Reparametrized generalized gamma partially linear regression with application to breast cancer data

Area of Science:

• Statistics
• Biostatistics
• Survival Analysis

Background:

• Regression models are crucial for analyzing complex data.
• Generalized gamma distribution offers flexibility in modeling survival data.
• Interpretable components are needed for practical applications.

Purpose of the Study:

• To develop a novel partially linear regression model.
• To utilize a reparametrized generalized gamma distribution for improved interpretability.
• To apply the new methodology to real-world breast cancer data.

Main Methods:

• Construction of a partially linear regression model.
• Reparametrization of the generalized gamma distribution.
• Parameter estimation using penalized maximum likelihood.
• Simulation studies to assess estimator accuracy and residual distribution.

Main Results:

• The proposed model provides easily interpretable systematic components.
• Penalized maximum likelihood estimation demonstrates accuracy.
• Simulations confirm the reliability of the estimators across various settings.
• The methodology effectively analyzes breast cancer survival data.

Conclusions:

• The new partially linear regression model based on the generalized gamma distribution is a valuable tool.
• The model offers accurate parameter estimation and interpretable results.
• The methodology shows promise for application in epidemiological studies, such as breast cancer analysis.