用片段常数法估算有机化合物在鱼体内的生物富集因子

根据337个有机化合物在鱼体内生物富集因子的实测数据建立了片段常数估算模型。涉及的有机物包括卤代苯、卤代联苯、氯代脂肪烃、多环芳烃、醇、醚、醛、酮、腈、苯酚、苯胺、硝基苯、呋喃、二误Yin、含氮氧硫的杂环混合和物、脲和有机磷农药等多种类别，logKow在0.39-8.60之间。模型可决系数0.982，标准差0.453，平均绝对误差0.322个对数单位。60种检验化合物的平均预测残差为0.320个对数单位。

Predicting of BCFs in fish for organic chemicals using fragment constant method

Based on 337 experimentally determined BCF values in fish for nonpolar and polar chemicals,a fragment constant estimation model for log BCF was established.In the first step of regression process,the whole body of the database was divided into two parts,one containg 277 chemicals was used for model development and the other of 60 chemicals for model validation.As a result,an average error of 0.320 log unit was produced when prediction of log BCF was made to the test set chemicals by applying the regression model derived from the train set data.The two sets of data then were combined to generate the final prediction model,whose determination coefficient was 0.982,with a standard error of 0.453 and a mean absolute residual of 0.322 log unit.