Incorporating connectivity among Internet search data for enhanced influenza-like illness tracking

Area of Science:

• Computational epidemiology
• Public health surveillance
• Big data analytics

Background:

• Internet data offers potential for societal trend analysis, particularly for infectious disease tracking.
• Existing disease tracking frameworks often fail to address complex connectivity within internet search data.
• Accurate infectious disease surveillance is crucial for public health decision-making.

Purpose of the Study:

• To propose ARGO-C (Augmented Regression with Clustered GOogle data), an integrative approach to enhance infectious disease tracking.
• To incorporate the clustering structure of internet search data for improved accuracy and interpretability.
• To demonstrate the effectiveness of ARGO-C for multi-resolution influenza-like illness (ILI) tracking.

Main Methods:

• Developed ARGO-C, a statistically principled method integrating clustered internet search data.
• Applied ARGO-C to multi-resolution %ILI tracking.
• Compared ARGO-C performance against benchmark methods across various geographical resolutions.

Main Results:

• ARGO-C demonstrated improved performance and robustness over benchmark methods for %ILI tracking.
• The approach showed enhanced accuracy and interpretability in disease tracking.
• Effectiveness was validated at multiple geographical scales.

Conclusions:

• ARGO-C offers a more accurate and interpretable framework for infectious disease surveillance using internet data.
• The method is adaptable for tracking various diseases beyond influenza, as well as other societal trends.
• This approach provides valuable tools for public health officials and researchers.