We are actively pursuing a "challenge-response" approach to study the mechanisms that underlie complex biological systems. In this approach, a controlled challenge is applied to the system, and its behavior is observed before, during, and after the challenge. The expectation is that well-designed challenges will cause the system to behave in a way exposes the underlying mechanism. This project is concerned with the designing the challenges that elicit the most informative behaviors. As an example of this approach, our group has quantified an association between cardiac electrophysiology and cardiac metabolism using a series of measurements on isolated rabbit hearts. We observed the time course of cardiac action potential duration (APD) as a result of intermittent cardiac anoxia, a metabolic challenge that simulates asphyxia. Evidence from these experiments support our hypothesis that supplementation with glutamine and glutamate mitigates the reduction in APD associated with cardiac anoxia. Our next challenge is to develop a framework for mechanism discovery, using mathematical models of alternative mechanisms, and to design the challenge-response experiments that best distinguish among the alternatives. Statistical methods for estimation, inference, and research design in the context of mechanistic models are a patchwork from the literature on mathematical modeling, nonlinear regression, longitudinal analysis, and model selection. There is a need to unify and tailor the methodology in this context, and to seriously consider the data complexities that commonly arise in challenge-response experiments, for example, discontinuous and nonlinear trends, heteroscedasticity in time, and nested sources of residual error. The optimal design methods that perform best under these complexities are unknown. The proposed work addresses this issue, by considering innovative approaches to optimal design in simulated in-silico studies, and using experiments with isolated rabbit hearts. Given two mechanistic models of challenge-response behavior, the experimental challenge may be tailored to discriminate between the two mechanisms. Challenge-response experiments are unique because the design parameters (e.g., the type, duration, and intensity of challenge) modify the response in a predictable way. Hence, these "active" design parameters are optimized such that the predicted responses are most divergent among the competing mechanistic models. We anticipate that optimal research design using both "active" and "passive" design parameters (e.g., the number and spacing of measurements in time) will accelerate biomedical discovery, and reduce the associated costs. We propose to study and experimentally validate an optimal design methodology in the context of challenge- experiments for the purpose of mechanistic model discrimination. We will simultaneously address three important questions in the modeling of cardiac electrophysiology.

Public Health Relevance

We anticipate that the biostatistical and experimental design strategies produced by this work will accelerate biomedical discovery in the study of complex biological systems, and reduce the associated costs (e.g. financial, human, laboratory animal). In addition, this research will address specific scientific questions that arise in our effort to understand cardiac electrophysiology and related therapeutics, such as cardiac defibrillation.

National Institute of Health (NIH)
National Heart, Lung, and Blood Institute (NHLBI)
Research Project (R01)
Project #
Application #
Study Section
Biostatistical Methods and Research Design Study Section (BMRD)
Program Officer
Lee, Albert
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Vanderbilt University Medical Center
Biostatistics & Other Math Sci
Schools of Medicine
United States
Zip Code