Responsible Artificial Intelligence governance in oncology

Area of Science:

• Oncology
• Medical Informatics
• Artificial Intelligence

Background:

• Artificial Intelligence (AI) adoption in healthcare, particularly oncology, is accelerating.
• Existing AI frameworks are generic, lacking specialized governance for oncology.
• There is a critical need for robust AI governance models in cancer care.

Purpose of the Study:

• To report on a Comprehensive Cancer Center's Responsible AI governance model for clinical, operational, and research applications.
• To share the outcomes of the first year of AI governance committee activities.
• To introduce novel tools and methodologies for AI governance in oncology.

Main Methods:

• Established an AI Governance Committee to oversee AI implementation.
• Registered and monitored 26 AI models, including large language models and ambient AI pilots.
• Reviewed 33 nomograms and developed management tools: program model, information sheet, risk assessment, and lifecycle management.
• Utilized case studies and an "Express Pass" methodology for select AI models.

Main Results:

• Successfully implemented a Responsible AI governance model over one year.
• Managed 26 AI models, 2 ambient AI pilots, and reviewed 33 nomograms.
• Developed and shared novel AI governance tools and an "Express Pass" methodology.
• Identified key lessons learned and open research questions in AI governance for oncology.

Conclusions:

• This report presents one of the first comprehensive, large-scale Responsible AI governance models in oncology.
• The developed model and tools provide a scalable framework for managing AI in cancer care.
• Further research is needed to address evolving AI technologies and their governance in oncology.