Introduction
The initial COVID-19 outbreak in Wuhan presented a unique challenge for epidemiological modeling. Traditional Susceptible-Exposed-Infectious-Recovered (SEIR) models often assume continuous transitions and static environments, which fail to capture the "jagged" reality of policy shifts and resource limitations. This paper introduces the e-ISHR model, an extension of the SEIR framework designed to account for discrete time delays, multi-stage government interventions, and the critical impact of hospital capacity.
The e-ISHR Model: Addressing Real-World Complexity
The e-ISHR model tracks five distinct populations: Exposed ($E$), Incubation ($I$), Symptomatic ($S$), Hospitalized/Home isolated ($H$), and Recovered ($R$). It addresses three primary gaps in standard modeling:
- Discrete Time Delays: Instead of continuous transitions, the model uses a Poisson distribution to simulate the average time spent in incubation, symptomatic, and recovery stages.
- 3-Staged Intervention Mechanism: The outbreak is divided into three phases:
- Stage 1 (Natural Spread): Before Jan 20, 2020—minimal protection and free movement.
- Stage 2 (Intermediate Control): Jan 21 to Feb 11—lockdowns and rising awareness, but medical resources are strained.
- Stage 3 (Late Control): After Feb 12—improved diagnostic criteria and sufficient medical resources.
- Hospital Resource Saturation: The model explicitly incorporates hospital bed capacity ($N_{hos}$). When symptomatic cases exceed capacity ($S > N_{hos}$), excess patients are isolated at home, where higher contact rates ($\beta_{hom}$) accelerate spread compared to controlled hospital environments.
Key Results and Accuracy
The e-ISHR model demonstrated significant predictive accuracy, particularly for the Wuhan trajectory. While traditional SEIR models exhibited error rates as high as 23.71%, the e-ISHR model achieved a Mean Absolute Percentage Error (MAPE) of 1.13%.
The model successfully predicted that the epidemic in Wuhan would peak and gradually subside after March 8, 2020, with total cases stabilizing around 55,000. A critical insight from the model is the "Hospital Overflow" feedback loop: insufficient hospital beds directly drive higher transmission rates by forcing home isolation.
Sensitivity Analysis: The Role of Infrastructure
Sensitivity analysis highlighted the quantitative impact of "flattening the curve" through infrastructure. As shown in Figure 6, increasing hospital capacity from 1,000 to 20,000 beds drastically reduces the total number of confirmed cases. This provides mathematical validation for the rapid construction of specialized facilities like the Huoshenshan and Leishenshan hospitals.
Conclusion
The e-ISHR model provides a more nuanced tool for public health assessment by integrating policy stages and resource constraints. While it does not account for inter-city migration or viral mutations, its ability to reflect the impact of diagnostic shifts and hospital capacity makes it a valuable framework for managing future pandemic responses.
Summary updated