Modeling influenza progression within a continuous-attribute heterogeneous population

We consider three attributes of an individual that are critical in determining the temporal dynamics of pandemic influenza: social activity, proneness to infection, and proneness to shed virus and spread infection. These attributes differ by individual, resulting in a heterogeneous population. We develop discrete-time models that depict the evolution of the disease in the presence of such population heterogeneity. For every individual, the value for each of the three describing attributes is viewed as an experimental value of a continuous random variable. The methodology is simple yet general, extending more traditional discrete compartmental models that depict population heterogeneity. Illustrative numerical examples show how individuals who have much larger-than-average values for one or more of the attributes drive the influenza wave, especially in the early generations of the pandemic. This heterogeneity-driven pandemic physics carries important policy implications. We conclude by using contact data in four European countries to demonstrate empirical uses of our model.

► We introduce a generalization of a discrete-time heterogeneous influenza spread model.
► Our model creates a continuum of characteristic “classes” in the population.
► As an outbreak progresses, the susceptible population becomes “low-activity” heavy.
► At the beginning of a flu outbreak, the infectious people are likely to be “high-activity”.
► Intervention should target people at the left tails of the activity-level distribution.