Adaptive advanced modelling approaches in the evolution of vector/cell manufacturing processes and analytics for patient-tailored cell-gene therapy pr

Cell gene therapy is a technique whereby an exogenous gene expression unit is introduced, using a vector, into the somatic cells of an organism in order to prevent, halt or reverse a pathological condition. Recent developments in vector safety and efficacy have brought about significant renewed interest in gene therapy, a number of patients with life threatening diseases have been successfully treated with gene therapy products, and GSK were granted approval for the first commercial gene therapy product in May 2016.

The development of gene therapy products is hindered by process complexity and variability; vector and cell manufacturing processes consist of a large number of manual tasks which introduce significant variability into the production methods and thereby product quality. It is likely that methods of data fusion with process-related information from vector and cell expansion and processing will bring significant benefits in process understanding, and increase the rate of development of new treatments. This project will make use of advanced data analytics and modelling techniques to identify critical relationships between patient data and vector/cell manufacturing processes, as well as their re-evaluation using data that becomes available over the patient's lifetime.

The suitability of advanced modelling approaches, such as multivariate data analysis (MVDA), stochastic, and hybrid modelling techniques, which proved to be invaluable in increasing the effectiveness of biologics manufacturing, will be explored. The novel semi-quantitative methods developed by the group of Prof Glassey as well as Bayesian approaches have proven to be effective in the context of data sparse environment and also will be investigated and further developed in this studentship. The possibility of combining first principles knowledge of the processes with MVDA process representation in semi-parametric hybrid models will in addition be explored in order to provide an effective tool for rapid process development and optimisation.