Mining online behavioural data to measure zoonotic spillover risk

Bríon Ó Conchubhair, Maxwell J Farrell , Rory Gibb, Luci Kirkpatrick, David W. Redding, Gregory F Albery

Efforts to predict zoonotic disease dynamics rarely directly quantify the fine-scale animal-human interactions that drive spillover. Public digital data (i.e., iEcology) could capture these behaviours at large scales, across time and space. Further, recent statistical advances allow us to semi-automate these processes, providing a framework for continuous exposure monitoring and risk prediction. Using online reviews describing macaque behaviour across southeast Asia, we demonstrate that iEcology provides a usable data source for risky animal-human interactions. We discuss challenges in operationalising these approaches, including ethical considerations and correcting spatial, taxonomic, and socioeconomic bias. In the Big Data era, semi-automated workflows like these could pave the way to dynamically updating risk maps for use in conservation and public health planning and interventions.