RAPID: III: Data Collection and Risk Evaluation Learning in Identifying High Risk Ebola Subpopulations for the Intervention and Prevention of Large-scale Ebola Virus Spreading

University Of Kansas Center For Research Inc, Lawrence KS

Investigators

Abstract

The 2014 Ebola epidemic is the largest in history, affecting multiple countries in West Africa, and now impacting the US and other countries worldwide. The US Center for Disease Control and Prevention (CDC) and partners are taking precautions to prevent the further spread of Ebola within the United States. There is a lack of public understanding of the risks associated with Ebola; witness the inconsistently applied local responses (such as quarantines) that do not match CDC recommendations. This project will develop technology to enable individuals to evaluate risks associated with their own past and planned activities and travel. This will both enable those at risk to take appropriate action, and reduce unwarranted demand on the healthcare system by reassuring those whose activities have not placed them at risk. This project will use data gathered from the CDC and other public sources to develop risk models, and develop a mobile app that will use this data along with the user's own location and activity history and plans to report individual risk to the user. An individual's data never leaves their own device, ensuring personal privacy. The resulting lessons learned will ease the process of developing similar individualized risk assessment tools for future epidemics, providing long-term benefits beyond the Ebola virus epidemic. The research will address three main issues. The first is focused crawling of structured (CDC Contact Tracing reports) and unstructured (social media, web blogs) information on time, location, and activities of Ebola patients. A second research challenge is patient activity modeling: Given the returned information, developing a time/space/activity model determining the risk of the patient acting as a transmission agent. Finally, the project will develop a mobile app that tracks time, location, and activities of the mobile device user, and retrieves the patient activity models developed from public data to determine if the user is at risk of infection. This is a complex problem, as the data may be non-specific and require inferential techniques to estimate risk (e.g., being in the same time/location as a transmission agent poses very different risk if the location is a sports stadium as opposed to a restaurant); the project will develop ontologies for activities to use in estimating risk. The project will use expert opinion to seed regression models for risk assessment. Lessons learned from this project will also identify challenges for future research in information integration, risk analysis, machine learning, and privacy preserving technologies.