Fully Automated Platforms for Drug Nanocrystals Manufacturing via Continuous-Flow, Data-Driven Antisolvent Crystallization

Lead Research Organisation: University College London
Department Name: Chemical Engineering

Abstract

The pharmaceutical industry is undergoing a period of unprecedented change in terms of product development, with increased digitization, greater emphasis on continuous manufacture and the rapid advent of novel therapeutic paradigms, such as personalized medicines, becoming more and more business critical. This change is amplified by Quality by Design considerations and the now routine use of the Target Product Profile approach to the design of patient-centred dosage forms. The recent advances in the range of available therapeutic strategies, alongside the breadth of diseases that can now be successfully treated, has resulted in the need for both new dosage forms and manufacturing approaches. Crucially, there has been a shift from high volume, low cost manufacture towards a more specialized, higher value product development. Consequently, ever more sophisticated approaches, not merely to producing medicinal products, but also to controlling their quality at every stage of the manufacturing process, have become paramount. These would be greatly facilitated by the emerging technologies, based on artificial intelligence and machine learning techniques, for enhancing online process analysis as well as real-time responsive process control. These technologies are particularly important for products where the financial and practical margins for manufacturing error are low, as is the case for an increasing proportion of new therapies.

In this proposal, we focus on a new way of screening, manufacturing and quality controlling drugs in the form of nanocrystals, that is, drugs prepared as nanosized crystalline particles stabilized by surface-active agents. In particular, we will combine continuous-flow processing, online advanced process analytical technology, real-time process control and quality assurance, design of experiments, advanced data analysis and artificial intelligence to deliver fully automated, self-optimizing platforms for screening and manufacturing drugs as nanocrystals via antisolvent precipitation. These dosage forms have attracted substantial interest as a means of delivering poorly water-soluble (and thus poorly bioavailable) drugs, a persistent and increasing problem for the pharmaceutical industry.

While nanocrystals offer a suitable test system for our approach, our methodology and the manufacturing platform we intend to deliver can be applied to other drug delivery systems. We focus on nanocrystals because they are of considerable therapeutic and commercial significance both nationally and internationally.

We intend to use continuous-flow small-scale (i.e. millifluidic) systems. These offer excellent process controllability, can generate crystals of nearly uniform size, and as the process is continuous, the product characteristics are more stable than in batch systems. Millifluidic systems are flexible (one platform can produce a larger variety of products) and agile - reacting rapidly to changes in market demands; they reduce the manufacturing time, speed up the supply chain and, being smaller, can be portable. These systems also expedite screening, curtailing the quantities of material required, benefits that design of experiments will amplify. This data-driven technique allows identifying the most informative experiments, maximizing learning while minimizing time and costs, advantages not fully exploited by the pharmaceutical industry. These technologies, coupled with online advanced process analytical methods, real-time process control, cutting-edge data analysis and machine learning methods, have the potential to disrupt the status quo, accelerate process development and deliver transformative platforms for the cost-effective and sustainable manufacturing of active pharmaceutical ingredients in solid dosage form, reducing the timeline from drug discovery to patient, and contributing to placing the UK at the forefront of innovation in the pharmaceutical sector.

Publications

10 25 50
publication icon
Besenhard M (2023) Non-fouling flow reactors for nanomaterial synthesis in Reaction Chemistry & Engineering

