Decoding influenza outbreaks in a rural region of the USA with archetypal analysis. (August 2021)

• Record Type: Journal Article
• Authors: Mokhtari, Elham Bayat; Landguth, Erin L; Anderson, Stacey; Stone, Emily
• Is Part Of: Spatial and spatio-temporal epidemiology. Volume 38(2021)
• Publication Date: 2021-08
• Subjects: Archetypal Analysis -- Seasonal Flu -- Spatial-temporal infectious disease spread; Epidemiology -- Statistical methods -- Periodicals; Epidemiology -- Periodicals; 614.4072
• DOI: 10.1016/j.sste.2021.100437
• Languages: English
• ISSNs: 1877-5845

Abstract

Highlights: We have developed a procedure for exploring epidemiological data using archetypal analysis. Our methodology decomposes epidemiological data into spatial patterns and time series representing their contribution to the flu count signal. Using Montana seasonal influenza data we discover regions of the state and aggregations of counties that experience simultaneous outbreaks. The spread of influenza across the state can be typified by a transfer across these spatial patterns. Abstract: We present the first application of archetypal analysis for influenza data from 2010 to 2018 in Montana, USA. Using archetypes, we decompose the data into spatial and temporal components, allowing for a more informed analysis of spatial-temporal dynamic trends during an influenza season. Initially, we reduce the dimension of the set of counties by using a mutual information measure on the influenza time series to create a smaller, maximal mutual information network. Archetypal analysis then describes the relationship between influenza cases across counties and regions in Montana. Finally, we discuss the potential implications this analysis can have for infectious disease modeling, particularly where data is sparse and limited.