Release profiles prediction of diclofenac sodium patches using Raman mapping technique

Abstract

Objective: To explore a rapid, nondestructive and accurate method for predicting the in vitro release profiles of transdermal system based on Raman mapping. Methods: Commercial diclofenac sodium patches of different production batches were purchased, Raman spectral data of the fixed surface area of all samples were collected and the distribution of active ingredients was visualized. Then the in vitro release of all patch samples was determined as the standard value using the method in the pharmacopeia. Then the partial least squares model and two convolutional neural network models were established to predict the in vitro release of patch samples at specific time points and the parameters of the equation fitting the in vitro release curve. Finally, the prediction accuracy of all models was evaluated. Results: All models showed excellent prediction accuracy for in vitro release fraction at specific time points. For the prediction of equation parameters, the prediction accuracy of partial least squares model is satisfactory. Conclusion: Based on Raman mapping, the establishment of models such as convolutional neural network is a fast and reliable method for characterizing the in vitro release degree of patches. It avoids many shortcomings of traditional methods and has great research potential and application value.