Figure 1.
Model used for the simulation of disease spread.
The disease spreads as a circle at km/day. Once the disease has reached a municipality centroid, each uninfected herd in the municipality can become infected with a daily probability
. Once a herd is infected, each cow from the herd can become infected with a daily probability
. Squares on the left-hand side represent herds. On the right-hand side, black dots and red dots represent uninfected and infected cows respectively.
Figure 2.
Simulation of milk losses caused by the disease represented on the theoretical lactation curve of a Holstein cow in her second lactation.
Milk production drops by a proportion on the day of disease onset and for a period of
days. Milk production gets backs to its normal value linearly over a period of
days. Milk losses were simulated on monthly recorded milk yields.
Figure 3.
Mean number of cows recorded per municipality per week in the dataset used for simulation.
The 2 blue crosses represent the disease starting locations used in the simulations.
Table 1.
Parameters used for disease spread simulation.
Table 2.
Parameters used for the simulation of disease induced milk losses.
Figure 4.
Difference between observed and predicted mean milk production per cow per week between 2003 and 2006.
The years 2003 to 2005 were used for model fitting: the curves represent the mean of residuals per week. The year 2006 was used for disease simulation and detection.
Figure 5.
Association between emergence model parameters and the projection of each scenario on the 2 principal components identified in the principal component analysis.
Each dot represents a scenario and each color represents a parameter level.
Figure 6.
Number of detected clusters of mean cow yield deviations for weeks 5 to 52.
Figure 7.
Groups generated using k-means algorithm from the scenarios 2 principal components.
On the left-hand side plot, each dot represents a scenario and each color represents one of the 8 retained scenario groups. The plot on the right-hand side represents the mean quantity of milk lost per cow in each of 8 geographical area and during the first 8 weeks after emergence. The same colors are used on both sides.
Figure 8.
Comparison of the mean quantity of milk lost per cow in each of the 8 zones and during the first 8 weeks after emergence between the 8 selected scenarios and difference between milk production as observed and as predicted from the model results.
Figure 9.
Detection of emergence induced milk losses by SaTScan for group 6 for weeks 2 to 8 after emergence.
The scan statistic was performed on the simulated milk losses to which random noise was added.
Table 3.
Parameters used for the simulation of the 8 selected scenarios representing 8 disease groups.
Table 4.
Number of weeks between disease emergence and cluster detection for groups of scenarios 1 to 8 for various levels of random noise added to the simulated milk losses.
Table 5.
Number of weeks between disease emergence and cluster detection for groups of scenarios 1 to 8 for 2 disease starting locations and 2 starting dates.
Figure 10.
Number of detected clusters for weeks 5 to 52 using the difference between observed milk production and milk production predicted from the model.
Figure 11.
Location of the detected clusters for weeks 5 to 52 using the difference between observed milk production and milk production predicted from the model. The number above each map is the week analysed.