Influenza predictions using county-level absences
We evaluated negative binomial models of seasonal variables (i.e.,
calendar week, average weekly temperature, and relative humidity) alone,
and including weekly all-cause county-level school absences at one-,
two-, and three-week lags. One- and third-week lagged absences had
similar model performance (Supplemental Table 2), therefore, we used
one-week lagged absences in all models to better reflect influenza’s
infectious period (i.e. one-week spread)(25). Compared to seasonal
models, AICs of in-sample models including calendar week, average weekly
temperature, average weekly relative humidity, and one-week lagged
weekly county-level all-cause absences either stayed the same or
slightly worsened (∆ AICc=2, 1, and 0, Table 1), whereas models
of calendar week, average weekly temperature, and one-week lagged weekly
absences had slightly improved fits (∆ AICc=-4, -4, -4, Table
1). For prediction performance, MAEs either stayed the same or decreased
when including one-week lagged weekly absences in models of calendar
week, average weekly temperature and relative humidity relative to
seasonal-only models (relMAE=0.95, 1.0, & 0.95, Table 1).
For individual influenza seasons, weekly-lagged country-level absence
multivariate models predicted atypical seasons poorly, but predicted
more typical seasons (i.e., 2013-2014, 2014-2015) with relatively high
accuracy (R2 of 0.91 and 0.57) (Figure 2A). Predicted
seasonal peaks were earlier and over-predicted during low transmission
seasons (i.e., 2010-2011 and 2011-2012), whereas during high
transmission seasons (2014-2015) had later predicted peaks, but of equal
magnitude (Figure 2A & 2C). Compared to seasonal models, predicted
cases from all-cause absence models varied (either increased or
decreased) over the five seasons (Figure 2B), with seasonal peak timing
varying most (Figure 2B). Calendar week, average weekly temperature, and
absence models varied the most across seasons (Figure 2B). The model
containing all seasonal variables and weekly absences had the smallest
changes in predicted cases. Lowest MAE models depended on the withheld
validation season (Supplemental Table 5). Given the consistently low
MAEs of the model including calendar week, average weekly temperature,
average weekly relative humidity and school absence, we present results
from this model.