BioKMS-HAG: A hierarchically guided biomedical and space science knowledge fine-grained mining system

Area of Science:

• Biomedical Informatics
• Artificial Intelligence
• Space Life Sciences

Background:

• Rapid growth in biomedical data and AI present opportunities for scientific discovery.
• Accessible tools are needed to integrate advanced AI methods for leveraging this data.

Purpose of the Study:

• Develop the Biomedical Knowledge Mining System (BioKMS) integrating knowledge graph construction and link prediction.
• Implement the Hierarchy Guided Approach (HAG) for fine-grained biomedical hypothesis generation.
• Enhance AI-driven knowledge discovery in biomedical and space life sciences.

Main Methods:

• Developed BioKMS, an interactive platform for exploring biomedical and space life science data.
• Integrated HAG, a hierarchical tree-guided Graph Neural Network model, into BioKMS for refined link prediction.
• Utilized knowledge graph construction and advanced link prediction techniques.

Main Results:

• Achieved significant improvements in predictive accuracy, especially for complex biomedical entities.
• Demonstrated reduced computational overhead in hypothesis generation.
• Enabled seamless knowledge discovery across disciplines.

Conclusions:

• The integration of HAG into BioKMS streamlines hypothesis generation and accelerates research efficiency.
• This user-friendly system showcases the transformative potential of combining AI methodologies for scientific research.
• Paves the way for more efficient biomedical and space science research through advanced AI integration.