Transmission pathway inference without genomic data

Before incorporating the use of genomic data in the inference methods of Pathogen.jl, we needed to see how well the standard individual level model (ILM) of disease transmission performed at the task of transmission pathway inference. A small extension to ILMs was required where the overall disease pressure on an individual was separated into the disease pressure contributions of each potential infection source. It was expected that without genomic information, transmission pathway inference would not be particularly accurate. Here’s a peak:

SEIR_animation_combined
The upper graphic shows the actual transmission network from a short epidemic simulation on a small population. The transmission network and disease state timing in the bottom graphic has been inferred based on only information about when individuals were observed to be symptomatic, and when they were observed to have recovered (observed with random lag). Specifically, a maximum aposteriori transmission network has been shown. In general we see a slight underestimation of the importance of internal spatial disease transfer, and a slight overestimation of outside/external disease pressure.

 

Leave a Reply