Technical Summary

Our research is aimed to provide input data for the digitalisation of lyophilisation processes in biopharma manufacture. Lyophilisation is the process by which the pharmaceutical and biopharmaceutical industries produce powder-like formulations, starting from liquid formulations. Solid formulations are known to be more stable than liquid ones; the shelf life is longer and the required temperatures for storage are usually higher. The lyophilisation process itself is a complex one which involves a fine-tuned interplay of temperature and pressure in order to achieve the ideal conditions for optimised removal of water from vials. The choice of set temperature and pressure throughout various parts of the lyophilisation cycle depend on several factors, such as vial type, vial size, fill, excipients type, concentration of active ingredients, vial disposition, among others. Given this is a batch-type of process (as opposed to continuous manufacturing), any mistake or error in a batch will result in a large loss of material and time, since most cycles last for about 3 days. Currently, optimization of a lyophilisation process is achieved by trial and error, which is wasteful and time consuming, this project aims to work towards the digitalisation of the lyophilisation process using AI tool for creating a digital twin – the intention is that a digital version of the lyophiliser at NIBRT will be constructed in a way that all parameters for a given formulation can be optimised to yield the most suitable lyophilisation cycle. The prediction of failures and required maintenance procedures will also be facilitated, thus providing value to this field of study in an applied manner. The data we’ll generate at NIBRT will be used to build a digital twin of the physical instrument.