Drug infusion control: An extended direct model reference adaptive control strategy

Medicine is one of the fields where control systems has not made significant inroads. The particular challenges of controlling physiological processes, while not insurmountable, are not trivial. Biological systems are highly nonlinear, and there is a significant amount of variability from one patient to another. The same individual can also exhibit different dynamics depending on other factors, such as environmental stresses.; A controller for drug infusion has to be able to achieve good performance for most, if not all, of the potential patient population. These controllers are designed for the general population, as no knowledge is available before hand for each particular patient. This necessitates perforce an algorithm that will be able to compensate for the differences between individuals and external factors. Of the possibilities, direct model reference adaptive control (DMRAC) is the choice for the results presented herein.; Safety and robustness issues must be considered, as a mistake in the infusion of drugs by a controller can be fatal. To this end, constraints must be enforced on drug infusion rates, and the design procedure must guarantee stability.; In this work the DMRAC algorithm is extended to enhance its applicability to drug infusion control. Methods to handle saturation and rate limits are introduced. Additional knowledge of the process is used both in the design procedure, as well as by the controller on-line. A method to relax setpoint constraints is implemented, which allows for performance objectives that are better aligned with the physician's specifications. The performance and stability for systems with time delay is also improved.; Hemodynamic control is the application of choice. The regulation of mean arterial pressure and cardiac output is required under a variety of clinical conditions. Experimental results from work with canines under various situations are presented.; A second application is the regulation of blood glucose. The particular issues involved are explored in detail. Of interest are the glycemic control of diabetic subjects and the treatment of stress hyperglycemia in critically ill patients. Modeling is covered in detail, including subcutaneous insulin infusion and glucose measurements, the carbohydrate metabolism, and glucose absorption from meals.