Fear, Access, and the Real-Time Estimation of Etiological Parameters for Outbreaks of Novel Pathogens

Nina H Fefferman; Eric Lofgren; Nianpeng Li; Pieter Blue; David Weber; Abdul-Aziz Yakubu

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020. 03.19.20038729 doi: medRxiv preprint Assuming that, at least initially, our surveillance cannot determine whether an uninfected 162 person is susceptible or recovered, and therefore 1 ) surveillance is perfect, then the measured case incidence will be equal to the 167 corresponding real case incidence, as we would hope). 168 169 Using this definition, we then correct our understanding of any disease incidence curve 170 once we have either measured or assumed appropriate functions/values for α , δ , λ , 1 ε , 171 and 2 ε . While this might at first seem straightforward, there arises the complication that 172 our health care seeking behavior functions are likely to be problematic in at least three 173 separate ways: (1) they are likely to be functions of the current perceived prevalence of 174 infection in the population (i.e. some function of m i ), (2) they are likely to be functions of 175 time since the beginning of the perception of the current outbreak, (3) they are likely to 176 be non-linear and, in some cases, not even continuous. We therefore propose the 177 following algorithm to produce a system of SIR curves which reflect the underlying 178 disease dynamics without the influence of behavioral shifts and/or testing inaccuracy; we 179 will denote this system as "Testing Neutral", TN. 180