Evaluation of an analytic, approximate formula for the time-varying SIS prevalence in different networks

One of the most important quantities of the exact Markovian SIS epidemic process is the time-dependent prevalence, which is the average fraction of infected nodes. Unfortunately, the Markovian SIS epidemic model features an exponentially increasing computational complexity with growing network size N. In this paper, we evaluate a recently proposed analytic approximate prevalence function introduced in Van Mieghem (2016). We compare the approximate function with the N-Intertwined Mean-Field Approximation (NIMFA) and with simulation of the Markovian SIS epidemic process. The results show that the new analytic prevalence function is comparable with other approximate methods.