Measuring the growth of infectious disease modelling publications and their impact on policymaking: a Large Language Model-assisted bibliometric review

BackgroundInfectious disease modelling (IDM) is increasingly being used to understand disease transmission and inform public health policy. However, its growth and policy influence has never been quantified, possibly because of the volume of literature involved. The development of large language model (LLM)-assisted reviewing allowed us to quantify the expansion of IDM publications over time, trends in policy citations of IDM research, and regional disparities in research contributions and citations in policy documents. MethodsAn LLM-assisted bibliometric review was conducted using Embase, Medline, and Web of Science, identifying IDM publications from database inception to December 2024 using GPT-4o. Inclusion criteria encompassed peer-reviewed studies employing mathematical, statistical, or mechanistic models for infectious disease outcomes. LLM accuracy was iteratively refined by human review. We extracted publication metadata, geographic scope, and policy citations using Overton, a global database of policy documents. Growth trends were analysed using negative binomial regression models, and geographic disparities were assessed based on World Bank income classifications. ResultsA total of 33,255 IDM publications were identified over 44 years, with distinct growth phases. The LLM selection and data extraction achieved 98% and 100% accuracy respectively, compared to human search. Publication volume increased from the time of HIV/AIDS emergence, experiencing steady expansion through multiple outbreaks (Ebola, SARS, H1N1, MERS, Zika), and surged sharply just before the COVID-19 pandemic before declining post-2021. Recorded policy citations accounted for 1.7% of IDM publications, closely following the overall publication trend, peaking during periods of heightened public health attention. Policy citations largely reflected national research outputs, with notable cross-regional adoption of IDM evidence in some settings. ConclusionStrengthening the integration of IDM evidence into policymaking processes may require addressing geographic disparities in research output (and its recording in international databases), enhancing cross-regional collaboration, and improving mechanisms for policy uptake. Following the COVID-19 pandemic, policy citations declined despite continued growth in IDM literature, suggesting a potential lag or shift in policymaking priorities.