Extended Data Fig. 1: Linking global burden of disease causes to research publications.
From: Global distribution of research efforts, disease burden, and impact of US public funding withdrawal
Schematic visual showing triangulated method for linking disease causes to disease-specific research publications, including a medical expert-derived ground truth, a crosswalk via International Classification of Disease Codes (ICD), and a large language model (LLM) approach using ChatGPT (Model gpt-4o). Accuracy statistics are provided at the MeSH-term level for cardiovascular diseases (467 C-branch terms) as well as at the publication level (1.6 million articles).
