Improved GA combined with GDBP algorithm for forecasting releasing behaviors of drug carrier

The bioadhesive drug delivery systems using starch-based colon-targeted drug carriers have drawn great attention in the field of pharmaceutical science in resent years. A Neural Network (NN) prediction model was developed based on hybrid method of improved genetic algorithms (GA) and conjugate gradient algorithm for backpropagation (GDBP) NN according to key factors that affect releasing behaviors of starch-based colon-targeted drug carrier. In particular, function approximation capability and high efficiency of GDBP NN is used to simulate nonlinear relation between key factors and drug carrier releasing behaviors. Furthermore, the simulation results indicate that compared with traditional GA-BP NN, training efficiency of GA-GDBP NN has been greatly improved. Consequently, the model finds a new way to predict drug carrier releasing behaviors and instructs factors setting in real experiments.