Barry Lorbetskie
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Barry Lorbetskie has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Michel Girard
Center for Oncology, Radiopharmaceuticals and Research, Biologic and Radiopharmaceutical Drugs Directorate, Health Products and Food Branch, Health Canada, Ottawa, ON K1A 0K9, Canada
2 shared publications - Simon Sauvé
Centre for Oncology, Radiopharmaceuticals and Research, Biologic and Radiopharmaceutical Drugs Directorate, Health Products and Food Branch, Health Canada and World Health Organization Collaborating Center for Standardization and Evaluation of Biologicals, Ottawa, ON K1A 0K9, Canada
2 shared publications - Changgui Li
National Institutes for Food and Drug Control, Beijing, China
1 shared publication - Huixin Lu
Health Canada
1 shared publication - Junzhi Wang
National Institute for Food and Drug Control, Beijing, China
1 shared publication - Laura Durno
Centre for Biologics Evaluation, Biologics and Genetic Therapies Directorate, Health Canada, 251 Sir Frederick Banting, Tunney's Pasture, Ottawa, Ontario, K1A 0K9, Canada
1 shared publication
Latest Publications
- Open Access
Advancing Reversed-Phase Chromatography Analytics of Influenza Vaccines Using Machine Learning Approaches on a Diverse Range of Antigens and Formulations
One concern in the yearly re-formulation of influenza vaccines is the time-consuming manufacturing of vaccine potency reagents, particularly for emergency responses. The continuous evaluation of modern techniques such as reversed-phase (RP) chromatography is an asset for streamlining this process. One challenge with RP methods, however, is the need to re-optimize methods for antigens that show poor separation, which can be highly dependent on analyst experience and available data. In this study, we leveraged a large RP dataset of influenza antigens to explore machine learning (ML) approaches of classifying challenging separations for computer-assisted method re-optimization across years, products, and analysts. Methods: To address recurring chromatographic issues—such as poor resolution, strain co-elution, and signal absence—we applied data augmentation techniques to correct class imbalance and trained multiple supervised ML classifiers to distinguish between these peak profiles. Results: With data augmentation, several ML models demonstrated promising accuracy in classifying chromatographic profiles according to the provided labels. These models effectively distinguished patterns indicative of separation issues in real-world data. Conclusions Our findings highlight the potential of ML as a computer assisted tool in the evaluation of vaccine quality, offering a scalable and objective approach to chromatogram classification. By reducing reliance on manual interpretation, ML can expedite the optimization of analytical methods, which is particularly needed for rapid responses. Future research involving larger, inter-laboratory datasets will further elucidate the utility of ML in vaccine analysis.
Open AccessAdvancing Reversed-Phase Chromatography Analytics of Influenza Vaccines Using Machine Learning Approaches on a Diverse Range of Antigens and Formulations
One concern in the yearly re-formulation of influenza vaccines is the time-consuming manufacturing of vaccine potency reagents, particularly for emergency responses. The continuous evaluation of modern techniques such as reversed-phase (RP) chromatography is an asset for streamlining this process. One challenge with RP methods, however, is the need to re-optimize methods for antigens that show poor separation, which can be highly dependent on analyst experience and available data. In this study, we leveraged a large RP dataset of influenza antigens to explore machine learning (ML) approaches of classifying challenging separations for computer-assisted method re-optimization across years, products, and analysts. Methods: To address recurring chromatographic issues—such as poor resolution, strain co-elution, and signal absence—we applied data augmentation techniques to correct class imbalance and trained multiple supervised ML classifiers to distinguish between these peak profiles. Results: With data augmentation, several ML models demonstrated promising accuracy in classifying chromatographic profiles according to the provided labels. These models effectively distinguished patterns indicative of separation issues in real-world data. Conclusions Our findings highlight the potential of ML as a computer assisted tool in the evaluation of vaccine quality, offering a scalable and objective approach to chromatogram classification. By reducing reliance on manual interpretation, ML can expedite the optimization of analytical methods, which is particularly needed for rapid responses. Future research involving larger, inter-laboratory datasets will further elucidate the utility of ML in vaccine analysis.