Using Bayesian networks to improve the decision-making process in public health systems

This paper proposes the use of Bayesian networks to support the decision-making process in public health systems. In particular, this paper presents LARIISA_Bay, a new component based on Bayesian networks that works together with LARIISA, a context-aware platform to support applications in public health systems. The main goal of the proposed component is to assist teams of health specialists in order to better diagnose diseases through data collected from users of LARIISA. As a case study, we focus on scenarios of dengue fever disease. We classify dengue cases into one of the following levels: normal, grave or emergency. Based on this classification, teams of health specialists can accurately make decisions, for example, to alert a health care agent to visit locations with a high incidence of the disease, to send a team of health specialists when a dengue emergency case has occurred, as well as give technical instructions on how to deal with specific cases. We present a prototype of LARIISA_Bay as well as the corresponding interfaces to support the interactions with the component. We compare the obtained results with real diagnosis of general practitioners. The results presented show the efficiency of the proposed approach.