Delivery of insulin to type 1 diabetics, control of anaesthesia and chemotherapy for cancer patients are typical examples of drug delivery systems. The main objective of a drug delivery system is to provide effective therapy while minimizing the side-effects. This can be facilitated by developing mathematical models to capture the pharmacokinetic and pharamacodynamic effects of the drugs on the state of the patient where state is a vector which, for example, can be given by blood glucose concentration and the mean arterial pressure. Due to the complex nature of these models, it is often non-trivial to design and implement controllers that can compute the optimal drug infusion rates on-line, for a given current state of the patient. We have developed advanced model based controllers that can take into account the model of the patient and constraints on the state of the patient and the drug infusion rates. These controllers are based upon the theory of multiparametric programming developed by our research group. This theory allows an optimal division of the multidimensional space of the state of the patient into a set of regions and each region is characterised by a unique drug infusion law which is an explicit function of the state in the corresponding region. These explicit functions can be stored on a simple and portable computational platform. The computing of the optimal drug infusion rates therefore reduces to simple function evaluations. These developments are expected to simplify controller implementation and result in tighter control of drug infusion rates and better lifestyle for patients.
+44 (0)20 7679 5063 - Copyright © 1999-2005 UCL