Systematic Determinants of Global COVID-19 Burden: Longitudinal Time-Series Analysis Using Big Data-Driven Artificial...

Area of Science:

• Epidemiology
• Machine Learning
• Public Health

Background:

• COVID-19 continues to resurge globally, influenced by viral evolution, population immunity, and environmental factors.
• The dynamic interplay shaping disease transmission and severity remains poorly understood.

Purpose of the Study:

• To quantify the relative contributions and temporal dynamics of viral variants, immunity, environmental conditions, and public health measures on COVID-19 burden.
• To analyze the impact on transmission (Rt), hospitalizations, ICU admissions, and deaths.

Main Methods:

• A retrospective, longitudinal, time-series analysis using big data and interpretable machine learning (XGBoost with Shapley Additive Explanations).
• Integrated global data from 38 countries (pandemic onset to Dec 2022) on variants, immunity, environment, policies, healthcare, and migration.

Main Results:

• Viral variants significantly drive transmission (Rt), with diminishing impact on severe outcomes.
• Natural infection and COVID-19 vaccination contributions increase with disease severity.
• Identified critical vaccination coverage thresholds for transmission reduction (29.9%) and ICU prevention (72.3%).
• Routine immunizations, like the yellow fever vaccine, showed cross-protective effects.
• Environmental factors like temperature and healthcare infrastructure also influenced disease burden.

Conclusions:

• This interpretable framework quantifies variant fitness, dual vaccination thresholds, and cross-protection from routine immunizations.
• It integrates multidomain data for a dynamic understanding of COVID-19 burden.
• Applications include adaptive vaccination strategies, targeted variant surveillance, and enhanced pandemic preparedness through routine immunization programs.