A linear dynamical perspective on epidemiology: interplay between early COVID-19 outbreak and human mobility

Area of Science:

• Epidemiology
• Complex Systems
• Data Science

Background:

• Human activity and mobility (HAM) are critical factors in epidemic dynamics.
• Early COVID-19 spread showed strong correlations with human mobility patterns.
• Understanding the interplay between HAM and disease transmission is crucial for effective control.

Purpose of the Study:

• To investigate the impact of HAM on epidemic spreading dynamics.
• To model the early spread of COVID-19 as a coupled system with HAM.
• To develop a data-driven approach for predicting epidemic trajectories based on mobility.

Main Methods:

• Application of the generalized Koopman framework with control inputs.
• Utilizing Hankel dynamic mode decomposition with Control (HDMDc) algorithm.
• Analysis of spatiotemporal data without requiring knowledge of physical laws.

Main Results:

• A locally linear model of disease spread incorporating HAM as a control input was derived.
• The model accurately captured the impact of local mobility on early COVID-19 dynamics.
• Accurate forecasting of new infections for 2-4 week prediction windows was achieved.
• A leader-follower relationship between mobility and disease spread was confirmed.

Conclusions:

• The proposed data-driven Koopman framework effectively models epidemic dynamics influenced by HAM.
• Human mobility is a significant driver and predictor of early-stage epidemic spread.
• The methodology offers a robust tool for real-time epidemic monitoring and short-term forecasting.