Skip to main content

Table 1 Outbreak detection simulation results.

From: Syndromic surveillance: STL for modeling, visualizing, and monitoring disease counts

   Sensitivity
Baseline Magnitude C1 C2 C3 GLM STL STL(90)
1 1.0 0.57 (5.44) 0.50 (5.50) 0.46 (5.48) 0.60 (6.58) 0.71 (6.44) 0.66 (6.65)
1 1.5 0.57 (5.42) 0.55 (5.65) 0.55 (6.05) 0.70 (6.80) 0.73 (6.54) 0.83 (6.57)
1 2.0 0.69 (5.39) 0.70 (5.64) 0.72 (6.13) 0.88 (6.24) 0.89 (6.22) 0.93 (6.07)
2 1.0 0.48 (5.43) 0.56 (5.90) 0.56 (5.89) 0.65 (6.01) 0.80 (6.31) 0.82 (6.90)
2 1.5 0.57 (6.03) 0.72 (6.68) 0.75 (6.81) 0.81 (6.72) 0.90 (6.63) 0.89 (5.96)
2 2.0 0.68 (5.11) 0.81 (5.39) 0.84 (5.53) 0.91 (5.36) 0.95 (5.19) 0.91 (5.93)
3 1.0 0.59 (5.51) 0.60 (5.74) 0.62 (5.80) 0.58 (6.20) 0.73 (6.62) 0.70 (6.74)
3 1.5 0.68 (6.01) 0.71 (6.37) 0.74 (6.49) 0.76 (7.00) 0.83 (6.82) 0.81 (5.93)
3 2.0 0.71 (5.47) 0.76 (5.82) 0.80 (5.87) 0.84 (6.21) 0.87 (6.00) 0.87 (5.70)
  1. Outbreak detection simulation results for three baselines and three outbreak magnitudes.
  2. Sensitivity is reported with mean days until detection in parentheses. The false positive rate was set empirically for each method and baseline to be 0.03. The superiority of the sensitivity of the STL method for these scenarios is evident, especially at the smallest outbreak magnitude. The results in the final column, STL(90), are from allowing only a 90 day historical baseline for each outbreak scenario.