Report of ABC for Irish hidden burden model

1 Method

1.1 Population

Sample of 2500 from breakdowns fitting selection criteria ending 2015 - 2018, excluding traders and herd size > 600.

1.2 Model

Allows density dependence. Includes GIF testing. Allows hurdle (reduced sensitivity) for first case of every test. (model14.cpp)

2 Posterior predictions

• Grey = data
• Yellow = model. Each dot represents one simulation of the model generating the summary statistic in the plot. The sets of parameters for each of the 100 simulation came from a weighted sample (according to proximity of model output to the data and difference to previous particles) of 100 particles from the ultimate round of the ABC.

2.1 Count skin cases

Figure 2.1: Skin case counts per breakdown. Max skin cases in data is 120. Plot is censored at 25 for clarity.

2.2 Count skin cases stratified by herd size category

Figure 2.2: Skin case counts per breakdown. Max skin cases in data is 120. Plot is censored at 25 for clarity.

2.3 Count of cases in first test of breakdown

Figure 2.3: Skin case counts from the first test of the breakdown. Max first test skin cases in data is 120. Plot is censored at 25 for clarity.

2.4 Count of cases in first test of breakdown stratified by herd size category

Figure 2.4: Skin case counts from the first test of the breakdown. Max first test skin cases in data is 120. Plot is censored at 25 for clarity.

2.5 Count of GIF positive cattle in GIF tested breakdowns

Figure 2.5: GIF case counts in GIF tested breakdowns. Max GIF reactors per breakdown in data was 81. Plot censored at 25 for clarity.

2.6 Count of GIF positive cattle in GIF tested breakdowns stratified by herd size category

Figure 2.6: GIF case counts in GIF tested breakdowns.Max GIF reactors per breakdown in data was 81. Plot censored at 25 for clarity.

2.7 Breakdown duration

Figure 2.7: Breakdown duration. Max breakdown duration in data was 1293 days. Plot cenosred for clarity.

2.8 Breakdowns with skin test reactors after the initial test of breakdown

Figure 2.8: Proportion of breakdowns with skin cases after the first test.

2.9 Recurrence

Figure 2.9: Recurrence within 1, 2 and 3 years of breakdown end.

2.10 First breakdowns initiated by a slaughterhouse case

Figure 2.10: First breakdowns initiated by slaughterhouse cases

2.11 Recurrent breakdowns initiated by a slaughterhouse case

Figure 2.11: Recurrent breakdowns initiated by slaughterhouse cases.

2.12 GIF tested breakdowns

In the model, a breakdown had to have four or more skin cases to be considered for GIF testing. Once this criterion was met, the decision to GIF test or not in the model was based on regression coefficients from an external generalised linear model explaining GIF testing with county, year end breakdown, count skin test reactors in triggering skin test, herd management and herd size. The GIF test metric here is a loose metric of breakdowns with four or more cases, and of the GIF testing process in the model versus reality.

2.13 GIF tested breakdowns

Figure 2.12: GIF tested breakdowns.

2.14 Prolonged breakdowns

Figure 2.13: Breakdowns prolonged beyond 240 days.

2.15 Breakdowns with greater than 1 case

Figure 2.14: Breakdowns with greater than one case

2.16 Mean count skin cases

The standard deviation of skin case count was much larger for the larger herd categories compared to the others. This created very wide confidence intervals for the largest herd category, which, for clarity, made me decide not to add to plots below. The case count bar plots, although they are also censored and exclude very large case counts, above are an alternative way to review the fit to this target metric. Case counts have a very long tail (right skew) in the largest herd category.

2.17 Mean count skin cases

Figure 2.15: Mean skin case count

2.18 Mean count skin cases in first test of breakdown

## Coordinate system already present. Adding new coordinate system, which will
## replace the existing one.

Figure 2.16: Mean skin case count for the first test of the breakdown.

2.19 Mean breakdown duration

## Coordinate system already present. Adding new coordinate system, which will
## replace the existing one.

Figure 2.17: Mean breakdown duration.

2.20 Mean GIF reactors in GIF tested breakdowns

Figure 2.18: Mean GIF reactors in GIF tested breakdowns .