The case for homebrew AI in diagnostic pathology

Area of Science:

• Digital pathology
• Artificial intelligence in medicine
• Cancer diagnostics

Background:

• Artificial intelligence (AI) holds significant promise for enhancing cancer diagnostics and patient care through digital pathology.
• Commercialization of AI tools requires regulatory clearance as in vitro diagnostic (IVD) devices, a complex and costly process.
• Many AI tools may not be commercially viable, limiting their adoption.

Purpose of the Study:

• To advocate for the use of 'homebrew' AI systems within existing quality frameworks for lab-developed tests.
• To explore the regulatory landscape for homebrew AI in digital pathology in the USA and EU.
• To propose strategies for the safe and effective integration of homebrew AI in pathology practice.

Main Methods:

• Analysis of current regulatory provisions in the USA and European Union regarding lab-developed tests and AI.
• Review of potential impacts of new US Food and Drug Administration (FDA) rules on homebrew AI.
• Development of recommendations for integrating homebrew AI into pathology workflows.

Main Results:

• Homebrew AI models are currently permissible under existing regulatory frameworks in the USA and EU.
• A new US FDA rule may pose challenges to the continued use of homebrew AI.
• The study outlines the regulatory status and potential integration pathways for homebrew AI.

Conclusions:

• Advancing homebrew AI within established lab-developed test frameworks can improve patient access to digital pathology tools.
• Addressing regulatory challenges, particularly concerning the new US FDA rule, is crucial for the successful implementation of homebrew AI.
• Facilitating the integration of homebrew AI can enhance diagnostic capabilities in pathology.