Files
Abstract
The epidemiology of seasonal influenza is shaped by mechanisms across ecological scales, from molecular interactions to global climate patterns. Misaligned data may greatly impact analytical inference, but spatial constructs characterizing larger scales, e.g., regions, lack concrete, standard definitions and, consequently, are often overlooked in influenza research.
In this dissertation, I analyze patterns in human mobility, disease incidence, and viral genetic evolution to holistically characterize spatial structuring within the United States related to seasonal influenza. In Chapter 2, I model commuting flows and influenza-like illness (ILI). Using an estimated critical distance of ~150km or ~93mi, I show that simple summary metrics of local mobility from county-level commutes informs some variation in state-level ILI epidemic intensity. In Chapter 3, I evaluate numerous regional delineations of the US for their ability to capture important patterns of worker commutes, ILI incidence, and viral population structure. From this network science community analysis, I find evidence suggesting that the US may be best represented with ~8 subnational regions which are not precisely captured by existing administrative regional delineations. In Chapter 4, I systematically describe local outbreaks of four seasonal influenza viruses across a decade of flu seasons in the US. I show that the average isolate diversities of local outbreaks exhibit weak spatial autocorrelation, and marginally, local outbreaks in more populous states tended to have less diverse viral isolates which may suggest either impactful differences in transmission patterns or isolate sampling.
Taken together, these analyses suggest that there is inherent structuring of local and regional scales within the US. Given these findings, I speculate that much of the observed variation in seasonal influenza epidemiology at the regional level could be explained by the underlying spatial organization of local populations. Additionally, this work shows that even with simple methodologies and crude conceptualizations of scale, we can abstract information from data at higher resolutions which is salient to patterns at larger scales and coarser resolutions. With continued effort, we may be able to identify systematic sources of variation in outbreak dynamics and viral evolution which would be invaluable when modeling an otherwise largely chaotic infectious disease system.
In this dissertation, I analyze patterns in human mobility, disease incidence, and viral genetic evolution to holistically characterize spatial structuring within the United States related to seasonal influenza. In Chapter 2, I model commuting flows and influenza-like illness (ILI). Using an estimated critical distance of ~150km or ~93mi, I show that simple summary metrics of local mobility from county-level commutes informs some variation in state-level ILI epidemic intensity. In Chapter 3, I evaluate numerous regional delineations of the US for their ability to capture important patterns of worker commutes, ILI incidence, and viral population structure. From this network science community analysis, I find evidence suggesting that the US may be best represented with ~8 subnational regions which are not precisely captured by existing administrative regional delineations. In Chapter 4, I systematically describe local outbreaks of four seasonal influenza viruses across a decade of flu seasons in the US. I show that the average isolate diversities of local outbreaks exhibit weak spatial autocorrelation, and marginally, local outbreaks in more populous states tended to have less diverse viral isolates which may suggest either impactful differences in transmission patterns or isolate sampling.
Taken together, these analyses suggest that there is inherent structuring of local and regional scales within the US. Given these findings, I speculate that much of the observed variation in seasonal influenza epidemiology at the regional level could be explained by the underlying spatial organization of local populations. Additionally, this work shows that even with simple methodologies and crude conceptualizations of scale, we can abstract information from data at higher resolutions which is salient to patterns at larger scales and coarser resolutions. With continued effort, we may be able to identify systematic sources of variation in outbreak dynamics and viral evolution which would be invaluable when modeling an otherwise largely chaotic infectious disease system.