 
Description The fourth industrial revolution is gaining momentum in the pharmaceutical industry. However, particulate (in particular, crystallization) processes and suspension handling are still challenging for automation and real-time particle size analysis. Furthermore, the development of crystallization processes is limited by expensive and time-intensive experimental screenings. To overcome these bottlenecks, we have developed a fully automated modular crystallization platform. The system combines sample preparation, automated crystallization, and immediate crystal size analysis via online laser diffraction, and provides a technology for rapidly screening the design space of crystallization process with reduced experimental effort (through data-driven machine learning classification and identification methods). In the laser diffraction measurements, to avoid multiple scattering, crystal suspension samples are diluted automatically. Multiple software tools, i.e. LabVIEW, Python and PharmaMV, and logic algorithms are integrated in the platform to facilitate automated control of all the sensors and equipment, enabling fully automated operation. A customized graphical user interface (GUI) is provided to operate the crystallization platform automatically and to visualize the measured crystal size and crystal size distribution of the suspension. Antisolvent crystallization of ibuprofen, with ethanol as solvent and water with additives as antisolvent, has been used as case study. The platform has been demonstrated for the crystallization of small ibuprofen crystals in a confined impinging jet crystallizer, performing automated pre-planned (open-loop) user-defined experiments with online laser diffraction analysis.
Exploitation Route The following technologies, which this project is advancing, are timely and of academic and industrial interest: continuous-flow crystallization, online particle characterization (e.g., laser diffraction, dynamic light scattering, Raman spectroscopy), full system automation (both open and closed loop modes), design of experiments, data-driven machine learning classification and identification methods, self-optimization. We are developing and combining these technologies for the continuous synthesis of API crystals, but the outcomes of this project are transferable to many other industrial processes.
Sectors Chemicals

Healthcare

Manufacturing

including Industrial Biotechology

Pharmaceuticals and Medical Biotechnology

 
Title Automated Screening of experimental conditions for fouling free operation 
Description An automated method was developed to screen the fouling-free feasible parametric range of the crystallization process using design of experiments, process automation, and pressure sensors. 
Type Of Material Improvements to research infrastructure 
Year Produced 2024 
Provided To Others? No  
Impact Significant reduction in time and labor required to screen experimental conditions for crystallization in a range of different reactors 
 
Title Automated laser diffraction-based online crystal size analysis method 
Description An online and automated analysis technique was developed and equipped with an optical flow cell for laser diffraction analysis of crystals. The automated platform can automatically collect samples, dilute them to the desired extent, and perform laser diffraction analysis, displaying crystal size distributions in real time. 
Type Of Material Improvements to research infrastructure 
Year Produced 2023 
Provided To Others? No  
Impact Automated laser diffraction-based online crystal size analysis method is able to rapidly screen the process parameters and identify the conditions suitable for nanonization of API crystals. The platform has been showcased to obtain 500 nm to 5 micron-sized ibuprofen nanocrystals. 
 
Title An algorithm for process automation in continuous flow systems 
Description Combination of flow chemistry, microreaction technology and computational methods for system identification has resulted in the development of automated platforms for rapid development and optimization of processes. One of the major challeneges involved in the development of such platforms is to develop algorithmic frameworks that allow seamless and secure data flow between various componenets of the platform and that establish self-triggered event flow for process automation. One of the objectives of our project is to develop automated platforms for drug nanocrystals synthesis in continuous flow systems. To achieve this, we have developed an algorithmic framework that employs Python programming language, LabVIEW software and open platform communication data access (OPC DA) technology. 
Type Of Material Computer model/algorithm 
Year Produced 2023 
Provided To Others? No  
Impact The methods to develop process automation in continuous flow systems are not commonly used both in academia and industry. One of the reasons for this is the lack of knowledge about the resources needed and the implementation of various methods involved. One of the expected outcome of our research is to make the methods for process automation in continuous flow systems, more accessible to users in academia and industry by developing an algorithmic framework applicable for process automation in continuous flow systems. 
 
Title LabVIEW virtual instruments for hardware control 
Description LabVIEW virtual instruments were developed for different equipment such as pumps, valves, magnetic stirrers, level sensors, and flow sensors enabling external control using a PC. 
Type Of Material Computer model/algorithm 
Year Produced 2022 
Provided To Others? No  
Impact LabVIEW virtual instruments was used for automated operation of a crystallization and online particle size measurement platform 
 
