General Bayesian Network Approach to Health Informatics Prediction: Emphasis on Performance Comparison

Abstract

Health Informatics is emerging as a promising research area. As average life expectancy increases due to medical technology development, health issues remain most sensitive agenda in most of countries in the world. However, heal th technology requires more intelligent mechanisms by which users’ requirement for more accurate prediction about their health problems can be ful- filled. Furthermore, such intelligent mechanisms must provide very flexible and robust procedures by which complicated but necessary decision support functions are allowed. In this sense, this paper proposes General Bayesian Network (GBN) to predict appropriate diets and restaurants that would benefit users’ health. We compared the performance of GBN with other competing techniques such as NBN (naive Bayesian Network), TAN (Tree Augmented naive Bayesian Network), and decision tree. Expe r- iments with real health dataset revealed that GBN results outperform other techniques with statistical validity.