Prediction of COVID-19 Outbreak in China and Optimal Return Date for University Students Based on Propagation Dynamics

Ganyu Huang, Qiaoyi Pan, Shuangying Zhao, Yucen Gao, Xiaofeng GaoJournal of Shanghai Jiaotong University (Science) 2020 · DOI

Introduction

In the early months of 2020, the rapid spread of COVID-19 in China necessitated sophisticated modeling to guide public policy. Traditional epidemiological models often struggled to capture the dynamic nature of government interventions and the evolving medical landscape. This paper introduces a specialized mathematical framework—the "4+1 penta-group model"—to predict the outbreak's trajectory and the "BAT model" to determine the optimal timing for university students to return to campus.

Methodology: The 4+1 Penta-Group Model

Standard SEIR (Susceptible-Exposed-Infectious-Recovered) models were deemed insufficient for the complexities of the COVID-19 outbreak. The researchers developed the 4+1 penta-group model, which augments the classic structure with a fifth compartment for Dead ($n_D$) individuals and introduces time-varying parameters to account for real-world changes in isolation and treatment.

Evolution of epidemiological models from basic SEIR to the proposed 4+1 penta-group model.

Key innovations in this model include:

  • Isolation Intensity ($p$): A parameter that quantifies the impact of government-mandated lockdowns and social distancing over time.
  • Dynamic Parameters: Unlike static models, recovery ($\gamma$) and mortality ($\mu$) rates were fitted as time-dependent functions ($R^2 > 0.93$), reflecting the continuous improvement in clinical treatments and healthcare capacity as the crisis progressed.

The BAT Decision Framework

To address the logistical challenge of millions of students returning to universities, the authors proposed the BAT model, a multi-criteria decision-making framework:

  • B (Back): A simulation of the "return rush" using population density data from major transit hubs like Shanghai Hongqiao Railway Station to adjust contact rates and infectiousness.
  • A (Analytic Hierarchy Process): A structured technique used to weight four critical factors: outbreak end date, total cases, economic impact, and student graduation.
  • T (TOPSIS): A ranking method used to identify the optimal return date by comparing various potential scenarios against a hypothetical "ideal" benchmark.

Key Insights and Results

The study found that traditional static models fail during dynamic crises because they ignore behavioral shifts. For instance, railway station data suggested that a return rush without precautions could quadruple daily contacts. However, simulations showed that maintaining strict precautionary measures (e.g., masks and isolation) would make the system resilient to secondary outbreaks or imported cases.

Simulation of the COVID-19 outbreak trajectory in China using the 4+1 penta-group model.

The model's predictions were highly correlated with real-world data ($r = 0.995$):

  • Outbreak End: The epidemic in China was predicted to effectively conclude around May 2, 2020, with a total of approximately 103,321 cases.
  • Optimal Return Window: The BAT model identified March 15 to March 22 as the ideal window for students to travel, recommending a university start date of March 23, 2020.

Conclusion and Limitations

The 4+1 penta-group model provided a robust tool for early pandemic prediction by incorporating the effects of social intervention and medical progress. While highly accurate for the initial wave, the authors noted that the model's validity depends on the continuation of strict public health measures. It does not account for later developments like viral variants or mass vaccination, which were not present during the study period.

Summary updated

BibTeX
@article{huang2020prediction,
  author = {Ganyu Huang and Qiaoyi Pan and Shuangying Zhao and Yucen Gao and Xiaofeng Gao},
  title = {{Prediction of COVID-19 Outbreak in China and Optimal Return Date for University Students Based on Propagation Dynamics}},
  journal = {Journal of Shanghai Jiaotong University (Science)},
  year = {2020},
  doi = {10.1007/s12204-020-2167-2},
}
Figure