Registries, Databases and Repositories for Developing Artificial Intelligence in Cancer Care.

Area of Science:

• Medical Informatics
• Artificial Intelligence
• Big Data Analytics

Background:

• Modern artificial intelligence (AI) shows promise in medicine but requires substantial data.
• Existing healthcare datasets (cancer registries, imaging, genetics, clinical data) are underutilized for AI development.
• The integration of healthcare big data with AI techniques faces significant, yet unaddressed, challenges.

Purpose of the Study:

• To identify and discuss structural reasons and barriers preventing AI from fully leveraging existing healthcare big data.
• To propose strategies for overcoming these obstacles and enabling progress in AI-driven healthcare research.
• To share practical experiences in data integration through a specific initiative.

Main Methods:

• Analysis of barriers using the framework of the 6Vs of Big Data.
• Application of FAIR (Findability, Accessibility, Interoperability, Reuse) data criteria.
• Case study of The Brain Tumour Data Accelerator initiative for data integration.

Main Results:

• Identified structural impediments hindering the combination of healthcare big data and AI.
• Demonstrated the utility of the 6Vs and FAIR principles in analyzing data integration challenges.
• Showcased successful data fragmentation and enrichment through The Brain Tumour Data Accelerator.

Conclusions:

• Overcoming data-related barriers is essential for unlocking AI's potential in medicine.
• The FAIR principles and structured data approaches are key to enabling AI development.
• Further efforts are needed to address the limitations of current data integration strategies for AI.