Figure 4
Effect of background case proportion (in range 0–0.9) on analysis of 300 simulated outbreaks with respect to (a) average correct identification (ID) rate top 3 rate, and top 10 rate with standard errors. (b) Proportional composition of each possible analysis outcome, and (c) boxplot of the number of cases analyzed by the model before returning the labeled outcomes. The upper and lower boundaries of the boxes indicate the group 75th (Q3) and 25th (Q1) percentile respectively, the black line within the boxes marks the group median, and whiskers extend to the least extreme data point within ± 1.5 *Interquartile range (Q3–Q1). Y-axis on log scale. Outbreak analysis was initialized at 5 cases and terminated if the number of cases exceeds 200 due to computational constraints.
