The intent is to create an artificial intelligence software system which accurately predicts chemical toxicity. The system will use a neural network to recognize patterns and correlation between chemical structure/substructures, physicochemical properties and toxicologic/carcinogenic findings. An Organ Toxicity Database containing data on known chemical toxins will be complied form a large subset of NTP TR and NTP TOX documents. The neural network will be trained using subsets of data from the database, and the ability of the neural network to predict the toxicity of previously tested chemicals will be evaluated. At the end of Phase I, the efficacy of the system in determining liver and kidney toxicants and carcinogens will be known. If accurate prediction is achieved, a complete system for predicting chemical toxicity will be built in Phase II and its performance will be statistically analyzed. The availability of such a system would reduce the exposure of the general public to possibly hazardous chemicals, would allow chemical manufacturers to quickly determine the hazards of new and existing products, and would save enormous amounts of time and money currently spent on animal testing.Awardee's statement of the potential commercial applications of the research: A neural network-based prediction system described here would be of value to chemical companies to assess toxicity of new compounds under development. The system could also be used to guide these companies and regulatory agencies in selection of chemicals for toxicity testing and developing plans that are economical in terms of cost and/or time.National Institute of Environmental Health Sciences (NIEHS)
