Design principles for One Health digital disease surveillance: lessons from three deployed systems

Early detection of zoonotic and emerging infectious threats depends on surveillance that integrates signals across human, animal, and environmental domains. Most digital surveillance infrastructure, however, is built and maintained by well-resourced institutions, leaving a gap for settings — and individuals — that cannot sustain such systems. This paper describes a reproducible architecture for One Health digital disease surveillance, derived inductively from three systems that the author designed, built, and deployed on commodity cloud infrastructure. Rather than presenting any single system as a finished product, the paper abstracts five design principles common to all three — reliance on public authoritative sources, automated collection, append-only persistence, keyword-based signal detection, and aggressive cost minimisation — and shows how each principle was implemented and progressively refined across the three deployments. The systems span an early aggregator that overwrote state on each cycle, a persistent core with a deduplicated relational store and an eight-job scheduler, and a globally scaled collector that grew from roughly 102 to approximately 118 registered sources across six rollout stages. The contribution is not a validated surveillance result but a transferable engineering pattern: a blueprint by which a single practitioner, at a monthly cost on the order of ten US dollars, can stand up an operating One Health surveillance pipeline. The paper states honestly what such systems can and cannot yet do, and identifies prospective operational validation as the necessary next step.