ABSTRACT

The contact interval in an ordered pair (i, j) is the time from the onset of infectiousness in i until infectious contact from i to j, where infectious contact is defined to be a contact sufficient to infect j if j is susceptible. It is shown that the contact interval can be used to calculate transmission probabilities, infectivity curves, and the basic reproductive number in network-based and mass-action stochastic epidemic models. When who-infects-whom is observed, parametric and nonparametric methods from survival analysis can be used to estimate the contact interval distribution in pairs of individuals. When who-infects-whom is not observed, an Expectation-Maximization algorithm can be used to account for all possible combinations of who-infected-whom. These methods can be extended to parametric and semiparametric regression models that allow simultaneous estimation of covariate effects on infectiousness and susceptibility.