Integration of qualitative and quantitative reasoning to support medical decisions in cardiac intensive care units

The main aspects of a new expert system for analysing the clinical status and evolution of postoperative cardiac patients in Intensive Care Units (ICUs) are described. Long-term knowledge is structured through a causal network which represents the main relationships between hemodynamic and metabolic quantities. the inference engine adopts a hybrid able to manage both quantitative and qualitative reasoning. The aim of the reasoning process is to identify all the possible configurations of the network, i.e. all the possible pathophysiological conditions coherent with the patient´s monitored quantities. The results of a simulation concerning a real patient at “high risk” are illustrated as an example of system diagnostic capabilities. The system imputes the severity of the patient to a decrease in cardiac performance, with a consequent increase in oxygen extraction from blood