Falsifying an Enzyme Induction Mechanism within a Validated, Multiscale Liver Model

The focus is an In Silico Liver ISL model family and an evolving suite of mechanistic hypotheses about rat liver-drug interactions. ISLs are multiscale and hierarchical. A medium grain Enzyme Induction mechanism was implemented. Validation falsification of complicated, knowledge-based models requires integrating distinct aspects and methods for multi-aspect validation. For ISLs, such integration has not been straightforward. Falsification is crucial for formulating, testing, and iteratively evolving hypotheses about liver mechanisms. During multi-aspect falsification, the authors can falsify a hypothesis in one aspect while simultaneously validating it in another aspect. The authors demonstrate a multi-scalar validation/falsification event in which they validate the mechanism against coarse grain measures of liver perfusate drug levels and falsify it against a medium grained measure of hepatic zonation. The authors also discuss how falsification is guiding mechanism hypothesis refinement. The ability to scale validation efforts is necessary for effective scientific use models such as ISLs.