Asosiy kontentga oʻtish
proceeding

Epidemic Outbreak Behavior Simulation Using SIR Models in Public Health Emergency Planning

Swarna VenkateshCMR Technical CampusCharos MadievaUniversity of Tashkent for Applied SciencesDhinakaran JSt.Joseph's Institute of Technology, OMRDedakhanov Abdumalik MutalliyevichTuran International UniversityDebarghya BiswasKalinga University
2026en
Ochiq kirishOltin

Litsenziya: cc-by

ABI

Annotatsiya

The planning for public health emergencies related to the outbreaks of new respiratory pathogens requires forecasting tools that work under the main structural feature that differentiates epidemic dynamics from standard time series: early growth with strong exponential characteristics, followed by a peak and decline that is driven mostly by saturation, which is the pattern that does not allow the classic statistical techniques generally used for extrapolation to achieve satisfactory results. The Susceptible-Exposed-Infected-Recovered (SEIR) model works along the lines of Kermack and McKendrick's model based on three groups and adds the group for incubation period. This paper calibrates and validates a region-specific SEIR framework across six administrative regions with a combined population of 31.2 million (Region A: 8.4M; B: 6.8M; C: 5.2M; D: 4.4M; E: 3.6M; F: 2.8M), using 120 days of historical surveillance data for parameter calibration (transmission rate$\boldsymbol{\beta}$, incubation rate$\boldsymbol{\sigma}=\mathbf{1} / \mathbf{5. 2}$days, recovery rate$\boldsymbol{\gamma}=\mathbf{1} / \mathbf{6. 8}$days) and a held-out 60-day validation window (days 121-180). The basic reproduction number,$\mathrm{R}_{0}$, varies across regions from 1.44 (Region F) to 2.84 (Region A) due to the varying population density and contact patterns among the regions. SEIR shows a significant improvement over linear extrapolation (RMSE$=\mathbf{0. 2 4 8} \boldsymbol{\%}$, error is$\mathbf{6 6. 1 \%}$higher), exponential growth fitting (RMSE$=\mathbf{0. 1 8 4 \%}$, error is$\mathbf{5 4. 3 \%}$higher), and an ARIMA(2,1,2) benchmark (RMSE$=0.124 \%$, error is 32.3% higher) and predicts the peak timing within a mean of 3.2 days error, compared to 18.4 days for linear extrapolation. Even with the earliest and strongest intervention scenario simulated (70% contact reduction beginning on day 42), there will be a peak demand for hospitalisations at 7.0 times the region's base capacity of intensive care beds, showing that model-informed planning will reduce but not eliminate the capacity gap during severe outbreak.

Hali tarjima qilinmagan

Identifikatorlar

Iqtiboslar va manbalar