Analysis of COVID-19 Prevention and Control Effects Based on the SEITRD Dynamic Model and Wuhan Epidemic Statistics

Yusheng Zhang, Liang Li, Yuewen Jiang, Biqing HuangInternational Journal of Environmental Research and Public Health 2020 · DOI

TL;DR

This paper introduces a novel SEITRD (Susceptible, Exposed, Infectious, Treatment, Recovered, Death) dynamic model to simulate COVID-19 transmission in Wuhan. By analyzing various control strategies like hospitalization rates, intervention timing, and interval control, the study provides data-driven insights into their effectiveness in mitigating the epidemic's impact.

What Problem Does This Paper Address?

The COVID-19 pandemic presented an unprecedented global health crisis, characterized by rapid spread and a lack of immediate effective treatments or vaccines. This necessitated a deep understanding of disease dynamics and the impact of various public health interventions. While many models exist for infectious diseases, the authors aimed to develop a model specifically tailored to COVID-19's characteristics, incorporating elements like a treatment stage and considering the nuances of real-world control measures. The core challenge was to quantify the effectiveness of different control strategies—such as varying hospitalization rates, intervention timing, and the duration/frequency of controls—to inform optimal public health policy decisions.

Methods

The researchers developed a compartmental dynamic model, extending the standard SEIR (Susceptible, Exposed, Infectious, Recovered) model. Their SEITRD model introduces two additional compartments: T (Treatment) to represent individuals receiving medical care, and D (Death) to track mortality. This model accounts for the incubation period, infectiousness of exposed and infected individuals, and the non-infectious nature of those in treatment or recovered stages.

Key methodological aspects include:

  • Model Extension: The SEITRD model builds upon existing SEIR frameworks by explicitly including treatment and death stages, which are crucial for understanding disease progression and outcomes.
  • Parameter Estimation: The model's parameters were estimated using real-world epidemic data from Wuhan, China, including confirmed cases, deaths, and recovery rates. This involved a combination of public data and estimations based on the epidemic's progression.
  • Simulation and Scenario Analysis: The core of the study involved simulating the SEITRD model under various hypothetical control scenarios. These scenarios manipulated key parameters representing:
  • Hospitalization Ratio (p): The proportion of exposed or infected individuals seeking medical treatment.
  • Intervention Time (t1, t2): The timing of implementing control measures.
  • Control Interval: The frequency and duration of control measures (e.g., strict vs. relaxed intervals).
  • Evaluation Metric: The impact of different control strategies was quantified using "Total Affected Days" (J), a weighted sum of days individuals spent in the E, I, and T stages, representing the overall burden of the epidemic.

Datasets

The study primarily utilized epidemic data from Wuhan, China, including confirmed cases, deaths, and recovery statistics. This data was used for parameter estimation and model validation. The authors also mention simulating the outbreak in Egypt using similar methods, referencing its COVID-19 statistics from April to October.

Key Insights and Ideas

  • The Importance of Early and Consistent Intervention: The simulations strongly suggest that implementing control measures earlier in the epidemic's trajectory leads to significantly better outcomes, reducing the overall number of affected individuals and the total disease burden.
  • Hospitalization as a Key Control Lever: Increasing the hospitalization ratio, especially in the early stages of the epidemic, proved to be a highly effective strategy for reducing the number of infectious individuals and mitigating the overall spread. This highlights the role of timely medical care not just for patient recovery but also for disease containment.
  • The Trade-off in Control Intervals: While strict, continuous control measures are most effective, the study also shows that even a well-timed, shorter control interval can offer substantial benefits compared to prolonged, less frequent interventions. This suggests a practical approach to balancing epidemic control with societal disruption.
  • Model Validation through Real-World Data: The SEITRD model's predictions for Wuhan's epidemic trend closely matched the actual observed data, lending credibility to its ability to simulate disease dynamics and evaluate control strategies.

Key Results

  • Hospitalization Ratio: Increasing the hospitalization ratio from 0.4 to 0.6 reduced total affected days by approximately 7.9% (scheme 2 vs. scheme 1). Conversely, a lower ratio (0.4) led to an 8.6% increase in affected days (scheme 3 vs. scheme 1). Higher hospitalization ratios in the early stage were particularly impactful.
  • Intervention Timing: Earlier intervention (scheme 3, t1=5 days, t2=15 days) resulted in a 1.5% reduction in total affected days compared to the baseline (scheme 1, t1=10 days, t2=20 days). Delayed interventions (scheme 4, 5) showed a slight increase in affected days.
  • Control Intervals: A strict control interval (1 day) reduced total affected days by approximately 4.0% compared to a normal interval (3 days), and 0.8% compared to a relaxed interval (5 days).

Limitations and Future Work

The authors acknowledge several limitations:

  • Model Simplifications: The model treats the population as homogeneous, not accounting for individual attributes like age, health status, or varying susceptibility. It also doesn't explicitly model inter-city traffic, which can be a significant factor in disease spread.
  • Parameter Uncertainty: While efforts were made to use real data, some parameters were estimated or calculated, introducing potential inaccuracies. The dynamic nature of parameters during an epidemic also poses a challenge.
  • Real-World Complexity: The study focuses on a few key control variables. In reality, a multitude of factors, including public awareness, mask-wearing, and individual behavior, play crucial roles and are difficult to fully capture in a model.

Future work could involve:

  • Incorporating more complex population heterogeneity (age, comorbidities).
  • Modeling inter-city and international travel dynamics.
  • Exploring a wider range of control strategies and their synergistic effects.
  • Developing more robust methods for real-time parameter estimation and model adaptation.
  • Investigating the balance between epidemic control and the disruption of normal life.

Summary updated

Abstract

Since December 2019, millions of people worldwide have been diagnosed with COVID-19, which has caused enormous losses. Given that there are currently no effective treatment or prevention drugs, most countries and regions mainly rely on quarantine and travel restrictions to prevent the spread of the epidemic. How to find proper prevention and treatment methods has been a hot topic of discussion. The key to the problem is to understand when these intervention measures are the best strategies for disease control and how they might affect disease dynamics. In this paper, we build a transmission dynamic model in combination with the transmission characteristics of COVID-19. We thoroughly study the dynamical behavior of the model and analyze how to determine the relevant parameters, and how the parameters influence the transmission process. Furthermore, we subsequently compare the impact of different control strategies on the epidemic, the variables include intervention time, control duration, control intensity, and other model parameters. Finally, we can find a better control method by comparing the results under different schemes and choose the proper preventive control strategy according to the actual epidemic stage and control objectives.
BibTeX
@article{zhang2020analysis,
  author = {Yusheng Zhang and Liang Li and Yuewen Jiang and Biqing Huang},
  title = {{Analysis of COVID-19 Prevention and Control Effects Based on the SEITRD Dynamic Model and Wuhan Epidemic Statistics}},
  journal = {International Journal of Environmental Research and Public Health},
  year = {2020},
  doi = {10.3390/ijerph17249309},
}
Figure