Digitalization Must Enable New Science, Not Just Go Paperless
Anna Codina, Senior Director at SciY (Bruker), argued that the true value of lab digitalization lies in enabling new scientific capabilities. She noted that while LIMS integration in regulated environments remains a significant challenge, the goal should be to make new things possible, “from predictive analytics to closed-loop experiments,” rather than simply making existing processes paperless.
- Integrating laboratory information management systems (LIMS) with existing instruments and higher-level platforms like electronic lab notebooks (ELNs) is a primary challenge, often requiring custom interfaces to handle differing data formats and communication protocols. Data migration from legacy systems presents a significant hurdle, with risks of data loss and the need for rigorous validation of transferred data to ensure accuracy and traceability in regulated environments. - The lack of standardized data management tools in the cell and gene therapy (CGT) space hinders efficiency, especially as production volumes increase and processes become more complex. This forces manufacturers to manage disparate data types, from batch records to supply chain tracking, increasing the risk of data silos. - Digital twin technology is being developed for viral vector manufacturing to simulate and predict process outcomes, which can help optimize yields. For instance, applying a digital twin for advanced process control has been shown to potentially increase productivity by 20%. - In viral vector development, machine learning (ML) algorithms are being used to accelerate capsid engineering and promoter optimization, significantly reducing development time and costs compared to traditional methods. Companies like Dyno Therapeutics are using AI to design and test vast libraries of AAV capsids to improve tissue selectivity and reduce immunogenicity. - Predictive modeling is being applied to bioreactor performance to tighten operator control and accelerate process development. These models use real-time data from tools like Raman spectroscopy to predict and control critical quality attributes (CQAs) such as glucose feed rate, which is particularly valuable in continuous manufacturing. - So-called "closed-loop" experiments, where results from one experiment automatically inform the parameters of the next, are becoming a reality through the integration of AI and robotics. This approach can enable running 200 to 400 experiments per day, drastically accelerating research and development timelines. - For autologous cell therapies, where the manufacturing process is patient-specific, Industry 4.0 principles are critical for managing the inherent variability of starting materials. Digital systems are needed to adjust process "recipes" in real-time to achieve consistent outcomes despite the biological differences between patient cells. - In GMP environments, the move is toward automated, "functionally closed" systems to minimize the risk of microbial and particulate contamination, a requirement of regulations like Annex 1. These systems, which can involve isolators and sterile single-use components, reduce the need for high-grade cleanrooms and manual interventions.
