Artificial intelligence-aided rational design and prediction model for progesterone-loaded self-microemulsifying drug delivery system formulations

Abstract

Artificial intelligence (AI) is now applied across various domains in nanomedicine. Self-microemulsifying drug delivery systems (SMEDDS) are isotropic mixtures of active compounds that can produce spontaneous oil-in-water emulsions. SMEDDS can improve the solubility of lipophilic drugs such as progesterone (PG). However, the physicochemical properties of SMEDDS are sensitive to various factors, depending on their components. This study generated a prediction model algorithm for PG-loaded SMEDDS to provide appropriate droplet size (DS), polydispersity index (PDI), zeta potential (ZP), and % drug loading (%DL). Various machine learning algorithms were compared for their accuracy, as reported by root mean square error (RMSE) and coefficient of determination (R2). The selected machine learning algorithms were implemented with an unseen training dataset, and the model performance was re-evaluated. The correlation of each factor was investigated. Self-micro emulsifying (SME) time, cloud point, pH, and viscosity of predicted PG-loaded SMEDDS were evaluated. Results showed that linear regression algorithms gave the highest accuracy and optimal prediction performance with the highest RMSE and R2. All components of PG-loaded SMEDDS correlated with DS, PDI, ZP, and %DL. The physical properties of predicted PG-loaded SMEDDS showed SME time within 39 s, cloud point at around 71.3 °C, pH between 5.53 and 6.10, and viscosity between 10.32 and 14.23 cP. This research outlined the application of a machine learning algorithm to build a prediction model to optimize PG-loaded SMEDDS drug delivery formulations.