Knowledge-driven drug repurposing using a comprehensive drug knowledge graph
Problem
New drug discovery and development is hugely expensive, so finding new uses for existing drugs is needed.
Why it matters
Repurposing existing drugs offers a cheaper, faster complement to de novo drug development.
What the AI did
Built a comprehensive drug knowledge graph from multiple drug knowledge bases and used path- and embedding-based representations as inputs to machine learning models predicting repurposing candidates.
Results
High predictive results for known diabetes mellitus treatments using only treatment information on other diseases; no numeric metrics stated
Impact
Demonstrated that knowledge-graph-driven ML can recover known diabetes treatments, supporting repurposing candidate prediction.
Record
| Venue | Health Informatics Journal |
|---|---|
| Published | 2020 |
| Work type | article |
| Authors | Yongjun Zhu, Chao Che, Bo Jin, Ningrui Zhang, Chang Su, Fei Wang |
| Citations | 73 (OpenAlex, at import time) |
| References | 46 |
| Model or method | path- and embedding-based machine learning models |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1177/1460458220937101 |
| Open access | gold |
| Keywords | knowledge graph, drug repurposing, machine learning, embeddings, diabetes |
Importance rating
significant — thousands to millions affected
Abstract states new drug discovery and development carry huge costs, making repurposing an important complement; industry-wide relevance for cheaper drug development, no scale quantified.