Title MLAPI - A machine learning-driven computational framework to assist automated drug crystals synthesis in flow 
Description We have developed a machine learning (ML)-driven computational framework to facilitate the development of nanocrystal drug formulations through automated continuous flow crystallizers. The framework incorporates classification algorithms, both discriminative and generative, for feasibility studies. It also features multi-task Gaussian Process (GP) models for surrogate modelling, active learning for optimal sequential experimental design, and Bayesian optimization for process optimization. All these techniques are implemented using advanced Python libraries, such as PyTorch and BoTorch. The framework has proven successful in the case study of Ibuprofen synthesis and is poised for application in another case study, focusing on the synthesis of the drug Ketoprofen. 
Type Of Material Computer model/algorithm 
Year Produced 2023 
Provided To Others? No  
Impact The development of the machine learning-driven modelling framework for nanocrystal drug formulations using automated continuous flow crystallizers has several notable impacts: 1. Efficient Feasibility Studies: The classification algorithms enhance the efficiency of feasibility studies, providing a quicker and more accurate assessment of the operating space for nanocrystal drug formulations. This accelerates the decision-making process during the early stages of drug development. 2. Enhanced Surrogate Modelling: The multi-task Gaussian Process (GP) models contribute to more effective surrogate modelling. This leads to improved predictive capabilities, allowing for better control and optimization of the nanocrystal synthesis process. 3. Optimized Experimental Design: Active learning facilitates optimal sequential experimental design, enabling the selection of experiments that provide the most valuable information. This not only streamlines the experimentation process but also ensures that each experiment contributes significantly to model refinement. 4. Process Optimization through Bayesian Optimization: The incorporation of Bayesian optimization enables automated process optimization. This contributes to the development of self-optimizing platforms, reducing the need for manual intervention and accelerating the overall drug development timeline. 5. Applicability to Multiple Case Studies: The framework's successful application to Ibuprofen synthesis and its planned application to the synthesis of Ketoprofen demonstrate its versatility. The potential for broad applicability across different drug formulations enhances its significance in the pharmaceutical research and development landscape. 6. Utilization of State-of-the-Art Python Libraries: By utilizing advanced Python libraries like PyTorch and BoTorch, the framework aligns with current best practices in machine learning. This ensures that the modelling techniques employed are cutting-edge and benefit from the latest advancements in the field. 7. Time and Resource Savings: The streamlined and automated processes facilitated by the framework result in significant time and resource savings. This is particularly valuable in the fast-paced and resource-intensive field of pharmaceutical research and development. 8. Improved Decision Support: The developed framework serves as a robust decision support tool, aiding researchers and developers in making informed decisions at various stages of drug development. This, in turn, contributes to the overall efficiency and success of the drug development process. 
 
Title Uncertainty-aware predictive machine learning models to support decision-making in automated drug synthesis platform 
Description Technological developments have led to the automation of decision-making processes in process plant operations. To use machine learning models (ML) for online decision-making processes, ML models require to provide reliable predictions with relatively small amount of data. One of the approaches to achieve this feat is to develop uncertainty-aware predictive models such as gaussian process regression models and then combine them with Bayesian inference and optimization to support online decision-making relating process control and optimizations. The combined framework allows to quantify the prediction uncertainty of ML models and to produce data for rapid validation of ML models. Our goal is to apply this ML modelling framework to develop autonomous platforms for drug nanocrystals synthesis. 
Type Of Material Computer model/algorithm 
Year Produced 2023 
Provided To Others? No  
Impact The modelling framework developed in this work can be used as a tool to support automated decision-making in smart and responsive manufacturing systems. 
 
