Fig 1.
Map of Nigeria showing our study area.
The colored area indicates the study area, and the gray area shows all States not included in the data collection. Each dot corresponds to a market surveyed: 6 in Plateau, 2 in Bauchi, and 2 in Kano.
Fig 2.
Steps used to build the probable path of transmission.
Fig 3.
Visual representation of the reference network (configuration A) and the extracted backbone (reference backbone).
A: Reference network: 144 nodes and 268 links. B: Reference backbone using the sparsify method [11], which contains 20 nodes from different States. Each color corresponds to the State or the District (node) in our study area (Plateau, Bauchi and Kano).
Table 1.
Impact of changes in the configuration on the reference backbone (backbone extracted from the reference network).
Fig 4.
Results of PPR simulations using the SI model, showcasing the animal mobility network.
The average final size in simulated epidemics is presented under different configurations (from A to E2), and for each transmission probability (ρ = 0.01, 0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.5, 0.75). The final size fluctuates significantly under condition C3 (when adding 40% of links) and under condition D1 (deleting the most centrally connected node).
Fig 5.
Kaplan-Meier curves showing the time required to reach peak infection as a function of each transmission probability and configuration.
This model was applied to the SI model’s result, including all simulated epidemics. p represents the p-value and ρ the probability of transmission.
Fig 6.
Number of seeds in each seed cluster per State obtained with the k-means classification.
The number of seeds is shown on the x-axis and the name of the State is on the y-axis. Each line corresponds to the distribution in clusters (color) of seeds in the States.
Fig 7.
Tornado plot showing the variation in the number of sentinel nodes for each transmission probability.
The gain or loss of the number of nodes according to the reference (here, it’s ρ = 0.1) is shown on the x-axis, and the different scenarios used in the study (configurations) are displayed on the y-axis.
Table 2.
Number of sentinel nodes shared between configurations and the backbone, with State-wise specification.
CommonRef indicates the set of nodes in the reference backbone (i.e. configuration A) that are also sentinel ones. Sentinel nodes are counted independently of all transmission probabilities.
Fig 8.
A: Gini Index showing the most important centrality measure per probability of transmission of the observed network. B: Most important centrality measure per probability of transmission of the sentinel node on the backbone. The result was obtained from the random forest classification. Iin_deg = in-degree, in_H = in-H index, eigenvec = eigenvector, in_ngb = in-neighborhood, in_clo = in-closeness, bet = betweenness, trst = transitivity.