Real-time projections of epidemic transmission and estimation of vaccination impact during an Ebola virus disease outbreak in Northeastern Democratic Republic of Congo

Lee Worden; Rae Wannier; Nicole A. Hoff; Kamy Musene; Bernice Selo; Mathias Mossoko; Emile Okitolonda-Wemakoy; Jean Jacques Muyembe-Tamfum; George W. Rutherford; Thomas M. Lietman; Anne W. Rimoin; Travis C. Porco; J. Daniel Kelly

All transmission events were assumed to be independent. The interval between date of 107 detection of each primary case and that of each of its secondary cases is assumed 108 gamma distributed with mean 14.5 days and standard deviation 5 days, rounded to the 109 nearest whole number of days, as above. 110 We used the (R initial , Ï„ ) pairs estimated from prior outbreaks to provide R values for 111 simulation. R initial values were sampled uniformly from the range of values estimated 112 from past outbreaks. We applied a linear regression to the values of R initial and log(Ï„ ) 113 estimated for prior outbreaks and used the resulting regression line to assign a mean Ï„ to 114 each R, used with the residual variance of log(Ï„ ) as a distribution from which to sample 115 Ï„ values for simulation given R initial . Note that the range of fast and slow quenching 116 scenarios modeled in this way is not limited to the exact combinations estimated from 117 past outbreaks, but extends over a continuous distribution that includes those values. generated by the above branching process with the given parameters R initial , Ï„ , and k, 125 and these were then filtered by discarding all proposed trajectories except those whose 126 cumulative case counts matched known counts of the current 2018 EVD outbreak on 127 known dates. In earlier, smaller data sets we filtered against all reported case counts, 128 while in later, more complete data sets we used a thinned series of case counts for 129 filtering, for computational tractability, by selecting five case counts evenly spaced in 130 the data set plus the final case count (Fig S1 in S1 Supporting Information). The 131 filtration required an exact match of the first target value, and at subsequent target 132 dates accepted epidemics within a number of cases more or less than each recorded 133 value. On the earlier data sets in which the beginning dates of the epidemic were 134 unknown, the first target value was allowed to match on any day, and subsequent target 135 dates were assigned relative to that day.