Virtual Human Initiative

Validation of many simulation models involves experimental testing on a physical model. For biological systems, much of this testing is done at ORNL's Mouse House.

The Virtual Human will integrate data, biophysical (and other) models, and advanced computational algorithms into a research environment used to investigate human responses to stimuli. This effort will go far beyond the visualization of anatomy produced by the Visible Human Project to incorporate physics (such as mechanical and electrical properties of tissue) and biology (from physiological to biochemical information) into a platform so that responses to varied stimuli (biological, chemical, physical, and--it is hoped--psychological) can be predicted and results viewed.

Neural Control of Breathing and Heartbeat

Breathing and heartbeat are continually adjusted to match an individual's needs. Brainstem neural circuits called central pattern generators (CPGs) establish respiratory and heartbeat patterns. Mechanoreceptors, bloodstream chemoreceptors, and other neural circuits, provide feedback to regulate these patterns. ORNL researchers are developing an integrated model of the respiratory and heartbeat circuitry that controls oxygen and carbon dioxide blood gas levels. The eventual goal is the description of both normal and abnormal breathing patterns.

The Virtual Human Respiratory System Model, a current ORNL Laboratory-directed R&D project, is representative of model integration challenges in the larger Virtual Human Initiative. It integrates a specific anatomical organ model (the lung) along with physiological models (respiratory and cardiovascular), in a heterogeneous computational grid environment.

Conditions modeled will include normal lung, asthmatic lung, and pneumothorax (punctured lung). The lung sound profiles generated will compare to actual human lung sounds recorded in diagnostic examinations.

The grid environment may consist of parallel supercomputers, workstations, high performance networks, high performance storage systems, and high end interactive visualization systems These resources can be integrated using software elements such as MPI, Globus, and Netsolve which support the multiple languages (e.g. Java, C, C++, Fortran), typically used in the development of human component models.

Collaborators: University of Tennessee, University of Kentucky, Boston University, Vanderbilt University, Walter Reed Medical Center

How will we get data for the Virtual Human? There is a world of data not currently available from humans using noninvasive measurement methods. We take the mouse for the prototype since its genetic characteristics and gene funtions have been and are still being so throughly studied. Where mice are not the best human model, we will have to pick the species that provides the best surrogate for human physiology or disease processes. Areas of study include:

• normal animals
• specific induced diseases in animals
• specific genetic types

Thus the "Virtual Mouse" will be

• a science tool
• a gateway to the human model for experimental data
• a means of determining what information is necessary rather than what information is easy to obtain
• a test bed for advanced data acquisition and analysis tools, such as injectable multifunction sensors and non-linear data analysis and fusion