Applications of experts' judgement to derive structure-biodegradation relationships.

The experts' judgement data on microbial degradation were used to develop the first general QSAR biodegradability model (Boethling and Sabljic, 1989) which is composed of a set of structural descriptors and a set of quantitative rules. Its evaluation and validation with experimental biodegradation data clearly show that the developed model gives a realistic and reliable account of structurebiodegradability relationship for organic chemicals. The same set of experts judgement data was used to develop structure-biodegradation rule by the application of an inductive machine learning method. An improved structure-biodegradation rule was derived from a larger training set of 160 chemicals, i.e. the combined experts' judgement and evaluated experimental biodegradation data. This rule has good predictive ability and discloses logical dependencies between structural features that have a strong influence on biodegradation of organic chemicals. Thus, the understanding of biodegradation processes will benefit from developed rule.