Accurate influenza forecasts using type-specific incidence data for small geographic units

Abstract

Influenza incidence forecasting is used to facilitate better health system planning and could potentially be used to allow at-risk individuals to modify their behavior during a severe seasonal influenza epidemic or a novel respiratory pandemic. For example, the US Centers for Disease Control and Prevention (CDC) runs an annual competition to forecast influenza-like illness (ILI) at the regional and national levels in the US, based on a standard discretized incidence scale. Here, we use a suite of forecasting models to analyze type-specific incidence at the smaller spatial scale of clusters of nearby counties. We used data from point-of-care (POC) diagnostic machines over three seasons, in 10 clusters, capturing: 57 counties; 1,061,891 total specimens; and 173,909 specimens positive for Influenza A. Total specimens were closely correlated with comparable CDC ILI data. Mechanistic models were substantially more accurate when forecasting influenza A positive POC data than total specimen POC data, especially at longer lead times. Also, models that fit subpopulations of the cluster (individual counties) separately were better able to forecast clusters than were models that directly fit to aggregated cluster data. Public health authorities may wish to consider developing forecasting pipelines for type-specific POC data in addition to ILI data. Simple mechanistic models will likely improve forecast accuracy when applied at small spatial scales to pathogen-specific data before being scaled to larger geographical units and broader syndromic data. Highly local forecasts may enable new public health messaging to encourage at-risk individuals to temporarily reduce their social mixing during seasonal peaks and guide public health intervention policy during potentially severe novel influenza pandemics.

Consistent pattern of epidemic slowing across many geographies led to longer, flatter initial waves of the COVID-19 pandemic

“…In previous studies on influenza and influenza-like-illness [ 28 – 30 ] we used a smoothly varying two-value functional form to describe the time-dependent reproduction number: R ( t ) = β ( t ) γ , where β ( t ) is the time-dependent transmission rate and γ is the total recovery rate. Here we extended this model to an arbitrary number of values: …”

“…Our model can also be made more flexible by including differences in quality of healthcare over time and location [48,49], and allowing for coupling between geographic regions [29]. Our previous work on forecasting ILI in the U.S. [28][29][30] highlights the important role that spatial coupling can play in respiratory disease transmission, and which we anticipate will become increasingly more important as travel restrictions have been relaxed and movement between states, countries, and continents significantly increases.…”

“…For the quantitative analysis, we developed an age-stratified compartmental model [26] that fits a flexible, smoothly varying reproduction number and used it to study the epidemic, from January to October of 2020, in 49 jurisdictions in the contiguous U.S. and 89 locations globally. Our model allows for multiple values of R(t) and is an extension of our previous work which used a smoothly varying twovalue functional form [28][29][30]. Although population disease profiles differ significantly temporally, we explored the idea that there may be similarities if studied from a common start time.…”

Comparing trained and untrained probabilistic ensemble forecasts of COVID-19 cases and deaths in the United States

“…To cite just a few examples, methods used include multiscale probabilistic Bayesian random walk models (Osthus & Moran, 2021), Gaussian processes (Johnson et al, 2018), kernel conditional density estimation (Ray et al, 2017;Brooks et al, 2018), and generalized additive models (Lauer et al, 2018). Other models have an implicit or explicit representation of a disease transmission process, such as variations on the susceptible-infectious-recovered (SIR) model (Shaman & Karspeck, 2012;Lega & Brown, 2016;Osthus et al, 2017;Pei et al, 2018;Turtle et al, 2021).…”