Feed

Objective: To determine the threshold for influenza epidemic intensity during the 2024-2025 season in Tianjin using the moving epidemic method (MEM). Methods: Collect weekly data on the percentage of influenza-like illness cases (ILI%), positive detection rate of influenza virus (PR), and PR×ILI% from sentinel hospitals in Tianjin for 2017-2025. Establish an influenza early warning model using MEM, select the optimal value of parameter δ, compare model fitting performance, evaluate the reliability of the three types of data, and determine the threshold for influenza epidemic intensity in Tianjin for 2024-2025. Model selection employs cross-validation, with evaluation metrics including sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, negative likelihood ratio, Matthews correlation coefficient, and Youden index. Results: The models based on PR and PR×ILI% showed better fitting effect than those based on ILI% alone, with the PR×ILI%-based model demonstrating the best performance. The MEM model constructed by using PR indicated that with δ set at 3.0, the epidemic start threshold for Tianjin in the 2024-2025 season was 16.05%, the end threshold was 19.68%, and the thresholds for medium, high, and very high epidemic intensity were 51.61%, 66.98%, and 75.16%, respectively. The model based on PR×ILI% showed that with δ set at 3.2, the epidemic start threshold was 105.23%, the end threshold was 128.08%, and the thresholds for medium, high, and very high epidemic intensity were 377.55%, 560.61%, and 667.64%, respectively. Conclusions: The moving epidemic method can be used to determine the influenza epidemic intensity thresholds in Tianjin. The influenza early warning model established based on PR×ILI% data better reflects changes in influenza activity intensity, providing a reliable basis for early warning.