Inovacije u formulaciji i procesu: QbD pristup i PAT alati podržani tehnikama veštačke inteligencije

Sažetak

QbD (Quality by Design) and PAT (Process Analytical Technologies) concepts significantly facilitate the implementation of new technologies in the pharmaceuticals´ formulation and processes development. From simple formulations to complex delivery systems, QbD approach allows identification of the critical process parameters and material properties affecting the pharmaceutical products quality. For the analysis of complex relationships, establishment of the design space and, most importantly, control strategies, modeling and simulation tools are of paramount importance. Hybrid models, which combine elements of mechanistic modeling and empirical approach, are particularly important for processing of large amount of data collected by monitoring the process with PAT tools. This enables the establishment of a virtual copy (digital twin), or cyber-physical system, which facilitates the optimization and continuous improvement of the process. Artificial intelligence techniques in formulation and process innovations involve different machine learning algorithms. They are used to solve regression or classification problems and to process data of various types (numerical, textual, images, etc). Artificial neural networks can be applied from the initial formulation development to the production of validation batches for which the bioequivalence predicted by models has been confirmed. Artificial intelligence technology is also very important for the design and application of virtual copies of continuous production processes or complex biotechnological processes. This facilitates the implementation of the Real Time Release Testing (RTRT) strategy. It is to be expected that good modeling practices will be more precisely defined through the official regulatory guidelines, in the context of the application of artificial intelligence techniques.