Evaluation the Performance of Exponentially Weighted Moving Average in the Detection of Cholera Outbreaks: Using the Reported Cholera Outbreaks in Literature

Abstract

Introduction: Timely Detection of outbreaks of infectious diseases can have a very important role in surveillance systems. the presence of appropriate methods can have a very important role for this purpose, the aim of the current study was to Evaluation The Performance of Exponentially Weighted Moving Average in the detection of cholera outbreaks using the reported cholera outbreaks in literature

Methods: In the current study the EWMA method was evaluated. To assess the performance of the mentioned methods the six real outbreaks algorithm reported in the literature were used. These reported outbreaks were the daily counts of cholera cases in different countries. After insertion of each outbreak, 7 days inserted as nonoutbreaks days. All analyses performed by MedCalc18.11, Stata version15 and excel 2010.

Results: the sensitivity of EWMA was 56.4% (95% CI: 54.3%- 58.5%). The highest sensitivity for outbreak detection was seen in EWMA1 79.18(73.56-84.09) and the lowest was seen in EWMA4 12.2(8.4-17.0). EWMA2 with λ= 0.2 had the best performance with sensitivity 69.8 (63.6-75.5) and specificity 91.4(76.9-98.2) and AUC= 0.80.

nd AUC= 0.80. Conclusion: The EWMA method can be very useful in the detection of outbreaks, but the use of this method along the other models may increase the sensitivity of outbreaks detection.