Toward a Theoretically Grounded Framework for Digital Epidemiological Enquiry: Integrating Behavioral Models, Information Diffusion, and AI-Driven Infoveillance Using a COVID-19 Vaccine Case Study

Digital epidemiology has emerged as a critical complement to traditional surveillance by leveraging digitally generated data to monitor population health behaviors and perceptions in near real time. Despite rapid methodological advances, the field remains dominated by tool-centric approaches, with limited integration of behavioral theory capable of explaining how information exposure translates into health-related decision-making. This article advances a theory-building, analytically grounded framework for digital epidemiological enquiry and empirically illustrates its application using a COVID-19 vaccine discourse case study from the United Kingdom and the United States. Drawing on the Health Belief Model (HBM), infodemiology, and information diffusion theory, the framework links perceived susceptibility, severity, benefits, barriers, and cues to action with AI-enabled sentiment and topic analytics. Using a large corpus of vaccine-related Twitter data, natural language processing and topic modeling were employed to operationalize behavioral constructs in digital discourse. Findings demonstrate that sentiment and thematic patterns can approximate key behavioral dimensions, while also revealing important limitations related to structural context, trust, and collective narratives that are only partially visible in digital traces. The study contributes a theoretically informed, empirically grounded approach for advancing digital epidemiology as an explanatory and policy-relevant discipline.