Description Advisory Board Meetings 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Advisory board meetings with project partners (GSK, Janssen, APC, Quotient Sciences, CPI). The meetings involved a) presentations from industrial partners aimed at clarifying their specific interests in the project and objectives for their involvement, b) presentations from researchers to update the partners about progress, achievements and challenges, c) discussion about progress against deliverables, project direction and possible support, d) discussion about further engagement opportunities (e.g., PDRA and PhD student co-supervision, MEng/MSc projects, secondments, and targeted one-to-one meetings). Impact: various technical suggestions taken on board; targeted one-to-one meetings organized to facilitate progress and overcome technical issues; exploration started for possible activities for further engagement and/or funding.
Year(s) Of Engagement Activity 2022,2023
 
Description Autonomous platforms for model identification: dream or reality? 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Industry/Business
Results and Impact This highlight talk was delivered withing the Inustrial Consortium Meeting (ICM) held on 8th of December 2023 at the Sargent Centre for Process Systems Engineering. The presentation covered the challenges in the integration of model identification algorithms and physical devices, relevant to NanoAPI. The presentation captured the interest of attendees ranging from industry to academia, who raised a number of interesting questions.
Year(s) Of Engagement Activity 2023
URL https://www.imperial.ac.uk/events/170171/sargent-centre-industrial-consortium-meeting-invitation-onl...
 
Description Autonomous platforms for online process modelling and model identification - Presentation at GSK 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Industry/Business
Results and Impact I have delivered a talk at GSK on the topic "Autonomous platforms for online process modelling and model identification". It was part of a research collaboration meeting organised between GSK and our research group - Galvanin System Identification Group (GSIG) at GSK.
Year(s) Of Engagement Activity 2023
 
Description CMAC Open day visit 
Form Of Engagement Activity Participation in an open day or visit at my research institution
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Visit of CMAC open days and visit of project collaborators.
Year(s) Of Engagement Activity 2022
URL https://cmac.ac.uk/events-database/cmac-open-day-2022-hub-and-articular-showcase
 
Description Galvanin System Identification (GSIG) group - presentation at GSK Stevenage 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Industry/Business
Results and Impact This presentation was delivered to introduce the range of research activities carried out in the Galvanin System Identification Group (GSIG), including the current research in NanoAPI related to model identification in a system for nanoparticles synthesis. This was an internal meeting within GSK at the Stevenage site, resulting in further collaborations within this area.
Year(s) Of Engagement Activity 2023
 
Description Guest lecture at University of Leeds 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Undergraduate students
Results and Impact Guest lecture at University of Leeds on "Nano and micro reactors: Flow chemistry for nanomaterial synthesis"
Year(s) Of Engagement Activity 2022,2023
 
Description Guest lecture at University of Leeds 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Undergraduate students
Results and Impact Guest lecture at University of Leeds on "Nano and micro reactors: Flow chemistry for nanomaterial synthesis". This year with the addition of automated crystallization.
Year(s) Of Engagement Activity 2023
 
Description ML-NanoAPI: A machine learning assisted automated platform for drug nanocrystals synthesis via antisolvent crystallization - Presentation at ECCE 14 & ECAB 7 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Postgraduate students
Results and Impact I have delivered a talk on the development of a computational framework that employs uncertainty-aware machine learning (ML) algorithms to drive an automated platform for the synthesis of drug nanocrystals to an audience of around 50 researchers and peers. I was happy to receive some good feedback from the audience. Also, the conference was a wonderful opportunity to share knowledge and skills.
Year(s) Of Engagement Activity 2023
 
Description Meeting with APC (Project Partner) 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Update on the progress of the research project. Discussion of related research activities, mutual interests and secondments. Discussion of avenues for further collaboration.
Year(s) Of Engagement Activity 2023
 
Description Model based design of experiments in autonomous model identification platforms: recent developments and open challenges 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact This plenary lecture was delivered at the BASF flow chemistry retreat held at Cumberland Lodge (WIndsor, UK) on 26th of June 2023 to an industrial audience of more than 50 people within the BASF chemical company, mostly from Germany. The presentation covered challenges in autonomous model identification in flow systems, a topic relevant to NanoAPI and received great attention, resulting in follow-up meetings for further collaborations.
Year(s) Of Engagement Activity 2023
 
