ABSTRACT

Tissue engineering has shown promise for providing alternatives to traditional surgical methods for reconstruction of damaged or resected tissues, but a number of fundamental issues remain before tissue engineering achieves routine clinical application. The study and design of engineered tissue necessitates the investigation of integrated tissue growth and neovascularization within a biopolymer scaffold. This system illustrates the complexity of natural and engineered systems critical to the survival of living species. The tissue engineering system must be studied in its entirety in order to assess its resiliency and fragility to external environmental changes and internal variations in the decision functions of its elements, to analyze and predict its growth, self-organization and sustainability, and to forecast the emergence of new behavior in its upper levels of hierarchical organization based on the decisions/behavior of its elements. While several individual aspects of tissue engineering have been studied rigorously and detailed models have been developed for these individual components, agent-based modeling provides an integrated framework for studying the interactions among these individual parts that typically invoke a response more complicated than the sum of the individual parts.

The goal of the proposed multidisciplinary research is to integrate experimental and computational studies in an evolutionary active learning framework to optimize engineered tissue growth. Simulations will be run in parallel with experiments to enable adjustments of experimental conditions for improving the growth process based on model predictions of final tissue properties. Three synergistic research activities are integrated. The first is to develop an active learning framework to coordinate the collection and interpretation of experimental and simulation results, refine simulation studies, develop a feedback loop between simulations and experiments, and enable modification of the conditions of ongoing experiments to optimize functional tissue growth. The second is to develop an open-source stochastic multiscale heterogeneous agent-based modeling framework in Java and Repast for modeling tissue growth and develop tissue engineering strategies based on agent-based model predictions. The third is to conduct experimental studies guided by active learning for engineering vascularized bone. The outcome will be a learning-decision-execution environment for integrated computational and experimental research to develop tissue engineering systems studied in their entirety by considering the interdependencies of vascular and tissue growth while tracking variations in individual cells.

The transformational research in this project is the development of an iterative "modeling-simulation-experiment" cycle guided by active learning (AL) to optimize engineered tissue properties by making adjustments both prior to and during tissue growth. One goal of this project is to develop an open-source agent-based modeling (ABM) environment to use agent-based models and experimental data for the development of tissue engineering strategies that can be rapidly screened by simulations to guide experimental work, with a particular emphasis on tissue engineered bone. The second goal is to develop an evolutionary AL framework and batch process supervision and control strategies for conducting realistic simulations and prediction of complex adaptive systems for engineering vascularized tissue. In vitro and in vivo (animal studies) engineered tissue formation will be modeled, with a particular emphasis on tissue engineering vascularized bone. Research results will contribute to the knowledge base in simulation-based engineering science and computational thinking, and in modeling, simulation and control of complex biomedical systems.

This iterative comprehensive "modeling-simulation-analysis-experimental validation" approach will provide ultimately a powerful method for healthcare providers to design better strategies for tissue regeneration and engineering. The proposed activity also contributes to promoting education and training in modeling of complex dynamic stochastic systems, hierarchical agent-based models, and tissue engineering for students at multiple levels, including K-12, undergraduate, and graduate students. The techniques for model development and assessment of the effects of stochastic variations can be used in complex adaptive systems in many fields central to national and global concerns - energy distribution systems, ecosystems, epidemics, and supply chains. Vascularized tissue growth presents an ideal testbed for computer science research in active learning of complex living systems. It also provides valuable material for STEM education at secondary and college levels. Modeling activities appropriate for middle, secondary, and collegiate students will not only introduce students to simulations that can be used to understand and contribute to solutions of complex problems, but will also contribute to the process of developing understanding of the nature of models and abilities to generate models. Further, the agent-based modeling and active learning that are core to this proposal will provide the need to go beyond the learning of equations that govern physical systems to adapting, revising and combining models for the purpose of addressing the functionality, precision and granularity needs for their intended use.

Please report errors in award information by writing to: awardsearch@nsf.gov.