Cluster-Based Analysis of Infectious Disease Occurrences Using Tensor Decomposition: A Case Study of South Korea

Area of Science:

• Epidemiology
• Data Science
• Public Health

Background:

• Infectious diseases like lower respiratory infections and diarrheal diseases cause significant social and economic losses.
• Existing methods for analyzing disease occurrences have limitations due to the complex interplay of numerous influencing factors.
• A need exists for advanced analytical techniques to better understand and predict infectious disease outbreaks.

Purpose of the Study:

• To propose a novel cluster-based analysis scheme for infectious disease occurrences.
• To identify commonalities and differences between disease outbreak patterns by grouping similar elements.
• To overcome the limitations of previous analyses by accounting for complex interactions between factors.

Main Methods:

• Collected and preprocessed infectious disease occurrence data based on time, region, and disease.
• Constructed a data tensor and applied Tucker decomposition to extract latent features across dimensions.
• Utilized k-means clustering on extracted latent features for pattern analysis.

Main Results:

• The cluster-based analysis scheme effectively grouped infectious disease occurrences based on similar patterns.
• Latent features extracted via Tucker decomposition provided insights into temporal, regional, and disease-specific dynamics.
• A case study using South Korean data demonstrated the scheme's practical application and effectiveness.

Conclusions:

• The proposed cluster-based analysis scheme offers a robust method for understanding complex infectious disease dynamics.
• This approach enhances the ability to identify commonalities and differences in disease outbreaks, aiding prevention strategies.
• The findings suggest improved proactive responses to reduce the social and economic costs associated with epidemics.