Stefan Klinken

Heinrich Heine Universität Duesseldorf, Faculty of Mathematics and Natural Sciences, Institute of Pharmaceutics and Biopharmaceutics, Universitätsstraße 1, 40225 Düsseldorf, Germany

Purpose

The purpose of this work is to identify material and process-specific influences on compressibility, tabletability, and compactibility. Additionally, the study aims to create a deeper understanding of correlations between tensile strength and compressibility.

Methods
Data from a compaction simulator (STYL’One Evolution) was evaluated using multivariate decomposition and AI-based methods. The constructed database contains data for more than 50 materials as well as for binary, ternary, and quaternary mixtures. In addition to different compression speeds and die geometries, tablets of different target weights were investigated.

Results
The results reveal new material-specific dependencies of tensile strength on the amount of compressed material. It could be successfully worked out that the tensile strength of certain materials was minimally influenced by the tablet mass, whereas for other materials a clear influence of the weight could be demonstrated. Additionally, novel methods in the utilization of AI for compression analysis are presented. Here, it can be demonstrated that neural networks with recurrent architecture are suitable for predicting properties of tablets across various materials and process settings.

Conclusions

The presented work demonstrates the complexity of the relationships within larger databases of compression data. However, with the assistance of AI tools, the evaluation of the data is facilitated, enabling the development of new models beneficial for pharmaceutical academia and industry.