Machine learning (ML) has seen a rapid growth of interest in the field of bio-therapeutics manufacturing over the years, aiding in advancements of drug discovery, process development, and upstream optimization. These applications have benefited greatly from the wealth of real-time sensing data that companies have accrued over many generations of discovery. Most of these ML applications have been applied to the front-end or R&D phase of bio-therapeutic manufacturing with its application to drug discovery alone expected to reach a market volume of $10 B by 2024. However, the growing introduction of miniaturized bio-therapeutic platforms utilizing rapid ‘cell-free’ technologies have uncovered a growing desire for instantaneous quality assessment or ‘real-time release’. Here we propose a ML framework for the real-time estimation of critical quality attributes (CQA’s) of bio-therapeutics using in-line process analytic technologies (PAT’s) in end-to-end bio-therapeutic manufacturing. The algorithm employs a deep layer artificial neural network (ANN) with a gaussian kernel applied at the feature input to the network. A genetic algorithm (GA) is used for the optimization of the parameters for the both the kernel and ANN to maximize the performance. The framework is verified on a miniaturized cell-free bio-therapeutic manufacturing platform that produces small batch therapeutics. The platform can incorporate real-time sensing such as multi-wavelength absorbance, fluorescence, conductivity, Raman spectroscopy, size exclusion chromatography (SEC), and pressure. Furthermore, a training data set is developed by perturbing repeated purification of a target protein from a lysate-based model system. The CQA’s can be measured with techniques such as Mass Spectrometry, SDS-PAGE, Capillary Electrophoresis, functional assay, and dynamic light scattering. The real-time assessment of quality attributes such as purity, aggregation, and integrity have strong implications on both the front-end and back-end of bio-therapeutic manufacturing as we inch closer to acquiring ‘real-time release’.

Machine learning techniques for real-time estimation of critical quality attributes (CQAs) in miniature bio-therapeutic manifacturing systems

Macromolecular Chemistry: The Second Century

[Author Benefits](https://scimeetings.acs.org/?utm_source=pubs_content_marketing&utm_medium=website&utm_campaign=0320_MCF_NPI_Launch_Spring_Homepage&ref=pubs_content_marketing)

[How to use SciMeetings](https://storage.googleapis.com/dl.morressier.com/acs/SciMeetings-ACSSpring2020-HowToGuide_M3020-Final.pdf)

[Frequently Asked Questions](https://storage.googleapis.com/dl.morressier.com/acs/SciMtng_ACSNMSp2020_FAQ_M3120_Final.pdf)

ACS Spring 2020 National Meeting & Expo

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.