USING NEUROFUZZY NETWORKS TO MIMIC ANESTHESIOLOGIST KNOWLEDGE IN DECISION MAKING ON PROPOFOL ADMINISTRATION

Abstract

The purpose of anesthesia is to maintain a steady state for specific clinical operations. In general, one anesthesiologist utilizes anesthetic drugs and anesthetic skills to make sure the depth of anesthesia (DOA) carefully in proper level such that a patient will not perceive pain during surgical procedure. It is complex to be treated as an art to reduce all sensations, whether it is the sense of pain, touch, temperature, or position. In this paper, utilizing the self-learning and the human-like reasoning ability of neurofuzzy networks, we design the virtual anesthesiologist to accommodate the knowledge and the experience of the real anesthesiologist in anesthetic drug administration. The heart rate and bispectral index are used as the input variables and the bispectral index target value (BIStarget) heart is treated as output variable. The anesthesia simulator is adopted to verify the virtual anesthesiologist's ability and to explore the patient status of the simulator. The pilot experiments and extended experiments have been carried out. The result showed that the virtual anesthesiologist was able to support the decision making on the maintenance of the patient DOA at BIStarget 60.