Biobanking for intelligent medicine: assessment and evaluation with the SHARE principle.

Abstract

Biobanks are essential for intelligent medicine but face fragmentation and heterogeneity. No standardized framework exists for assessing biobank data value using public information; this study addresses this gap. We systematically evaluated 94 global biobanks (2010-2024) through a literature review and structured data extraction. Based on 12 international standards, we developed the 5-dimensional SHARE principle (Standardization, Hierarchical structuring, Analytical compatibility, Regulatory compliance, Evolutionary adaptability), operationalized into 10 indicators with a 4-tier scoring system. GPT-4o provided AI-supported prescoring, which was validated by 10 experts and through case studies, including RARPKB. The SHARE principle and a classification map of 94 biobanks were generated. AI and expert scoring showed substantial consistency (κ = 0.62; 95% CI, 0.54-0.70). Biobanks were categorized into 4 tiers: Traditional (60-69), Data-Driven (70-79), Knowledge-Guided (80-89), and Generative and Reasoning-oriented Biobank (90-100). Case validation confirmed utility for disease-specific biobanks. We highlight the principal findings, critically examine the reliance on public documentation, propose mitigation strategies, and discuss indicator weighting and implications for translational informatics. The SHARE principle provides a scalable, standardized method for assessing biobank data value, supporting biobank development, resource discovery, and the development of AI-driven biomedical ecosystems for intelligent medicine.