Predictive Modelling of Stress and Tensile Properties of Oral Dispersible Film Using Artificial Neural Network

Abstract

An Oral Dispersible Film (ODF) remains a polymeric device designed for easier drug administration in the treatment course, especially for paediatric and geriatric therapy patients, for whom adherence remains a challenge owing to swallowing difficulties. The study employs an Artificial Neural Network (ANN) to assess the tensile properties, specifically the stress and Young’s modulus, of PVA electrospun ODFs with κ-carrageenan and ascorbic acid. The ODFs were characterised and their mechanical properties assessed using a tensile mechanical testing apparatus. The ANN model was created using a feed-forward backpropagation neural network. The network was trained with the Levenberg–Marquardt algorithm. The dataset was divided into the training set, validation set, and testing set.  These findings provided indications of the ANN’s predictive power, demonstrated by R2 = 0.999 with very low mean square error. The highest mean square error was 4.737% for stress and 2.081% for Young’s modulus, meaning the model is very accurate. This research shows that modelling based on ANN is time-effective for predicting ODFs' mechanical properties. It optimizes invaluable data elements to formulate process designs related to the prediction and mechanical property relationship to ODFs in the pharmaceutical industry.