Knowledge Graph-based Neurodegenerative Diseases and Diet Relationship Discovery
This work addresses the challenge of finding preventive dietary strategies for neurodegenerative diseases, which lack effective treatments, but it is incremental as it applies existing methods to a new dataset.
The researchers tackled the problem of identifying potential dietary influences on neurodegenerative diseases by constructing a knowledge graph from 4,300 publications using PubTator for annotations and node2vec for embeddings, resulting in the discovery of several food-related species and chemicals that may impact these diseases.
To date, there are no effective treatments for most neurodegenerative diseases. However, certain foods may be associated with these diseases and bring an opportunity to prevent or delay neurodegenerative progression. Our objective is to construct a knowledge graph for neurodegenerative diseases using literature mining to study their relations with diet. We collected biomedical annotations (Disease, Chemical, Gene, Species, SNP&Mutation) in the abstracts from 4,300 publications relevant to both neurodegenerative diseases and diet using PubTator, an NIH-supported tool that can extract biomedical concepts from literature. A knowledge graph was created from these annotations. Graph embeddings were then trained with the node2vec algorithm to support potential concept clustering and similar concept identification. We found several food-related species and chemicals that might come from diet and have an impact on neurodegenerative diseases.