Researchers at Johns Hopkins University and Duke University have developed PandemicLLM, an AI model that leverages large language modelling (LLM) to forecast infectious disease spread with unprecedented accuracy. Unlike traditional statistical models that often struggle amid rapidly changing conditions, PandemicLLM integrates diverse data streams, including epidemiological trends, public health policies, genomic surveillance, and state-level demographics, to generate nuanced predictions.
Retrospective evaluation during the COVID-19 pandemic across US states demonstrated that PandemicLLM outperformed existing state-of-the-art forecasting tools, especially during dynamic periods such as variant emergence or policy shifts. Its ability to incorporate real-time data and reason about interactions among variables allows it to accurately predict infection and hospitalization trends up to three weeks in advance.
This model’s multidimensional approach addresses critical limitations exposed during past outbreaks, enabling more agile and informed public health responses. Beyond COVID-19, PandemicLLM’s adaptable framework holds promise for forecasting other infectious diseases like avian influenza, monkeypox, and RSV.
Moreover, the researchers are exploring behavioral modelling integrations to predict individual health decisions, which could further refine disease management strategies. As infectious threats continue to evolve, tools like PandemicLLM represent a vital step forward in harnessing AI for proactive, data-driven public health.
Article written by Tecnhology Networks team
06/06/2025
Source:
Technology Networks