Optimal resource allocation with spatiotemporal transmission discovery for effective disease control

Area of Science:

• Epidemiology
• Computational Biology
• Public Health

Background:

• The rapid spread of the SARS-CoV-2 Omicron variant strains global healthcare systems.
• Limited testing, vaccination, and hospital capacity exacerbate the challenge of controlling widespread outbreaks.
• Accurate forecasting of transmission dynamics is crucial for effective public health interventions.

Purpose of the Study:

• To develop a machine learning method for inferring daily transmission risks of the Omicron variant.
• To create a data-driven strategy for prioritizing healthcare resources during outbreaks.
• To mitigate the impact of Omicron variant outbreaks in densely populated regions.

Main Methods:

• Constructed spatiotemporal transmission intensity matrices using Omicron infection data.
• Employed Gaussian process modeling to forecast daily district-level transmission risks.
• Developed a transmission-guided resource prioritization strategy for optimal allocation.

Main Results:

• Risk mapping revealed irregular and dynamic spatiotemporal transmission patterns of the Omicron variant.
• The proposed strategy, with 300,000 daily testing capacity, reduced infection peaks by 87.1% compared to population-based allocation.
• A 24.2% reduction in infection peak was observed compared to case-based allocation, easing healthcare system strain.

Conclusions:

• Computational characterization of transmission patterns enables effective risk mapping and resource prioritization.
• Adaptive strategies are vital for timely outbreak control, especially with limited healthcare capacity.
• The method offers a framework for managing current Omicron outbreaks and future variant-driven epidemics.