Prediction of plasma protein binding of cephalosporins using an artificial neural network.

ABSTRACT

An artificial neural network model is developed to predict the fraction of cephalosporins bound to plasma proteins (f(b)) from their molecular structural parameters. These molecular structural parameters are the molecular weight (MW), the surface area occupied by oxygen and nitrogen atoms (S(O),N), and the surface area occupied by hydrogen atoms attached to oxygen or nitrogen atoms (S(H)). For a training set of 20 cephalosporins and a test set of 3 cephalosporins, root mean squared errors (RMSE) between experimental fb values and calculated/predicted fb values are 0.036 and 0.045, respectively.