Abstract:Effective epidemic forecasting is critical for public health strategies and efficient medical resource allocation, especially in the face of rapidly spreading infectious diseases. However, existing deep-learning methods often overlook the dynamic nature of epidemics and fail to account for the specific mechanisms of disease transmission. In response to these challenges, we introduce an innovative end-to-end framework called Epidemiology-Aware Neural ODE with Continuous Disease Transmission Graph (EARTH) in this paper. To learn continuous and regional disease transmission patterns, we first propose EANO, which seamlessly integrates the neural ODE approach with the epidemic mechanism, considering the complex spatial spread process during epidemic evolution. Additionally, we introduce GLTG to model global infection trends and leverage these signals to guide local transmission dynamically. To accommodate both the global coherence of epidemic trends and the local nuances of epidemic transmission patterns, we build a cross-attention approach to fuse the most meaningful information for forecasting. Through the smooth synergy of both components, EARTH offers a more robust and flexible approach to understanding and predicting the spread of infectious diseases. Extensive experiments show EARTH superior performance in forecasting real-world epidemics compared to state-of-the-art methods. The code will be available at <a class="link-external link-https" href="https://github.com/Emory-Melody/EpiLearn" rel="external noopener nofollow">this https URL</a>.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is that existing deep - learning methods fail to fully consider the dynamic nature of epidemics and specific transmission mechanisms when predicting the spread of infectious diseases. Specifically: 1. **Ignoring the dynamic evolution of epidemics**: Existing deep - learning methods usually overlook the dynamic changes in epidemic systems, such as the impact of changes in people's behavior (e.g., lockdowns) on inter - regional transmission. 2. **Unable to handle irregular sampling observation intervals**: In the real world, the data reporting frequencies in different regions may be inconsistent, resulting in irregular time intervals in the data, while existing methods are often only able to handle fixed time intervals. 3. **Lack of learning of global infection trends**: Existing models usually only focus on local transmission patterns and ignore the global infection trends caused by external factors (such.g., global vaccination trends, international travel restrictions, etc.). To solve these problems, the author proposes an innovative end - to - end framework - Epidemiology - Aware Neural ODE with Continuous Disease Transmission Graph (EARTH), aiming to more accurately capture and predict the spread of infectious diseases by combining neural ODE and epidemiological mechanisms. ### Main problem summary - **How to combine neural ODE and epidemiological mechanisms?** - Existing methods fail to explicitly learn epidemic mechanisms and are difficult to provide decision - support. - **How to flexibly learn continuous disease transmission patterns under limited observations?** - In the real world, data on various states of the disease (e.g., susceptibles) may be incomplete, and a flexible method is required to learn the underlying transmission patterns. - **How to model global infection trends and learn dynamic regional transmission patterns?** - Epidemics not only exhibit local spatial transmission patterns but are also affected by global trends. For example, changes in the global vaccination rate will affect local transmission. ### Solutions To address the above challenges, the EARTH framework makes the following improvements: 1. **Epidemiology - Aware Neural ODE (EANO)**: - Combine the classical SIR model and neural ODE to implicitly capture the continuous evolution of the regional transmission graph. - Initialize the disease - state features and input them into the EANO module to learn the underlying disease - transmission patterns. 2. **Global - guided Local Transmission Graph (GLTG)**: - Obtain a long - term view of epidemic progress and establish relationships between regions with similar development patterns. - Use GNN to fuse global - trend - indicating features, generate a more fine - grained dynamic transmission graph, and guide the disease - transmission in EANO. 3. **Cross - attention mechanism**: - Integrate global consistency (epidemic trends) and local nuances (disease - transmission patterns) to achieve more accurate predictions. Through these improvements, EARTH can more flexibly capture time - continuous epidemic mechanisms and has achieved better performance than existing methods on multiple epidemic - prediction data sets.

Epidemiology-Aware Neural ODE with Continuous Disease Transmission Graph

Predicting the Epidemics Trend of COVID-19 Using Epidemiological-Based Generative Adversarial Networks.

Graph Neural Ordinary Differential Equations for Epidemic Forecasting

EpiGNN: Exploring Spatial Transmission with Graph Neural Network for Regional Epidemic Forecasting

DDE: Deep Dynamic Epidemiological Modelling for Infectious Illness Development Forecasting in Multi-level Geographic Entities

DDE: Deep Dynamic Epidemiological Modeling for Infectious Illness Development Forecasting in Multi-level Geographic Entities

Epicasting: An Ensemble Wavelet Neural Network (EWNet) for Forecasting Epidemics

DeepDynaForecast : Phylogenetic-informed graph deep learning for epidemic transmission dynamic prediction

Epicasting: An Ensemble Wavelet Neural Network for forecasting epidemics

NeuralCODE: Neural Compartmental Ordinary Differential Equations Model with AutoML for Interpretable Epidemic Forecasting

EINNs: Epidemiologically-Informed Neural Networks

A Novel Multi-scale Spatiotemporal Graph Neural Network for Epidemic Prediction.

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

Backbone-based Dynamic Graph Spatio-Temporal Network for Epidemic Forecasting

Epidemiology-informed Graph Neural Network for Heterogeneity-aware Epidemic Forecasting

An early warning indicator trained on stochastic disease-spreading models with different noises

Epi-DNNs: Epidemiological Priors Informed Deep Neural Networks for Modeling COVID-19 Dynamics.

MepoGNN: Metapopulation Epidemic Forecasting with Graph Neural Networks

EpiLearn: A Python Library for Machine Learning in Epidemic Modeling

Machine learning for data-centric epidemic forecasting