Room for Optimization
Richard Bourne’s research group, located at the University of Leeds’ Institute of Process Research and Development, focuses on the optimization of pharmaceutical processes using automated systems. Recently, he has been working on developing automated flow systems that combine online analysis, feedback control, and evolutionary algorithms that help with process understanding and optimization.
What does your work focus on?
Continuous flow processes are more cost-effective than traditional production methods – and the pharmaceutical industry is becoming increasingly interested in their use. I’ve spent several years developing an automation platform that can help companies to more rapidly develop their processes. Pharmaceutical processes are fascinatingly complex, with many stages involved in the synthesis of therapeutics. I often work with innovator companies who generate new treatments, but face the pressures of limited development times due to patent lifetimes. Many companies also struggle with the challenges of clinical trials. Automated technologies can automate and optimize some processes, helping companies to more effectively develop their therapies.
How did you get started working with Arcinova?
The collaboration followed a meeting with the Arcinova team at a Dial-A-Molecule Symposium in 2018. The Dial-A-Molecule network (https://bit.ly/2SrrmkF) was established by the UK’s Engineering and Physical Sciences Research Council to “promote research aimed at a step change in our ability to deliver molecules quick and efficiently”. It’s an amazing initiative that has helped develop new links between academia and industry. Following this event, Gareth and the Arcinova team visited our facilities. Since then, we’ve been working together to develop new projects that integrate industry 4.0 developments within their portfolio, such as implementing machine learning algorithms that help to optimize reactions. We are also working on developing our automation platforms to include other types of reactor including miniaturised systems such as continuous stirred tank reactors to broaden the scope of chemistry that can be studied with our Industry 4.0 approach.
At what stage is the work? What are the main challenges right now?
The optimization platforms we’ve developed work very well for homogeneous chemistries with fixed chemical molecules – we can typically optimize new processes within a week, including a day of experimental time on the platform. However, it’s a real challenge to develop these automated platforms so that they can tackle heterogenous (solid/liquid) processes, but we hope to develop this in the future. We also need new algorithms and robotic systems to be able to change discrete parameters (such as solvents or catalysts) so we can find an optimal process with the optimized conditions.
What are your thoughts on the future of automation and Industry 4.0 in the pharma industry?
I think there are real opportunities in using new technology and Industry 4.0 approaches to transform how pharmaceutical processes are developed, but there are concerns that it may lead to a loss of synthetic chemists within the pharmaceutical industry. I believe that this is simply not the case. There will always remain challenging chemistries that cannot be performed by robots and we will also still require chemists to interpret the data generated and propose new experiments. The future will likely be chemists that are augmented with digital capability so that they can focus on the challenging work and automated platforms can perform the routine/laborious actions.
What other research projects connected to pharma are you involved with?
We collaborate with a number of companies, including AstraZeneca, GlaxoSmithKline, and Dr. Reddy’s Laboratories. Examples include:
- Developing self-learning reactor systems for the automated development of kinetic models, with AstraZeneca. We’re using mixed integer linear programming techniques, capable of kinetic model discrimination, to create an autonomous system that can evaluate and develop scalable process models.
- The Cognitive Chemical Manufacturing (CCM) EPSRC project. This project involves the University of Leeds, AstraZeneca, IBM, Swagelok, University College London, the University of Nottingham and Promethean Particles. We’re developing an Industry 4.0 approach to revolutionize the transfer from laboratory to production using data-rich and cognitive computing technologies.
- Developing automated self-optimizing reactors for multistage processes, with Dr Reddy’s. We are looking to expand the applicability of self-optimizing systems and explore multi-step optimizations.
You can read more about these projects on our lab website (www.bournelab.co.uk)
I’d like to highlight my current Senior Research Fellowship with the Royal Academy of Engineering, in collaboration with AstraZeneca, which is enabling us to explore methods to quickly generate optimized libraries of compounds for clinical trials.
Further Reading
AD Clayton et al., “Automated self-optimisation of multi-step reaction and separation processes using machine learning,” Chemical Engineering Journal, 384 (2020).
AM Schweidtmann et al., “Machine learning meets continuous flow chemistry: Automated optimization towards the Pareto front of multiple objectives,” Chemical Engineering Journal, 352, 277-282 (2018).
N Holmes at al., “Self-optimisation of the final stage in the synthesis of EGFR kinase inhibitor AZD9291 using an automated flow reactor,” Reaction Chemistry & Engineering, 1, 366-371 (2016).
