Optimization of the Prescription of Cassiae Semen Umbilical Patch Based on Box-Behnken Response Surface Method Combined with BP Neural Network

OBJECTIVE To optimize the prescription of Cassiae Semen umbilical patch.

METHODS Taking appearance, the sum of peak areas of aurantio-obtusin and chrysophanol, and the total peak area of active ingredients as evaluation indexes. A single factor experiments, Box-Behnken response surface methodology combined with BP neural network were used to optimize the PEG6000∶PEG4000, drug loading, and menthol dosage in umbilical patch formulations, and prepare Cassiae Semen umbilical patch.

RESULTS The test results of Box-Behnken response surface method and the prediction results of BP neural network were tested each other, and the average comprehensive evaluation index values of the validation results were 0.976 and 1.040, respectively. Compared with the Box-Behnken response surface validation test, the comprehensive evaluation index of BP neural network was higher and the different was statistically significant(P<0.01). The prediction result of BP neural network were selected as the optimal prescription for umbilical patch, that was, PEG6000∶PEG4000 ratio of 3∶2 (g∶g), drug loading of 14.8%, and menthol dosage of 4.2%.

CONCLUSION Cassiae Semen umbilical patch prepared with this prescription has a smooth and flat appearance, no bubbles, and good skin permeability.