9522639 Kaufman Critical care patients have often suffered a disturbance to the normal operation of their cardiovascular system; this disturbance could have been generated by surgery or some sort of trauma. A responsibility of the critical care physician is to maintain certain patient states within an acceptable operating range. Often, the anesthesiologist will infuse several drugs into the patient to control these states close to the desired values, historically, this has been done using a manual (open-loop) strategy. Towards the important goal of using feedback for aiding the anesthesiologist, the Departments of Biomedical (BME), Chemical (CHE), and Electrical, Computer, and Systems (ECSE) Engineering at Rensselaer Polytechnic Institute are proposing an interdisciplinary research project in the development and analysis of a human-aided supervisory adaptive control system and its application to a representative drug delivery system. This system will regulate the infusion of multiple drugs in order to simultaneously steer several hemodynamic states (eg blood pressure, cardiac output) towards desired values. This system can also be considered as an interactive human aided supervisory system in which the human operator or anesthesiologist commands the desired state and interacts with the control drug selection logic. From a set of measured signals, the patient state is constructed and then used by the anesthesiologist to command a desired state and with the help of a supervisory program, to select the controls or drugs. These drugs, along with the commanded values, are then used by the control system to steer the states towards their commanded values. Fuzzy and adaptive control concepts are to be used and an expert system will be used to select the proper (adaptive or fuzzy) controllers for the drug infusion rates. The novel contributions expected to result from the proposed research include: A system that manipulates the infusion of a variety of drugs, in response to some trauma, in a manner so as to steer the hemodynamic states towards their desired values. Procedures for studying the behavior of complex nonlinear systems that include a mixture of fuzzy controllers and conventional and/or adaptive controllers. Procedures for deciding what controls or drugs should be used and which should be nulled out for a specific state and mission. Control strategies that incorporate almost continuous measurements of cardiac output. ***
