International Journal of Engineering and Technology

A robust reagents prediction is able to support the service improvement in laboratories. In this paper, Radial Basis Function Networks (RBFN) method with (3, Q, 1) architecture is used to predict two types of reagents needs, i.e. SD Bioline HBsAg and SD Bioline Anti HCV. Data of reagents from 2012 - 2013 are used as training data, whereas 2014 data are used as comparative data for the prediction result. In RBFN training, the best condition obtained when the spread value is 4 with RMSE 1.63E-06 for both types of reagents. The prediction results with RBFN methods reached 99% with correlation value of 0.99 for each reagents. RBFN method shows better prediction result compared to BPNN method with prediction of 92%.