1. Locally Estimated Scatterplot Smoothing (LOESS) Across Texas: The Novel Application of an Accessible-Use Methodology to Detect Trends in COVID-19 Incidence Across Disparate Geographic Regions.
- Author
-
Blackledge, Sabrina, Webb, Christopher, Nunley, Karen, Archer, Natalie, and Pont, Stephen J.
- Subjects
- *
COVID-19 pandemic , *LOESS , *COVID-19 , *INFECTIOUS disease transmission , *COMMUNICABLE diseases - Abstract
The COVID-19 pandemic has presented unpreceded challenges to conventional surveillance systems, including high day-to-day reporting variability, timeliness, cost, manpower, and interpretability of large-volume data. A review of the advantages and limitations across common visual surveillance approaches to SARS-CoV-2/COVID-19 disease progression are discussed. A novel surveillance methodology using a two-month moving locally estimated scatterplot smoothing (LOESS) procedure, laid out in a small-multiple panel format, was successfully able to address many surveillance issues that occurred with onset of the COVID-19 pandemic. This surveillance system was used throughout the COVID-19 response by the state of Texas to detect daily incidence trends across disparate regions. The LOESS panel methodology can be applied to daily new cases or other variables of interest across multiple levels of aggregated regions for COVID-19 or other diseases. Use-case testing indicated the LOESS strikes a successful balance between daily variability versus early detection of true change in overall country and regional COVID-19 trends beyond that of a weekly moving average. In sum, the paneled LOESS smoothing methodology described herein presents a highly useful addition to other conventional methods of trend detection. This approach can be applied to COVID-19 and other rapidly spreading infectious diseases across multiple regions of interest and could be an additional highly accessible tool available for utilization by many state and regional jurisdictions. [ABSTRACT FROM AUTHOR]
- Published
- 2023