Colossus: bridging the gap between big data and radiation epidemiology

Area of Science:

• Radiation Epidemiology
• Computational Biology
• Statistical Modeling

Background:

• Addressing complex questions in radiation epidemiology requires advanced software for big data analysis.
• Existing tools may lack the flexibility and computational power needed for intricate exposure uncertainties.

Purpose of the Study:

• To introduce Colossus, an open-source R package designed for complex modeling in radiation epidemiology.
• To enhance Colossus with methods for propagating covariate and exposure uncertainties in risk models.

Main Methods:

• Development of the open-source R package Colossus.
• Implementation of multi-core processing for faster analyses.
• Integration of methods for analyzing multiple exposure realizations, including frequentist model averaging and Monte Carlo maximum likelihood, guided by NCRP Commentary 34.

Main Results:

• Colossus leverages R's flexibility and multi-core systems for efficient big data analysis.
• The package now incorporates advanced techniques to handle covariate and exposure uncertainties.
• New methods allow for the application of complex risk models to datasets with intricate exposure uncertainties.

Conclusions:

• Colossus provides a flexible and powerful platform for radiation epidemiology research using big data.
• The inclusion of uncertainty propagation methods enhances the reliability of risk assessments in radiation studies.
• Colossus is designed for future expansion, promising continued advancements in the field.