Deep Dynamic Epidemiological Modelling for COVID-19 Forecasting in Multi-level Districts

Ruhan Liu,Jiajia Li,Yang Wen,Huating Li,Ping Zhang,Bin Sheng,David Dagan Feng

2023-06-21

Abstract:Objective: COVID-19 has spread worldwide and made a huge influence across the world. Modeling the infectious spread situation of COVID-19 is essential to understand the current condition and to formulate intervention measurements. Epidemiological equations based on the SEIR model simulate disease development. The traditional parameter estimation method to solve SEIR equations could not precisely fit real-world data due to different situations, such as social distancing policies and intervention strategies. Additionally, learning-based models achieve outstanding fitting performance, but cannot visualize mechanisms. Methods: Thus, we propose a deep dynamic epidemiological (DDE) method that combines epidemiological equations and deep-learning advantages to obtain high accuracy and visualization. The DDE contains deep networks to fit the effect function to simulate the ever-changing situations based on the neural ODE method in solving variants' equations, ensuring the fitting performance of multi-level areas. Results: We introduce four SEIR variants to fit different situations in different countries and regions. We compare our DDE method with traditional parameter estimation methods (Nelder-Mead, BFGS, Powell, Truncated Newton Conjugate-Gradient, Neural ODE) in fitting the real-world data in the cases of countries (the USA, Columbia, South Africa) and regions (Wuhan in China, Piedmont in Italy). Our DDE method achieves the best Mean Square Error and Pearson coefficient in all five areas. Further, compared with the state-of-art learning-based approaches, the DDE outperforms all techniques, including LSTM, RNN, GRU, Random Forest, Extremely Random Trees, and Decision Tree. Conclusion: DDE presents outstanding predictive ability and visualized display of the changes in infection rates in different regions and countries.

Machine Learning,Artificial Intelligence,Populations and Evolution

What problem does this paper attempt to address?

The main objective of this paper is to propose a new method to address several key issues in COVID-19 pandemic forecasting. Specifically, the study aims to tackle the following aspects:

1. **Improving prediction accuracy**: Traditional SEIR model-based methods often fail to accurately fit real-world data because model parameters are usually fixed and do not consider the impact of changing factors such as social distancing policies and quarantine measures.

2. **Enhancing model interpretability**: While machine learning-based methods can fit data well, their "black box" nature makes it difficult to understand the reasons behind the predictions, limiting the application of these models.

3. **Creating a highly generalizable model**: Existing improved SEIR models are usually designed for specific regions and are difficult to directly transfer to other regions. Therefore, a more general model is needed that can be flexibly applied across different countries and regions.

To address the above issues, the authors propose a method called Deep Dynamic Epidemiological (DDE). DDE combines the advantages of Neural Ordinary Differential Equations (Neural ODE) and epidemiological equations, achieving high-precision data fitting and good visualization capabilities. Additionally, the authors propose four variants based on the SEIR and SIR models (SIRD, SEIRD, SMCRD, SEMCRD), which consider different scenarios and data sources, thereby improving the adaptability and flexibility of the models.

Through experimental comparisons, the DDE method demonstrated significant advantages in fitting COVID-19 data across multiple regions. It not only achieved better performance in Mean Squared Error (MSE) but also excelled in the Pearson coefficient, indicating the method's superiority in prediction accuracy and model interpretability.

In summary, the goal of this paper is to develop a method that can both improve prediction accuracy and enhance model interpretability, while being widely applicable across different geographical regions.