Mojtaba Sepandi; Yousef Alimohamadi; Mousa Imani
Abstract
Background: Outbreak detection algorithms could play a key role in public health surveillance.Objectives: This study aimed to compare the performance of three algorithms (EWMA, Cumulative Sum (CUSUM), and Poisson Regression) using the reported COVID-19 data for outbreak detection.Methods: Three outbreak ...
Read More
Background: Outbreak detection algorithms could play a key role in public health surveillance.Objectives: This study aimed to compare the performance of three algorithms (EWMA, Cumulative Sum (CUSUM), and Poisson Regression) using the reported COVID-19 data for outbreak detection.Methods: Three outbreak detection algorithms were applied to the data of COVID-19 daily new cases in Iran between 19/02/2020 and 20/06/2022, and 344 simulated outbreak days were injected into the data sequences. The Area Under the Receiver Operating Characteristics (ROC) Curve (AUC) and its 95% confidence intervals (95% CI) were also computed.Results: EWMA9 had the lowest AUC (51%). Among the different algorithms, EWMA9 with λ = 0.9 and CUSUM 1 had the highest sensitivity with 100 and 87% (95% CI: 84%-91%), respectively.Conclusion: According to the results, CUSUM, EWMA, and poison regression showed appropriate performance in detecting the COVID-19 outbreaks. These algorithms can be extremely helpful for health practitioners and policymakers in the detection of infectious disease outbreaks.