Enhancing COVID-19 forecasting precision through the integration of compartmental models, machine learning and...

Area of Science:

• Epidemiology
• Computational Biology
• Machine Learning

Background:

• Accurate epidemic forecasting is crucial for effective public health interventions.
• Computational models offer insights into disease progression and early detection.
• Existing models struggle with dynamic changes like new variants and trend shifts.

Purpose of the Study:

• Introduce Sybil, a novel framework for epidemic prediction.
• Evaluate Sybil's accuracy in forecasting trend changes and variant emergence.
• Compare Sybil's performance against traditional data-centric methods.

Main Methods:

• Developed Sybil, integrating machine learning with variant-aware compartmental models.
• Utilized a hybrid approach combining data-driven and analytical methodologies.
• Validated Sybil using COVID-19 data from European and U.S. states, including case counts, fatalities, and variant data.

Main Results:

• Sybil accurately forecasts shifts in pandemic trends and their magnitude.
• The framework successfully predicts the future prevalence of new variants.
• Sybil demonstrates superior performance compared to conventional data-centric forecasting approaches.

Conclusions:

• Sybil provides a robust and accurate method for predicting epidemic evolution.
• The framework's ability to handle dynamic changes improves public health preparedness.
• Sybil offers a valuable tool for informed decision-making during infectious disease outbreaks.