Artificial Intelligence for Multi-Scale Predictions in Infectious Health Emergencies – AIMPI

Timely detection of emerging diseases is essential for preventing widespread outbreaks, minimizing public health impacts, and assessing the effectiveness of interventions. This research program introduces an innovative mathematical framework that integrates cutting-edge artificial intelligence (AI) and machine learning (ML) technologies to enhance the accuracy and efficiency of early warning systems (EWS) of emerging diseases. To date, monitoring for potential outbreaks has relied on surveillance-based approaches like identifying zoonotic hosts, using phylogenetics to locate animal reservoirs, and characterizing viruses with pandemic potential in labs, with data feeding into advanced EWS that apply models (e.g., by fitting a transmission model) to detect changing and irregular patterns. However, these model-based systems are limited by uncertainty and sparsity of data, lack the flexibility to capture complex disease transmission dynamics, and struggle to effectively incorporate multi-modal, heterogeneous, and unstructured data. By leveraging rapid advancements in AI and ML, the proposed framework will create a state-of-the-art EWS capable of identifying anomalous and atypical patterns in heterogeneous, unstructured datasets, such as hospital data, wastewater epidemiology, social media trends, event-based surveillance, and high-throughput sequencing data, alongside traditional surveillance data. Key objectives are to (i) quantify emergence risk and establish a threshold of emergence, considering factors such as pathogen evolution, declining vaccination coverage, waning immunity, mobility patterns; (ii) use heterogeneous datasets to dynamically learn key parameters indicative of an emerging outbreak. Outcomes of the framework (including developing commercial applications) include rapid responses to infectious disease outbreaks, improved simulation capabilities for complex disease spread scenarios, and more effective health programs.