Description Oral Presentation in European Congress of Chemical Engineering, Berlin 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Postgraduate students
Results and Impact Oral presentation at the 14th European Congress of Chemical Engineering, Berlin about "Automated continuous crystallization of ibuprofen in Impinging Jet Reactor with laser diffraction-based online crystal size analysis"
Year(s) Of Engagement Activity 2023
URL https://ecce-ecab2023.eu/Programme.html
 
Description Pharmaceutical manufacturing at UCL - preparing for the next centenary 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Industry/Business
Results and Impact A presentation was delivered by Prof. Eva Sorensen on the range of pharma applications covered by the systems engineering group at UCL, including NanoAPI activities. More than 50 attendees from industry and academia were present this Pharma Forum organised by the Sargent Centre for Process Systems Engineering (SCPSE) at Holiday Inn Hotel, Kensington, London.
Year(s) Of Engagement Activity 2023
URL https://www.imperial.ac.uk/process-systems-engineering/courses-and-seminars/pharmaceutical-manufactu...
 
Description Poster Presentation in symposium - Responsive Manufacturing Showcase - University of Strathclyde 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Showcase event organised by University of Strathclyde for those who were successful in the 2021 EPSRC Responsive Manufacturing Call with the aim of networking, discussing the research, and identifying new synergies and opportunities for collaboration. In the morning, each PI gave a presentation; in the afternoon, there was a researcher poster session.
Year(s) Of Engagement Activity 2022
 
Description Poster presentation at the Pharmaceutical Manufacturing Forum 2023 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Postgraduate students
Results and Impact Poster presentation at the Pharmaceutical Manufacturing Forum 2023 in Imperial College London
Year(s) Of Engagement Activity 2023
URL https://www.imperial.ac.uk/process-systems-engineering/courses-and-seminars/pharmaceutical-manufactu...
 
Description Presentation at CFRT conference in Dublin 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Poster presentation as well as short talk on "How to Close the Loop: Autonomous Flow Chemistry for Process and Material development"
Year(s) Of Engagement Activity 2023
URL https://www.cfrt-tks.com/
 
Description Presentation at the Medical School at the University of Sheffield 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Presenation on "Reproducible and Scalable Synthesis of Nanomaterials" to the Medical School, as well as lab visits and scientific exchange.
Year(s) Of Engagement Activity 2023
 
Description Presentation to industrial board of Sargent Centre for Process Systems Engineering 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Industry/Business
Results and Impact Presentation to industrial board on "Digital manufacturing: From the beaker to machine learning"
Year(s) Of Engagement Activity 2022
URL https://www.imperial.ac.uk/process-systems-engineering/industrial-consortium/
 
Description Project Progress Meeting with CPI 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Industry/Business
Results and Impact Presentation of Project Progress and future opportunities with CPI Team
Year(s) Of Engagement Activity 2024
 
Description Responsive Manufacturing Showcase - University of Strathclyde 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact Showcase event organised for those who were successful in the 2021 EPSRC Responsive Manufacturing Call. Aim: networking, discussing the research and identifying new synergies and opportunities for collaboration. In the morning, each PI gave a presentation; in the afternoon, there was a researcher poster presentation.
Year(s) Of Engagement Activity 2022
 
Description Spring into STEM presentation to motivate future scientists to engage in these disciplines. This was a faculty of engineering lecure series organised by UCL. 
Form Of Engagement Activity A broadcast e.g. TV/radio/film/podcast (other than news/press)
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Media (as a channel to the public)
Results and Impact Presentation (made public on YouTube) on "Smart Manufacturing in the Digital Era"
Year(s) Of Engagement Activity 2023
URL https://www.youtube.com/watch?v=Cm1MVi-jlEQ
 
Description Talk and pannel discussion at "Digital Precision: Lab Self-Optimisation for Nanoparticle Manufacturing" event organised by CPI 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Professional Practitioners
Results and Impact Presentation given on reactor technology for automated material synthesis and participating in pannel discussion
Year(s) Of Engagement Activity 2024
URL https://cdn2.assets-servd.host/smart-hoopoe/production/content/docs/V5-Nanoman-event-agenda.pdf?utm_...