In the breeding of Bellamya,weight data of some individuals are often missing.To make best use of information on all individuals with excellent breeding performance,an artificial neural network was trained on four morphological traits (including shell height,shell width,aperture height and aperture width) and body weight data of 784 individuals from five geographical populations including Yangcheng Lake,Jiangyin,Guanlian Lake,Hong Lake and Xiantao.After this,data of 261 individuals from Tai Lake were used to test the artificial neural network model.In the end,an artificial neural network model for predicting missing body weights of Bellamya was successfully established.In addition,the artificial neural network model was used to predict the missing body weights of 201 Bellamya from Weishan Lake,and the determination coefficient of this method was compared with those of two other prediction methods (i.e.,the predicted mean matching method and the random forest prediction method).The results showed that the determination coefficient of the artificial neural network model constructed in this study was 0.96 for predicting the missing body weight,which was obviously higher than those of the predictive mean matching method (0.87) and the random forest prediction method (0.85).This study could provide an efficient method for the prediction of missing values of body weight involved in the breeding process of the Bellamya.