<![CDATA[Background: Good decisions in policy-making rely on acquiring the best possible understanding at the fast pace of what is happening and what might happen next in the population. Immediate measurements and predictions of disease spread would help authorities take necessary action to mitigate the rapid geographical spread of potential emerging infectious diseases. Unfortunately, measuring human behavior in nearly real-time, specifically at a large scale, has been labor-intensive, time-consuming, and expensive. Consequently, measurements are often unfeasible or delayed in developing in-time policy decisions. The increasing use of online services such as Twitter generates vast volumes and varieties of data, often available at high speed. These datasets might provide the opportunity to obtain immediate measurements of human behavior. Here we describe how the patterns of population mobility can be associated with the number of COVID-19 cases and, subsequently, could be used to simulate the potential path of disease spreading.Methods: Our analysis of country-scale population mobility networks is based on a proxy network from geotagged Twitter data, which we incorporated into a model to reproduce the spatial spread of the early phase COVID-19 pandemic in Indonesia. We used aggregated province-level mobility data from January through December 2019 for the baseline mobility patterns from DKI Jakarta as the origin of the 33 provinces' destinations in Indonesia.Result: We found that population mobility patterns explain 62 percent of the variation in the occurrence of COVID-19 cases in the early phases of the pandemic. In addition, we confirm that online services have the potential to measure human behavior in nearly real time.Conclusion: We believe that our work contributes to previous research by developing a scalable early warning system for public health decision-makers in charge of developing mitigation policies for the potential spread of emerging infectious diseases.]]>
