Review Article

Digital Epidemiology and Mathematical Modelling for HIV and Infectious-Disease Early Warning: Integrating AI, Genomics, Mobility, and Routine Surveillance

Abstract

The growing availability of clinical, genomic, mobility, environmental, and digital data is expanding the capacity of infectious-disease surveillance beyond retrospective description toward earlier detection and prediction. However, greater data availability and increasingly sophisticated analytical methods do not automatically produce effective early warning. This narrative review critically examines the roles of artificial intelligence (AI), mathematical epidemiological modelling, genomic epidemiology, mobility data, digital signals, and routine surveillance in HIV and broader infectious-disease early warning. Across these domains, the evidence indicates that individual data streams provide complementary but incomplete views of epidemic dynamics. AI can identify complex patterns, mathematical models can represent transmission processes, genomics can reveal pathogen evolution and transmission relationships, and mobility data can characterize population connectivity; yet each remains constrained by data quality, representativeness, uncertainty, transportability, and implementation challenges. The central limitation of the field is therefore not simply predictive performance, but the difficulty of converting fragmented signals into validated intelligence that can support timely public-health decisions. To address this gap, this review proposes an Integrated Digital Epidemiology Early-Warning Framework that links heterogeneous data inputs with integration, analytical modelling, epidemic intelligence, validation, warning, public-health action, and continuous evaluation and learning. The framework adds an explicit validation gate between prediction and warning, rules for reconciling conflicting signals and propagating uncertainty, tiered criteria for progressing from signal to warning and action, and a staged pathway for implementation and prospective validation in resource-limited settings, including for HIV surveillance. Future systems should be evaluated not only by predictive accuracy but also by lead time, calibration, external validity, equity, actionability, and effects on population outcomes. The next frontier is to build integrated systems capable of transforming heterogeneous signals into timely, trustworthy, equitable, and actionable public-health intelligence. 

Authors affiliation:

Ismaila Muhammed (ORCID): Department of Mathematics, Khalifa University, Abu Dhabi, United Arab Emirates.

How To Cite: I. Muhammed. Digital Epidemiology and Mathematical Modelling for HIV and Infectious-Disease Early Warning: Integrating AI, Genomics, Mobility, and Routine Surveillance. Infectious, 3(2):43–58, 2026. https://doi.org/10.58612/infectious325