In the biopharmaceutical industry, deviation management in manufacturing processes has become an essential aspect to ensure product quality and patient safety. Facing this challenge, Merck & Co., Inc.’s (MSD) Digital Manufacturing Data Science team has identified a unique opportunity to optimize this management using emerging technologies such as vector databases and generative artificial intelligence, supported by AWS services.
The goal of this new methodology is to leverage past deviations as a valuable source of knowledge, with the expectation of significantly reducing the time and resources required to investigate and resolve new deviations. This advancement promises to increase operational efficiency while ensuring compliance with the high standards required by Good Manufacturing Practices (GMP).
The trend towards the implementation of advanced technologies in the pharmaceutical industry is not new. However, the use of artificial intelligence in deviation management, a process that historically has required significant human effort and expertise, is beginning to show clear benefits. In this context, MSD has developed an innovative solution that employs generative AI to tackle the challenge of deviations.
The new platform is based on creating a comprehensive knowledge base from previous deviation reports, providing work teams with quick and accurate access to relevant information about similar incidents. To ensure effective search, a hybrid mechanism has been implemented with Amazon OpenSearch Service.
However, the solution has also taken into account the risk of overreliance on artificial intelligence. Therefore, human supervision protocols and quality controls have been incorporated to ensure that the generated information is up-to-date and verified. Additionally, security measures such as encryption and access controls are being used to protect sensitive and proprietary information.
Among the anticipated benefits of this new architecture, a significant reduction in the time and effort required to investigate deviations is highlighted, as well as easier access to historical information and more efficient data management. Founded on a Augmented Generation by Recovery (RAG) approach, this solution combines semantic search with relational databases to offer precise and contextualized information.
Looking to the future, MSD intends to integrate advanced semantic search capabilities through Amazon Bedrock Knowledge Bases. This move is part of a broader strategy aimed at establishing a new standard not only in deviation management, but also in improving quality processes in the manufacturing of biopharmaceutical products. With these efforts, MSD is on track to revolutionize its deviation management process and pave the way towards a more efficient and knowledge-based industry.
Referrer: MiMub in Spanish
