Steps Toward Controlling Blood Pressure during Surgery using Evolutionary Programming

The problem of controlling patient blood pressure during surgery is considered. As a first step, evolutionary programming experiments are conducted using simplified models of patient response to sodium nitroprusside in the absence of stochastic perturbations. The models are evolved on-line as data is sampled during closed-loop control. The results of two sets of experiments indicate that such evolutionary control methods may yield better performance than other techniques which have been previously tried, such as CMAC. Extensions of the current effort are discussed.