Efficient Bayesian Expert Models for Fever in Neutropenia and Fever in Neutropenia with Bacteremia

Abstract

Bayesian expert models are very efficient solutions since they can encapsulate in a mathematical consistent way, certain and uncertain knowledge, as well as preferences strategies and policies. Furthermore, the Bayesian modelling framework is the only one that can inference about causal connections and suggest the structure of a reasonable probabilistic model from historic data. Two novel expert models have been developed for a medical issue concerning diagnosis of fever in neutropenia or fever in neutropenia with bacteremia. Supervised and unsupervised learning was used to construct these two the expert models. The best one of them exhibited 93% precision of prediction. © 2020, Springer Nature Switzerland AG.