Microfluidic Synthesis of Indomethacin-Loaded PLGA Microparticles Optimized by Machine Learning.

Abstract

Several attempts have been made to encapsulate indomethacin (IND), to control its sustained release and reduce its side effects. To develop a successful formulation, drug release from a polymeric matrix and subsequent biodegradation need to be achieved. In this study, we focus on combining microfluidic and artificial intelligence (AI) technologies, alongside using biomaterials, to generate drug-loaded polymeric microparticles (MPs). Our strategy is based on using Poly (D,L-lactide-co-glycolide) (PLGA) as a biodegradable polymer for the generation of a controlled drug delivery vehicle, with IND as an example of a poorly soluble drug, a 3D flow focusing microfluidic chip as a simple device synthesis particle, and machine learning using artificial neural networks (ANNs) as an in silico tool to generate and predict size-tunable PLGA MPs. The influence of different polymer concentrations and the flow rates of dispersed and continuous phases on PLGA droplet size prediction in a microfluidic platform were assessed. Subsequently, the developed ANN model was utilized as a quick guide to generate PLGA MPs at a desired size. After conditions optimization, IND-loaded PLGA MPs were produced, and showed larger droplet sizes than blank MPs. Further, the proposed microfluidic system is capable of producing monodisperse particles with a well-controllable shape and size. IND-loaded-PLGA MPs exhibited acceptable drug loading and encapsulation efficiency (7.79 and 62.35%, respectively) and showed sustained release, reaching approximately 80% within 9 days. Hence, combining modern technologies of machine learning and microfluidics with biomaterials can be applied to many pharmaceutical applications, as a quick, low cost, and reproducible strategy.