...