Sensor-based localization of epidemic sources on human mobility networks

Area of Science:

• Epidemiology
• Mathematical Modeling
• Network Science

Background:

• Source detection is critical for epidemic control.
• Waterborne diseases like cholera pose significant public health challenges.
• Understanding disease transmission dynamics is essential for effective interventions.

Purpose of the Study:

• To develop a mathematical framework for identifying epidemic sources.
• To apply a Bayesian approach incorporating human mobility data for source localization.
• To demonstrate the method's efficacy using a real-world cholera outbreak.

Main Methods:

• Reformulated source detection as identifying components in a multivariate Gaussian mixture model.
• Calibrated model parameters using human mobility networks within a stochastic, spatially explicit epidemiological model.
• Employed a Bayesian perspective for incorporating prior information on source location.
• Utilized first-arrival times from observers at a subset of nodes for posterior-based inference.

Main Results:

• The proposed method successfully estimates epidemic source locations or regions.
• The approach is robust and requires minimal observer data (first-arrival times).
• Demonstrated effectiveness in analyzing the 2000-2002 KwaZulu-Natal cholera outbreak.

Conclusions:

• The developed Bayesian mixture model offers a powerful tool for epidemic source detection.
• Incorporating human mobility networks enhances the accuracy of epidemiological models.
• This method provides valuable insights for public health strategies in managing waterborne disease outbreaks.