Evaluation of Artificial Intelligence Based Models for Chemical Biodegradability Prediction.docVIP

Molecules 2004, 9, 989-1004 molecules ISSN 1420-3049 Review Evaluation of Artificial Intelligence Based Models for Chemical Biodegradability Prediction James R. Baker , Dragan Gamberger , James R. Mihelcic and Aleksandar Sablji? 1,* 2 1 2 1 Department of Civil and Environmental Engineering, Michigan Technological University, 1400 Townsend Drive, Houghton, Michigan, USA Rudjer Boskovic Institute, POB 180, HR-10002 Zagreb, Croatia 2 * Author to whom correspondence should be addressed; email: jrbaker@ Received: 26 October 2004 / Accepted: 12 December 2004 / Published: 31 December 2004 Abstract: This study presents a review of biodegradability modeling efforts including a detailed assessment of two models developed using an artificial intelligence based methodology. Validation results for these models using an independent, quality reviewed database, demonstrate that the models perform well when compared to another commonly used biodegradability model, against the same data. The ability of models induced by an artificial intelligence methodology to accommodate complex interactions in detailed systems, and the demonstrated reliability of the approach evaluated by this study, indicate that the methodology may have application in broadening the scope of biodegradability models. Given adequate data for biodegradability of chemicals under environmental conditions, this may allow for the development of future models that include such things as surface interface impacts on biodegradability for example. Keywords: Biodegradation, QSBR, environmental fate, QSAR, interfaces Introduction Biodegradation is an interfacial phenomenon influenced by a chemical’s tendency to partition to various phases in the environment. Equili