Skip to main content

Table 1 Summary of simulation results

From: Detection of influenza-like illness aberrations by directly monitoring Pearson residuals of fitted negative binomial regression models

Signal to noise

Evaluation measure

Methods

SPR with thresholdz1 − 0.025

SPR with thresholdz1 − 0.005

CUSUM with pearson residuals

CUSUM with observations

θ = 5

Mean days to detection

0.5

0.8

1.9

6.1

Non-detection probability

0.0

0.0

0.013

0.082

Mean false alarm rate

3.4%

1.0%

5.2%

5.2%

θ = 3

Mean days to detection

1.3

2.7

4.7

11.7

Non-detection probability

0.0

0.0

0.105

0.158

Mean false alarm rate

3.6%

1.1%

5.2%

5.2%

θ = 1

Mean days to detection

6.6

12.6

11.9

19.2

Non-detection probability

0.004

0.120

0.373

0.178

Mean false alarm rate

3.9%

1.2%

5.1%

5.3%