Quantifying transmission of emerging zoonoses: Using mathematical models to maximize the value of surveillance data

Monique R. Ambrose; Adam J. Kucharski; Pierre Formenty; Jean-Jacques Muyembe-Tamfum; Anne W. Rimoin; James O. Lloyd-Smith

When comparing the overall fraction of cases attributed to each source type (zoonotic 311 versus within-locality versus between-locality transmission), the sampled transmission trees 312 closely match the true transmission patterns (Fig 4) . On average, the difference between the true 313 fraction of cases caused by zoonotic spillover and the fraction inferred in a tree was 0.022 314 (standard deviation 0.018), the difference for within-locality transmission was 0.006 (standard 315 deviation 0.005), and the difference for between-locality transmission was 0.022 (standard 316 deviation 0.018). values. If the fraction of transmissions for each source is perfectly inferred, points will lie exactly 324 on the transition between bar colors. B. Box plots summarize the error in the inferred fraction of 325 cases originating from each source type. The error size is small across all parameter sets, 326 especially for within-locality human-to-human transmission. The upper whisker was calculated 327 as min(max(x), Q 3 +1.5*IQR) and the lower whisker was calculated as max(min(x),Q 1 -1.5*IQR). 328