Fraunhofer Institute for Production Technology IPTResearch project "Qua²ntum"
Production quality and control are increasingly supported by AI systems to ensure consistent, efficient, and resource-efficient production. In highly regulated markets such as the pharmaceutical industry, strict quality control ensures the safety and efficacy of medicines. In order to reap the benefits of AI, an additional management system is required to control and monitor the AI systems themselves.
Quality² - Quality assurance with and for AI
The research project "Qua²ntum" aims to enable the use of AI systems for quality assurance in the pharmaceutical industry. In order to create transparency in decision-making processes and to meet strict regulatory requirements, the traceability of computer-aided decisions must be ensured.
The task of the "Qua²ntum" project is therefore to develop a comprehensive quality management system (QMS) that is specifically tailored to the requirements of AI systems. This system defines processes, responsibilities and documentation during the development of AI systems in order to provide the required traceability. It ensures transparency and robustness of the AI methods and thus achieves traceability of the decisions made.
AI quality management system in the pharmaceutical industry
The first step in the project is to determine the requirements and research the applicable regulations in detail. Both the AI quality management system and the AI algorithms are then developed on this basis. Three use cases from the pharmaceutical industry, including the inspection of empty bottles, bottle neck finishes, and crimp caps, demonstrate the effectiveness of the system. In particular, for empty bottle inspection, the system checks that there are no particles in the empty bottle. Neck finish inspection checks the integrity of the bottle neck surface, and crimp cap inspection ensures proper seal positioning on the filled bottle. These use cases demonstrate not only the functionality of the QMS, but also the practical applicability and effectiveness of the AI methods.
The increased transparency of the decisions made by the AI models should not only increase trust in these systems, but also ensure that they perform the tasks assigned to them in a reliable and traceable manner. In this way, the project contributes to improving the quality of production in highly regulated markets and to ensuring the safety and efficacy of products that affect millions of people worldwide.
Funding
The research project "Qua²ntum" is funded by the German Federal Ministry of Education and Research (BMBF) within the framework of the "KI4KMU" guideline for the funding of projects on the topic "Research, development and application of artificial intelligence methods in SMEs".
Project partner
- Fraunhofer Institute for Production Technology IPT, Aachen
- OCTUM GmbH, Ilsfeld
- Aimino Tech GmbH, Karlsruhe
- Sartorius AG, Göttingen
- Groninger Holding GmbH & Co. KG, Crailsheim
- ROTZINGER PharmaPack, Waiblingen