Unlocking biomedical data sharing: A structured approach with digital twins and artificial intelligence (AI) for open...

Area of Science:

• Biomedical Informatics
• Computational Biology
• Data Science

Background:

• Data sharing is crucial for scientific advancement but hindered by privacy concerns.
• Sensitive data, especially in biomedicine, requires secure and ethical sharing methods.
• Lack of computational expertise limits data utilization for many scientists.

Purpose of the Study:

• To foster FAIR (Findable, Accessible, Interoperable, Reusable) sharing of sensitive biomedical data.
• To develop an integrated computational approach for data utilization and enrichment.
• To enable scientists without coding experience to work with sensitive datasets.

Main Methods:

• An in silico pipeline generating synthetic data for controlled material sharing.
• A cyberinfrastructure facilitating computational notebook sharing without local installation.
• A digital twin model using cancer datasets as a use case for open data availability.

Main Results:

• A metadata approach for generalizable computational model descriptors and validation using existing data.
• A virtual lab book within a cloud-based system for user interaction and data management.
• Qualitative testing indicating a need for comprehensive guidelines to enhance user acceptance.

Conclusions:

• The framework promotes Open Science through reproducible data generation and incomplete data interpolation.
• The system is expandable from biomedical to other domains.
• Future enhancements, including a graphical user interface, could broaden interdisciplinary applicability.