You are here:
COMPUTER SUPPORT SYSTEMS FOR ESTIMATING CHEMICAL TOXICITY: PRESENT CAPABILITIES AND FUTURE TRENDS
Citation:
Richard, A M. COMPUTER SUPPORT SYSTEMS FOR ESTIMATING CHEMICAL TOXICITY: PRESENT CAPABILITIES AND FUTURE TRENDS. Presented at Ninth Int'l Workshop on Quantitative Structure-Activity Relationships in Environmental Sciences: OSAR2000-Crossroads to the XXI Century, Bourgas, Bulgaria, Sept 16-20, 2000.
Description:
Computer Support Systems for Estimating Chemical Toxicity: Present Capabilities and Future Trends
A wide variety of computer-based artificial intelligence (AI) and decision support systems exist currently to aid in the assessment of toxicity for environmental chemicals. These have taken the forms of web-accessible, public toxicity data compilations (NTP, IRIS, RTECS, etc ), commercial toxicity data bases (MDL/Toxicity), chemical property calculators (BIOBYTE, AQUIRE, SPARC, etc.), metabolism predictors and data bases (META, MetabolExpert, MDL/Metabolite), toxicity prediction systems (e.g. TOPKAT, CASE/MULTICASE, ECOSAR, ASTER), and expert rule- based systems ( e.g. DEREK, HazardExpert). At present, available tools fall far short of providing accurate toxicity prediction capability across toxicity endpoints, routes, species and chemical classes, or providing optimal access to relevant information in support of toxicity estimates. This presentation will provide a survey of existing AI and decision support tools currently being applied to toxicity estimation, discuss some limitations of these efforts, and consider current trends and developments likely to have major impact for the future. An important overriding issue is balancing the need for toxicity models resolved to appropriately defined domains of study (e.g.common modes of action in defined biological systems), with the need to think more globally and provide effective linkages between data from different domains of study ( e.g. bioavailability, metabolism, multiple toxicity endpoints). Some problems include widely distributed sources of data and expertise, lack of structure-searchability, and incompatibility of data base formats. Finally, interfacing existing toxicity data and modeling capabilities with more refined and mechanistically informative biological measures, including genomic information pertaining to toxicity, is a rising challenge.
This abstract does not necessarily reflect EPA policy.