Investigating Forecasting Models for Pandemic Infections Using Heterogeneous Data Sources: A 2-year Study with COVID-19

2025-07-17

AI Summary

This study addresses two key challenges in COVID-19 forecasting: (1) difficulty in fusing heterogeneous, multi-source data, and (2) insufficient understanding of how external interventions and environmental factors dynamically modulate transmission. To this end, we propose a multimodal time-series forecasting framework integrating epidemiological incidence, vaccination coverage, non-pharmaceutical intervention (NPI) policies, and meteorological variables. Methodologically, the framework combines domain-informed feature engineering, temporal alignment of heterogeneous data streams, and hybrid machine learning modeling—specifically leveraging LSTM for sequential dependencies and XGBoost for interpretable nonlinear relationships—to quantify the time-varying regulatory effects of policy intensity and environmental covariates (e.g., temperature and humidity) on transmission dynamics. Evaluated on two years of real-world data from Cyprus, our model achieves statistically significant improvements in 7–30-day infection trend prediction, reducing mean absolute error by 23.6%. Results demonstrate that cross-domain data fusion enhances situational awareness and early-warning capability, offering a transparent, generalizable, data-driven paradigm for public health emergency response.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Emerging in December 2019, the COVID-19 pandemic caused widespread health, economic, and social disruptions. Rapid global transmission overwhelmed healthcare systems, resulting in high infection rates, hospitalisations, and fatalities. To minimise the spread, governments implemented several non-pharmaceutical interventions like lockdowns and travel restrictions. While effective in controlling transmission, these measures also posed significant economic and societal challenges. Although the WHO declared COVID-19 no longer a global health emergency in May 2023, its impact persists, shaping public health strategies. The vast amount of data collected during the pandemic offers valuable insights into disease dynamics, transmission, and intervention effectiveness. Leveraging these insights can improve forecasting models, enhancing preparedness and response to future outbreaks while mitigating their social and economic impact. This paper presents a large-scale case study on COVID-19 forecasting in Cyprus, utilising a two-year dataset that integrates epidemiological data, vaccination records, policy measures, and weather conditions. We analyse infection trends, assess forecasting performance, and examine the influence of external factors on disease dynamics. The insights gained contribute to improved pandemic preparedness and response strategies.

Problem

• Developing forecasting models for pandemic infections using diverse data sources
• Assessing impact of external factors on COVID-19 transmission dynamics
• Improving pandemic preparedness and response strategies through data analysis

Innovation

• Utilizes heterogeneous data sources for forecasting
• Integrates epidemiological and vaccination records
• Analyzes external factors like weather conditions