Establishment and application of SEIAR model

ObjectiveTo establish a SEIAR model by using software Berkeley Madonna and analyze the modeling steps and code.

MethodsData on the onset time of an influenza outbreak in a city were collected. The Berkeley Madonna software was used to establish a non-intervention model (SEIAR0), a model for case isolation (SEIAR_I), a model for class suspension (SEIAR_C), and a model for comprehensive intervention (SEIAR_M). The cumulative attack rate (TAR) and duration of epidemics (DO) were used to compare the impact of interventions on outbreaks. The parameter β of the model was obtained by curve fitting, and other parameters were based on the same kind of research and sensitivity analysis on the stability of the result.

ResultsThe influenza epidemic occurred on November 12, 2017, and the control measures for case isolation and class suspension were taken on November 17 and November 21 respectively. The optimal value of β was 0.00105, the TAR of the SEIAR0 model was 85.90%, and the DO was 49 days. SEIAR_M, SEIAR_I, and SEIAR_C decreased by 95.25%, 89.94%, and 73.32%, respectively, compared with SEIAR0. DO was 29 days, 78 days, and 39 days, respectively. The sensitivity analysis showed that the TAR was between 71.42% and 86.02% when the parameters changed.

ConclusionThe SEIAR model established with Berkeley Madonna software is simple and easy to use, which is suitable for the evaluation of intervention effects.