Date of Award

Fall 1999

Document Type


Degree Name

Master of Science in Biomedical Engineering - (M.S.)


Biomedical Engineering Committee

First Advisor

Alex Y. Bekker

Second Advisor

Stanley S. Reisman

Third Advisor

Arthur B. Ritter


The purpose of this thesis project is the development and testing of a simulated closed loop drug delivery (CLDD) system that consists of a pharmacokinetic, physiological, and feedback-controlling model. The focus of this study is on the control of milrinone inflision to maintain cardiac output at desired setpoint range for patients suffering from congestive heart failure (CHF). The simulated CLDD system is written in VisSim dynamic simulation language for an IBM-compatible PC.

Milrinone pharmacokinetics are represented by a three compartment model. The physiological model consists of the cardiovascular system model linked to the pharmacodynamic submodel of milrinone. The feedback-controlling model consists of a cascade controlling mechanism incorporating a PID controller.

Validation of system dynamics was performed by comparison of simulated results of the loop model (pharmacokinetic and physiological model) to available experimental data. Pharmacokinetic and hemodynamic responses showed that the behavior of the simulated open ioop model was similar to that of CHIF patients under milrinone administration.

The addition of the feedback-controlling model to the open loop model resulted in the development of the CLDD system. Performance of the cascade controller was optimized with tuning of PIP controller. A two-hour control performance was monitored as the CLDD system underwent the following situations: (1) target CO was modified (transient response), (2) perturbation was incorporated as circulatory vessel resistances were changed, and (3) randomization of system parameter was achieved by varying the elimination rate constant. Onset delay, time taken for controller to bring CO within set boundaries, and percentage overshoot of cardiac output from target were the underlining results analyzed in understanding the performance of the controller.

Aside from some minor refinements, the overall performance of the controller showed it to be robust in responding to the changes in the system by adjusting milrinone inftision so that cardiac output could track to the setpoint. The simulated CLDD system as a whole was observed to correctly represent clinical automated drug delivery. The results of the simulated controller also lead into the possibility of developing an automated control milrinone infusion system for maintaining cardiac output for CHF patients.