Correlating Digital and Experimental Chemical Space to Pharmaceutical Manufacturing Processes

Bulk properties are influenced by particle attributes, such as particle size, shape, and chemistry, defined during crystallization and/or milling processes. Understanding how particle attributes affect the pharmaceutical manufacturing process performance remains a significant challenge for the industry, adding cost and time to developing robust production routes. This places strict demands on bulk material flow properties such as blend uniformity, compactability, and lubrication, which need to be satisfied. Consequently, making the flow prediction of pharmaceutical materials during early-stage development is increasingly important. Currently, the suitability of raw materials and/or formulated blends for product development requires detailed, time-consuming experimental characterisation of the bulk properties.
The project aims to demonstrate the applicability of predictive models towards product manufacturing using particle informatics and to improve explainability/model confidence in the results. The project will use the novel coupling of experimental characterisation and computed molecular and particle features to achieve this. This will culminate in a framework that allows for an explainable and interpretable machine learning model.
This project will deliver a novel, interpretable machine-learning model for predicting powder flow considering physical, chemical, and computed molecule/particle features. The output of this model will facilitate the rapid development of new medicines manufacturing.