Biomedical knowledge graphs
Connect drugs, targets, publications and evidence into one graph.
Research · Experimental
Vink Drug
Biomedical Intelligence
Vink Drug researches biomedical knowledge graphs for drug interactions and evidence discovery. It keeps what is known separate from what a model predicts.
Approach
Known evidence is never mixed with predictions
Every statement is shown as one or the other, and predictions are never presented as established biomedical facts.
Known evidence
Traceable to sources
- Relationships reported in publications and databases
- Provenance recorded for each relationship
- Shown with the source it comes from
Model predictions
Hypotheses, not facts
- Possible interactions suggested by graph models
- Clearly labelled as predicted
- Meant to guide research, not to inform treatment
Research areas
Under development
Drug-drug interaction modelling
Model known interactions and how they relate to each other.
Evidence discovery and provenance
Find supporting evidence and keep track of where it comes from.
Graph machine learning
Learn patterns in the graph that may point to unexplored interactions.
Future interaction prediction
Explore predictions of interactions that are not yet documented.
Status
Research, not a clinical product
Vink Drug is a research initiative. It is experimental and under development, and nothing here is medical advice or a clinical tool.
We are interested in collaborations with researchers and organisations working on drug interactions and evidence.