Intelligent Veterinary Disease Management Driven by Knowledge Graph for Conservation Breeding of Captive Forest Musk...

Area of Science:

• Veterinary Medicine
• Bioinformatics
• Artificial Intelligence

Background:

• Forest musk deer farming faces high disease morbidity (abscess, enteritis, pneumonia, parasitic infections).
• Early disease symptoms are often subtle and difficult to diagnose manually.
• Current diagnostic methods can stress animals and delay treatment.

Purpose of the Study:

• To construct a comprehensive disease knowledge graph for forest musk deer.
• To improve the accuracy and efficiency of disease diagnosis and management in farmed musk deer.
• To integrate fragmented clinical and literature data into a structured format.

Main Methods:

• Utilized an improved BRW-GPLinker joint entity-relationship extraction approach.
• Developed a Boundary-Aware Module for precise entity detection.
• Implemented a Relative Distance Bias Module and a Weighted Sparse Multi-label Cross-Entropy loss function.
• Extracted disease entities, symptoms, etiological associations, and treatments from diverse data sources.

Main Results:

• Achieved an F1 score of 0.887 on a custom dataset.
• Demonstrated strong generalization capabilities on external medical datasets.
• Successfully transformed unstructured data into a structured disease knowledge graph.
• Facilitated rapid retrieval of forest musk deer disease information.

Conclusions:

• The developed knowledge graph significantly enhances veterinary decision-making for forest musk deer.
• This approach aids in proactive forest musk deer health management.
• The methodology offers a scalable solution for organizing animal disease information.