System Identification Of Dynamical Models For Signals Related To The Human Use Of Ethanol

Abstract

The influence of genetics on the risk for alcoholism is a major theme in alcoholism research. Genetic research depends on phenotyping. However, accurate phenotyping of human use of alcohol is difficult. What are essentially video games with alcohol as a reward are being used to examine human use of alcohol in controlled circumstances. A generative model (containing parameters with unknown values) of a simple game involving a progressive work paradigm is described along with the associated pointprocess signal processing that allows system identification of the model. The system is demonstrated on human subject data. The same human subject playing the game under different circumstances, e.g., with and without a psychoactive drug, is assigned different parameter values. Potential meanings of the different parameter values are described. Physiologically based pharmacokinetic models have been used to describe the distribution and elimination of ethanol after intravenous administration. Mathematically, these models are nonlinear ordinary differential equations. These equations are solved and optimized, by using their gradient, to formulate and refine parameter identification and control strategies. The Hessian information is then used to design an optimal input to the system.