Prediction of in Vitro Hepatic Biliary Excretion using Stochastic Agent-Based Modeling and Fuzzy Clustering

We present a method for estimating (predicting) parameter values for an agent-based model of in silico hepatocytes (ISH). The method enables the ISH to interact with simulated drugs to reasonably match results from in vitro hepatocyte excretion studies. Further, we make the estimation method available to the model, itself, to enable it to reasonably anticipate (predict) the biliary transport and excretion properties of a new compound based on the acceptable parameter values for previously encountered compounds. We use Fuzzy c-Means (FCM) classification algorithm to determine the degree of similarity between previously tuned compounds and the new compound. Specifically, a set of simulation parameters for enkephalin was predicted using the tuned parameter values of salicylate, taurocholate, and methotrexate. The feature space for the FCM classification is the physicochemical properties of the compounds