COVID-19 [electronic resource] : Spatiotemporal social data analytics and machine learning for pandemic exploration and forecasting

• Ngôn ngữ: eng
• Ký hiệu phân loại: 614 Forensic medicine; incidence of injuries, wounds, disease; public preventive medicine
• Thông tin xuất bản: Washington, D.C. : Oak Ridge, Tenn. : United States. National Nuclear Security Administration ; Distributed by the Office of Scientific and Technical Information, U.S. Dept. of Energy, 2021
• Mô tả vật lý: Size: 9 p. : , digital, PDF file.
• Bộ sưu tập: Metadata
• ID: 261710
• Chủ đề: 1. 60 applied life sciences; 2. 97 mathematics and computing

Tóm tắt

This task focused on developing a preliminary approach to use machine learning (ML) to explore the relationship between county-level societal variables and COVID-19 parameters, including COVID-19 cases rates and counts and COVID-19 death rates and counts. The objective was to develop and test a prototype approach for linking COVID-19 and county-level data. The task focused on enhancing and applying existing LANL ML techniques to COVID-19. Our novel ML methods have been a subject of a recently approved U.S. patent. The codes based on these methods are already open-source released. Our ML tools (NMFk/NTFk) are applied to extract hidden features (signals, waves) in the analyzed datasets and automatically identify their optimal number. The features are extracted by identifying counties that have similarities between the county-level societal variables and the COVID-19 parameters. These demonstration analyses will facilitate the ongoing pandemic simulations and predictions performed by Los Alamos other institutions, as well as lay the groundwork for future work.