The past decade witnessed significant progress in the development of our knowledge and understanding regarding relationships between the sub-structure of chemicals and their toxic end-point activity, as determined by in vitro assays (e.g., Salmonella mutagenicity) and long-term carcinogenesis assays. A key element of this progress has been the laborious development of toxicology data bases that were large enough to provide information across non-congeneric chemical classes. Concomitantly, computer-based, """"""""expert"""""""" systems, that utilize artificial intelligence software, have been developed to analyze such toxicology data bases. This application of computer technology to the analysis of various toxicology data bases shows great potential for transforming toxicology from a science based on retrospective, empirical-based observation to one based on modelling, hypothesis testing, and validation of theory. In the short range, this new approach to structure-activity relationship (SAR) research promises to provide insights into toxicology that will help to eliminate the need for testing in some cases, reduce the number of animals used in testing, and optimize the effectiveness of the information obtained from testing that is conducted. A new project was initiated to extend the application of knowledge-based, """"""""expert"""""""" computer systems to the SAR analysis of organ toxicity data generated from NTP short- and long-term, in vivo toxicology studies. Over the last two quarters, data fields and decision rules for an organ toxicity data base have been set up and relevant data from NTP Technical Reports have been entered into it for recently-reported chemicals. By the end of the year, preliminary SAR analyses will be run on organ toxicity data for about 30 chemicals. This will be done in collaboration with Dr. H. Rosenkranz at the Univ. of Pittsburgh, using a particularly promising knowledge-based computer system called CASE (computer assisted substructure evaluation) that has been developed by Drs. Klopfman and Rosenkranz.
