Compliance expectations for colony counting on agar plates have evolved over the years resulting in increased scrutiny throughout the pharmaceutical industry. Regulatory concerns exist over the reliability of plate counts associated with manual human reading, as well as potential data integrity issues. However, despite these concerns the adoption of automated systems for colony counting remains limited within the industry, due in part to the challenge of validating new instrumentation to demonstrate the benefits of the chosen technology outweigh risks of adoption and performance is demonstrated to be non-inferior to current processes.

In this paper, the validation considerations for an AI-based technology, APAS Independence, for the automated reading of culture plates are discussed. Whilst no compendial methods exist for the adoption of technologies that incorporate AI, the testing outlined in the compendial methods for the assessment of alternative microbiology methods provides a good basis for assessing the performance of such technologies. These tests have been refined, using the company’s understanding of the technology as well as previous discussions with regulators, to create an approach with sufficient scientific rigour to demonstrate the performance of the system. Following this primary validation, it is incumbent on the end user to consider any site-specific performance qualification (PQ) requirements to demonstrate suitability for use in real world operations.

Journal: American Pharmaceutical Review

Date: July / August 2024

Authors: Vanessa Figueroa, Andrew Gravett, Karen Capper, Steven Giglio

Citation: Vanessa Figueroa, Andrew Gravett, Karen Capper, Steven Giglio. 2024. Evolving Validation Strategies for AI-based Colony Detection using the APAS® Independence. American Pharmaceutical Review. Volume 27, Issue 5.