Optimal tight glycaemic control supported by differential geometric methods

Optimal control of an Intensive Care Unit (ICU) metabolic model from nonlinear control point of view is presented in the current paper. The transformation of a clinically validated nonlinear model into a series of integrators via exact linearization and asymptotic output tracking is performed. Both methods need the value of the state variables; therefore Kalman-filter extended for nonlinear systems is applied. Finally, linear optimal LQ method is applied on the ICU model handled with differential geometric approach. Results are demonstrated on input data recorded in actual clinical environment. © 2011 Springer-Verlag Berlin Heidelberg.