Chemical and Material Toxicity Evaluation Module

The U.S. Air Force deploys and encounters a highly diverse spectrum of chemicals and materials. The health hazards of these materials must beevaluated to protect personnel and to minimize life cycle costs for weaponsystems. To evaluate health risks, chemical toxicity information is needed.However such information is costly and requires several years of testing toobtain. As a result, reliable data are available for only a few percent ofthe chemicals in use today. The research under this Phase I SBIR project isdirected at demonstrating the feasibility of using artificial intelligenceapproaches to estimate human health risks from exposure to chemical hazards. To accomplish this feasibility demonstration, a chemical toxicity database will be designed and developed to underpin predictions of genotoxicity, carcinogenicity, anddevelopmental/reproductive toxicity for chemicals of interest to the Air Force. Both a neural network and statistical techniques will be used to predict these three toxicity endpoints, and the performance of the two approaches compared. Technicalfeasibility will be demonstrated by achieving a 90 percent or better accuracy in predicting toxicity for chemicals of interest to the Air Force. This level of performance representsa significant improvement over currently available systems.If proven feasible during Phase I, YAHSGS' approach to predicting chemical toxicity will provide a rapid and inexpensive means to determine the health risks of new chemicals; the risks of existingchemicals for which inadequate toxicity data exist for conducting health hazard assessments (~80% of chemicals in commercial use); and risks for chemicals encountered in military and industrial situations. The Chemical and Material Toxicity EvaluationModule (CAMTEM) will be sold either as an application for installation on a user's system or as a tool to be used via Online Analytical Processing (OLAP). The OLAP approach will greatly expand the user group and market potential by reducing the user'shardware, software, and, potentially, training requirements. Two specialized modules will be developed to expand the CAMTEM's usage and marketability to government and industrial users. The first will predict environmental effects of subject chemicalsincluding the risks associated with chemical daughter products formed or released to the environment. The second will provide reverse engineering capabilities that will identify ways for chemical designers to eliminate or reduce health risks fromchemicals or products being developed. The overall benefit from the CAMTEM approach will be the availability of significantly better health risk information for the 80% or more of the chemicals now in commercial use for which there are currentlyinadequate toxicity risk related information and for new chemicals being developed. That information will enhance governments' and industries' ability to recognize specific chemical toxicity related risks and take responsible steps to avoid exposingpeople and the environment to harmful chemicals.