Programs and Communities

Hybrid system models - those devised from two or more disparate sub-system models - provide a number of benefits in terms of conceptualization, development, and assessment of dynamical systems. The decomposition approach helps to formulate complex interactions that are otherwise difficult or impractical to express. However, hybrid model development and usage can introduce complexity that emerges from the composition itself.

To improve assurance of model correctness, sub-systems using disparate modeling formalisms must be integrated above and beyond just the data and control level; their composition must have model specification and simulation execution aspects as well. Poly-formalism composition is one approach to composing models in this manner.

This dissertation describes a poly-formalism composition between a Discrete EVent System specification (DEVS) model and a Cellular Automata (CA) model types. These model specifications have been chosen for their broad applicability in important and emerging domains. An agent-environment domain exemplifies the composition approach. The inherent spatial relations within a CA make it well-suited for environmental representations. Similarly, the component-based nature of agents fits well within the hierarchical component structure of DEVS.

This composition employs the use of a third model, called an interaction model, that includes methods for integrating the two model types at a formalism level, at a systems architecture level, and at a model execution level. A prototype framework using DEVS for the agent model and GRASS for the environment has been developed and is described. Furthermore, this dissertation explains how the concepts of this composition approach are being applied to a real-world research project.

This dissertation expands the tool set modelers in computer science and other disciplines have in order to build hybrid system models, and provides an interaction model for an on-going research project. The concepts and models presented in this dissertation demonstrate the feasibility of composition between discrete-event agents and discrete-time cellular automata. Furthermore, it provides concepts and models that may be applied directly, or used by a modeler to devise compositions for other research efforts.

The tables and graphs in this chartbook were created using data collected by Camden Public Schools for the school year 2008-2009. Rutgers Center for State Health Policy obtained de-identified data from the schools and computed a BMI score and a BMI percentile (BMIPCT) for each child. Weight status is defined using the following BMIPCT categories.

BMIPCT

BMIPCT < 85

BMIPCT ~ 85

BMIPCT ~ 95

BMIPCT ~ 97

Weight Status

Not Overweight or Obese

Overweight and Obese

Obese

Very Obese

BMIPCT categories are presented at the city level and in sub-group analysis by age, gender, and race. Aggregate data are also presented at the school level, with notation, where representativeness of the data was a concern.

Tables and graphs on pages 5, 7, 9, and 11 show comparisons with national estimates (National Health and Nutrition Examination Survey, 2007-2008). The national data are representative of all 2-19 year old children in the US.

Each graph and table is accompanied by brief summary statements. Readers are encouraged to review the actual data presented in tables and graphs as there is much more detail.

The New Jersey Childhood Obesity Study, funded by the Robert Wood Johnson Foundation, aims to provide vital information for planning, implementing and evaluating interventions aimed at preventing childhood obesity in five New Jersey municipalities: Camden, Newark, New Brunswick, Trenton, and Vineland.

These five communities are being supported by RWJF's New Jersey Partnership for Healthy Kids program to plan and implement policy and environmental change strategies to prevent childhood obesity.

Effective interventions for addressing childhood obesity require community-specific information on who is most at risk and on contributing factors that can be addressed through tailored interventions that meet the needs of the community.

Using a comprehensive research study, the Center for State Health Policy at Rutgers University is working collaboratively with the State Program Office for New Jersey Partnership for Healthy I<ids and the five communities to address these information needs. The main components of the study include:

• A household survey of 1700 families with 3 -18 year old children

• De-identified heights and weights data from public school districts

• Assessment of the food and physical activity environments using objective data

Data books and maps based on the results of the study are being shared with the community coalitions in the five communities to help them plan their interventions.