Toward Combatting COVID-19: A Risk Assessment System

Area of Science:

• Epidemiology
• Data Science
• Public Health

Background:

• The COVID-19 pandemic has generated vast datasets, yet granular data for local areas remains scarce.
• Existing COVID-19 data is often aggregated at macrolevels (state, county), limiting neighborhood-level situational awareness.
• The Internet of Medical Things (IoMT) offers potential for real-time data collection and dissemination.

Purpose of the Study:

• To develop a novel two-level risk assessment system for COVID-19.
• To generate fine-grained, microlevel risk indices for geographical coordinates.
• To overcome limitations of macrolevel COVID-19 data availability.

Main Methods:

• Proposed a two-level risk assessment system defining a "risk index."
• Developed the MK-DNN model, integrating multikernel density estimation (MKDE) and deep neural networks (DNN).
• Trained the MK-DNN model at the macrolevel to predict microlevel risk indices, validated using a heuristic method.

Main Results:

• The MK-DNN model successfully generated accurate microlevel risk indices for specific geographic coordinates.
• Simulations on real-world data confirmed the system's accuracy and validity.
• The system enables a more granular understanding of disease spread in local communities.

Conclusions:

• The proposed two-level risk assessment system effectively addresses the need for fine-grained COVID-19 data.
• MK-DNN provides a reliable method for calculating microlevel risk indices, enhancing disease surveillance.
• This approach can significantly improve public health responses and situational awareness at the community level.