Interactive tool for clustering and forecasting patterns of Taiwan COVID-19 spread

Area of Science:

• Data Science
• Epidemiology
• Public Health

Background:

• Effective COVID-19 data analysis is crucial for policy decisions and outbreak management.
• Traditional time series methods struggle with high-volume COVID-19 data and interactive application demands.
• There is a need for scalable and compelling clustering and forecasting techniques for infectious disease data.

Purpose of the Study:

• To develop a web-based interactive tool for clustering and forecasting COVID-19 confirmed infection cases in Taiwan.
• To implement a computationally efficient and user-friendly parametric forecasting method.
• To assist policymakers and medical researchers in understanding disease spread patterns.

Main Methods:

• Utilized Model-based (MOB) tree for clustering Taiwan COVID-19 data based on domain-relevant attributes.
• Employed Ordinary Least Squares (OLS) for forecasting, using models generated by the MOB tree within each cluster.
• Developed an R Shiny App for a user-friendly interface, allowing parameter selection for clustering and forecasting.

Main Results:

• The proposed tool effectively clusters and forecasts COVID-19 infection cases.
• The MOB tree and OLS approach provide a scalable and computationally inexpensive solution.
• The interactive web application facilitates easy access to clustering and forecasting results.

Conclusions:

• The developed web-based tool offers a practical solution for analyzing and forecasting COVID-19 spread in Taiwan.
• This approach aids in identifying disease patterns, supporting public health interventions and medical research.
• The user-friendly interface empowers practitioners to explore different analytical parameters and gain insights from the data.