In this section, we begin with the seminal case of Google Flu Trends. It was a surveillance tool that Google launched in 2008 to estimate influenza activity in near-real time. GFT models raised hope of faster and easier estimates than "old-school" methods of data collection and statistical analysis, claiming the ascendance of data-driven approaches. In 2015, however, after several major stumbles in subsequent influenza seasons, GFT stopped publishing estimates.
But scientific efforts on harnessing search queries to nowcast or forecast epidemics have continued to the present. We review publications along these lines of research. We focus on one issue, the "blue team dynamics". This describes a process where the algorithm producing the data has been modified by the service provider in accordance with their business model, inducing specific user behaviors and introducing patterns into data.
More generally, beyond the "blue team dynamics", we also discuss benefits and biases, and promise and potential perils of social research using big data for disease nowcasting and forecasting.