Colossus: software for radiation epidemiological studies with big data

Area of Science:

• Epidemiology
• Biostatistics
• Radiation Science

Background:

• Growing need for survival analysis software for large-scale radiation epidemiology data.
• Existing software limitations in handling millions of rows of data.

Purpose of the Study:

• Introduce Colossus, an R package for scalable survival analysis.
• Provide total and relative rate equations for use with regression models.
• Validate Colossus performance against existing software and published results.

Main Methods:

• Development of the Colossus R package.
• Implementation of total and relative rate equations.
• Application of Cox proportional hazards, Poisson, and Fine-Grey regression models.
• Comparative analysis with existing software (Epicure) and published data.

Main Results:

• Colossus successfully analyzed large radiation epidemiology datasets (tens of millions of rows).
• Results obtained using Colossus were in agreement with existing software.
• Validation confirmed Colossus's accuracy compared to previous publications.

Conclusions:

• Colossus provides a scalable and accurate solution for survival analysis in radiation epidemiology.
• The R package is suitable for analyzing large datasets, such as those from the Million Person Study.
• Colossus performance is validated and comparable to established methods.