Biopharmaceuticals and personalized therapies require a different approach to production which in turn impacts manufacturing processes and inspection. Inspection solutions need to be flexible and scalable. Small batches, for example, require require modular filling systems and a modular inspection approach.

Italy-based Stevanato Group, a producer of glass primary packaging and provider of capabilities for drug delivery systems, recently launched a modular, robotic inspection system that incorporates artificial intelligence with accrued machine learning capabilities. The Vision Robot Unit (VRU) is an autonomous module capable of automatically inspecting particle properties without human intervention.

Raffaele Pace, Engineering Head of Operations for Stevanato Group, discusses today’s operational challenges, the new Vision Robot Unit, and large-scale primary packaging for potential COVID-19 vaccines. –KB
 

Contract Pharma: What are some of today’s operational challenges and how does the Vision Robot Unit help to overcome them?

Raffaele Pace: The main drivers of change in the pharma industry are the development of biotech drugs and personalized therapies. Both these elements require a completely different approach to production and impact the manufacturing processes, including the inspection phase.

Operational challenges – especially in high-value products like biotech and personalized drugs – include production of very small batches, which requires a critical combination of flexibility and scalability; a natural variability in production in which vendors are required to deliver “intelligent” inspection solutions; and emerging packaging platforms, such as soft bags.

To address these challenges, Stevanato Group’s VRU provides enhanced, AI-backed inspection for products with varying characteristics – an obstacle that frequently leads to heightened false rejection rates.

Small batches require modular filling systems and a modular inspection approach. On the other hand, sometimes pharma companies’ inspection capabilities are too oversized, leading to maintenance costs and pricing. VRU represents a solution to both scenarios. It’s a modular and scalable inspection system, as multiple vision inspection units can be added to increase throughput

Often re-inspection of gray or rejected items is performed by manual operators. VRU represents an alternative and reliable solution as it achieves high inspection performance thanks to Artificial Intelligence. Further, some products require particular movements for preparing them for inspection. Using robots, the VRU ensures significant flexibility in these movements, simulating human-like gestures.

CP: What role does artificial intelligence play in optimizing pharma visual inspection processes? Where is there potential to further the use of AI?

RP: Artificial Intelligence applied to visual inspection offers consistent improvement in inspection quality. Our case studies show how Detection Rate (DR) increases compared to the standard analysis tools while reducing the False Rejection Rate (FRR). This allows pharma companies to bring to market safer products reducing false positives and benefit from cost savings, decreasing false negatives.

Additionally, using Artificial Intelligence (AI) could unlock significant potential. For example, the model can systematically embed a much larger knowledge base in comparison to a classic rule-based solution. 

Furthermore, AI models can generalize acquired information to classify several similar objects, a knowledge abstraction that can be used to optimize procedures and therefore costs. For example, different syringe flanges from various suppliers can be inspected using the same AI model.

CP: What challenges do you anticipate for potential COVID-19 vaccine/s and large-scale primary packaging?

RP: We are all working in unprecedented circumstances that require extraordinary efforts in a condensed timeframe of 12-18 months. We are providing global coverage with all our production sites, hiring more people and redesigning shifts to ensure we protect our staff even while accelerating production.

At the same time, we are investing in additional capacity. We estimate the global glass primary packaging market consumption at 45-50 billion units, with 25% of the overall consumption represented by glass vials. There are many possible scenarios for vaccination strategies among geographies, which can be confirmed once vaccines, treatments and dosages are defined. We estimate an increased need of 1-2 billion vials (monodose or multidose) linked to COVID-19, which we believe can be provided by the current main producers of glass primary packaging.

Thanks to our integrated capabilities, we are ready to support the inspection needs of pharma companies as well. We have partnered with major players on this landscape, providing the equipment needed to face this emergency. Further, we believe that - leveraging on the extensive database of existing vaccines, which are currently inspected in the market - Artificial Intelligence could support the fast realization of an inspection recipe for COVID-19 candidate vaccines.