On Improving Toll Accuracy for COVID-like Epidemics in Underserved Communities Using User-generated Data

This paper envisions using user-generated data as a cheap way to improve accuracy of epidemic tolls in underserved communities. The global widespread of COVID-19 pandemic has imposed several unprecedented challenges. One of these challenges is constantly monitoring the unprecedented epidemic widespread at a fine-granular spatial scale, so experts can model, understand, and prevent disease transmission and field personnel can reach and treat infected people. Unfortunately, the limited resources compared to the pandemic widespread has led to a significant number of unreported cases in underserved communities and developing countries, including a large number of severe cases. We propose in this paper enhancing epidemic case reporting in underserved communities through exploiting the power of data that are posted by people on web. Our vision is building a data analysis pipeline that filters and categories use-generated data objects to provide informal estimates for tolls in unreachable regions and enhance estimates in other regions. The pipeline consist of five stages, that starts with filtering epidemic-specific data to visualize advanced aggregates to end users. We also discuss several technical challenges that face different stages of the pipeline.