Traditional Approach
Previous studies rely on patent citations to measure innovation impact.
Limitation: Citations only capture explicit references—they miss when research influences innovation without direct citation.
This underestimates the true impact of basic science on pharmaceutical R&D.
Our Approach
We use BioBERT (biomedical language model) to extract gene and disease mentions directly from patent text.
This captures the actual scientific content being used in patents, not just what's cited.
Result: A comprehensive map of which gene-disease pairs appear in pharmaceutical patents—enabling direct measurement of knowledge translation.