Battelle and Ohio State University Partner on
New-Tech Bayesian Modeling

A shared belief in the value of collaborative research has energized a new partnership between Battelle’s Measurement and Data Analysis Sciences (MDAS) Product Line and the Spatial Statistics and Environmental Sciences (SSES) Program, Department of Statistics, at The Ohio State University (OSU). Participants have shared recent environmental research through presentations and technical discussions during the fall 2002-winter 2003 academic quarters. Analysis of remote sensing (satellite) data, physical-statistical modeling, and combining monitoring data with model predictions were among the topics. This spring, participants will identify opportunities for collaborative research.

Interest in the power and flexibility of hierarchical Bayesian modeling (HBM), an emerging decisionmaking and information-synthesis technology, is a central theme of the collaboration. While HBM is not new, only in recent years has it emerged as a practical solution to an array of problems. Its power lies in its probabilistic approach of breaking highly complex, multi-dimensional relationships into conceptually simpler and logically intuitive hierarchies. HBM’s flexibility results from its ability to incorporate all available information (data), including but not limited to, qualitative scientific knowledge, disparate data structures, and multiple data sources. Another strength is HBM’s accurate characterization of uncertainty, which in turn supports more informed decision-making. As additional information becomes available, this tool facilitates increasingly improved decision-making over time.

HBM technology is not limited to environmental settings. A recent MSNŽ article (http://msn.com.com/ 2100-1104-984921.html) suggests that HBMs are integral to the data-search and information-retrieval technologies of companies such as Google and Autonomy. HBMs are also used “. . . to determine correlations between specific symptoms and diseases, create personal robots, and develop artificially intelligent devices that ‘think’ by doing what data and experience tell them to do.” Microsoft has already incorporated HBM into several existing applications and plans to embed this technology in future software that lets communication devices automatically filter messages and derive strategies for reaching people. Later this year, Intel intends to provide a tool for constructing HBMs. One existing application involves cameras that warn doctors if patients are at near-term risk of suffering a stroke.

Over the past three years, Battelle’s MDAS Product Line has developed capability and experience in HBM through EPA project support, internally funded research, and the recent hire of Dr. Christopher Holloman, an experienced HBM practitioner. Participating OSU faculty members—Dr. Noel Cressie, Director of the SSES Program, Dr. L. Mark Berliner, and Dr. Tom Santner—are nationally recognized for scientific contributions in this area.

For more information, contact Dr. Steven Bortnick at (614) 424-7487, bortnick@battelle.org.