United States and European Union Artificial Intelligence Administrative Arrangement

On January 27, 2023, the United States (U.S.) and European Union (EU) signed an Administrative Arrangement to collaborate on research using artificial intelligence (AI), computing, and privacy-related technologies.

This arrangement represented a ground-breaking agreement designed to advance the use of AI in a number of fields, including in cancer research. In brief, the agreement sought to:

• create new opportunities for U.S. and EU subject matter experts (SMEs) from diverse AI fields to work together and explore innovative ways to use AI-focused research for the common good.
• enhance collaborative efforts between the U.S. and EU to advance technology by applying methods such as federated learning.
• give researchers in the U.S. and EU access to more detailed and data-rich models.
• stimulate discussion around other aspects related to AI such as preserving privacy, engaging the community, including underserved communities, as well as ethics and trustworthiness. 

NCI was fully engaged in helping advance this unique AI collaboration. Dr. Sylvia Gayle directed NCI’s efforts under the “Health and Medicine” focus group, providing leadership to five additional focus groups aiming to use AI to advance cancer research.

International AI Working Group

A trans-NCI working group launched the “Equitable and Engaged AI to Advance Biomedical Research” meeting series to feature speakers from the U.S. and EU discussing a variety of topics about AI in cancer research. For example:

• techniques for preserving privacy in AI.
• AI technologies and tools to foster patient engagement in cancer research.
• AI ethics and how technology could help address the needs of underserved communities.

International AI Challenges

NCI worked with fellow agencies to host international challenges to establish opportunities for U.S. and EU researchers to explore innovative ways to address cancer research pain points with AI. Two challenges included:

• Medical Image De-Identification Workshop 
  NCI hosted a 2-day workshop looking at challenges in de-identifying medical images to ensure patient privacy. The workshop brought together more than 600 developers, researchers, and data scientists from around the United States and Europe.
• Annual Brain Tumor Segmentation (BraTS) Challenges
  NCI, the U.S. Food and Drug Administration (FDA), the University of Pennsylvania, and Sage Bionetwork kicked off the 2023 International BraTS Challenge. That year, the challenges focused on AI techniques that used BraTS data sets to address diversity, glioma, meningioma, brain metastases, and pediatric tumors.

Interagency AI Collaborations

NCI further supported the initiative by collaborating with other government agencies aiming to use AI to advance cancer research.

NCI’s Center for Cancer Research (CCR) and the FDA   
CCR and the FDA worked together to make clinical trials more efficient and cost effective. The investigators are using AI/machine learning (ML) in digital oncology studies to aid in clinical cancer research and care. CCR and FDA also are piloting a method that automatically extracts common data elements from electronic health records to help in filling out electronic case report forms. If successful, this technology, along with a secure cloud-based system, could be useful for keeping clinical trial participants up to date on findings. 

Past Activity Leads:  Dr. Paolo Ascierto (National Tumor Institute, Naples, Italy), Dr. James Gulley (CCR), Dr. Mark Lawler (Queen’s University-Belfast), Dr. Jason E. Levine (CCR), Dr. Ignacio Melero (University of Navarra, Spain), and Dr. Sjoerd Van Der Burg (Leiden University-Netherlands)

NCI-Department of Energy (DOE) Collaborations 
NCI and DOE collaborated to explore the use of AI/ML and other advanced technologies. They were aligned with the work under the U.S.-EU agreement, and included:

• AI-Driven Multi-Scale Investigation of the RAS/RAF Activation Lifecycle (ADMIRRAL): Used AI to develop effective strategies for cancer diagnosis and therapy.
• Innovative Methodologies and New Data for Predictive Oncology Model Evaluation (IMPROVE): Offered a framework for comparing and evaluating cancer drug response, which was a key research area in the AI Administrative Arrangement.
• Modeling Outcomes Using Surveillance Data and Scalable AI for Cancer (MOSSAIC) and NCI’s Surveillance, Epidemiology and End Results (SEER) Program: Actively researched AI/deep learning and natural language processing.

Past Activity Lead: Dr. Emily Greenspan (NCI CBIIT)

If you have additional questions about the past project or its outputs, email NCI CBIIT.