OBJECTIVE: To understand the epidemiological trend on the number of influenza-like cases and to explore the feasibility of early warning systems of influenza in Gansu province. METHODS: Based on data from the influenza sentinel surveillance program, a sequence chart was used to analyze the epidemiological trend on the number of influenza-like illness (ILI) cases. Both control chart and mobile percentile method were used to select the threshold of premium alert for the ILI of sentinel surveillance program. Warning effects were assessed by statistical model. RESULTS: The prevalence of influenza were both low in 2007 and 2008. Alert thresholds for ILI of Sentinel surveillance was built. The thresholds were higher alert in winter, but lower in summer. Both Seasonal Exponential Smoothing Model and Multiplicative Seasonal ARMA Model (1, 1, 1) (0, 1, 0) were used to dynamically predict the weekly percentage of outpatient visits for influenza-like illness (ILI%) of 2011. The concordance rates (predicted = actual) were 100% for both of them. According to the RMSE values, the dynamically predicted effect of the seasonal exponential smoothing model was superior to ARIMA. CONCLUSION: Dynamic prediction on the number of influenza-like cases could reflect the epidemiological trend of influenza in Gansu province, but with some limitations.
OBJECTIVE: To understand the epidemiological trend on the number of influenza-like cases and to explore the feasibility of early warning systems of influenza in Gansu province. METHODS: Based on data from the influenza sentinel surveillance program, a sequence chart was used to analyze the epidemiological trend on the number of influenza-like illness (ILI) cases. Both control chart and mobile percentile method were used to select the threshold of premium alert for the ILI of sentinel surveillance program. Warning effects were assessed by statistical model. RESULTS: The prevalence of influenza were both low in 2007 and 2008. Alert thresholds for ILI of Sentinel surveillance was built. The thresholds were higher alert in winter, but lower in summer. Both Seasonal Exponential Smoothing Model and Multiplicative Seasonal ARMA Model (1, 1, 1) (0, 1, 0) were used to dynamically predict the weekly percentage of outpatient visits for influenza-like illness (ILI%) of 2011. The concordance rates (predicted = actual) were 100% for both of them. According to the RMSE values, the dynamically predicted effect of the seasonal exponential smoothing model was superior to ARIMA. CONCLUSION: Dynamic prediction on the number of influenza-like cases could reflect the epidemiological trend of influenza in Gansu province, but with some limitations.