Behavior Risk Prediction for Psychiatric Patients Using Data Mining Techniques

Abstract

In recent years, data mining technology has gradually been used in various applications of medical systems. How to extract useful information from various complicated factors within medical records and use classification and prediction technology in data mining will become an important application of machine learning in medicine. The extracted knowledge and factors ranked by significance will provide physicians as the reference for making diagnostic decision before the costly and invasive therapy. Moreover, the establishment of disease and risk prediction model for patients is also a critical topic on disease and public health management. This study attempts to build the prediction model of risk behavior for psychiatric patients from the clinical data of cooperative hospitals, and the results show that support vector machine learning technique behaves superior to neural network and logistic regression in average with small sample situations.