Identification and control of intrinsic bias in a multiscale computational model of drug addiction

Personalized medicine is rapidly evolving with the objective of providing a patient with medications based on the "use of genetic susceptibility or pharmacogenetic testing to tailor an individual's preventive care or drug therapy" [1]. It is reasonable to foresee that this domain will incorporate sources of biological knowledge other than genetics including computational modeling of diseases. For this purpose, a critical issue is how to identify and control systematic biases that may arise. In this paper, a multiscale computational model of drug addiction is presented and the interpretations of the simulated behavioral profiles of a virtual subject are discussed. These outcomes are analyzed using mathematical analytical techniques with particular attention directed to minimization of systematic biases. The simulations exemplify how a structural analysis of the model, prior to the actual simulations, may benefit the overall framework in terms of accuracy. While this paper focuses on an equation-based model for drug addiction, a similar methodology could be applied to other types of computational models for other diseases.