The generalized F distribution: an umbrella for parametric survival analysis

Area of Science:

• Statistics
• Biostatistics
• Epidemiology

Background:

• Parametric survival analysis is crucial for modeling time-to-event data.
• The generalized gamma (GG) distribution is a flexible platform for survival analysis, encompassing common distributions and hazard function types.
• The generalized F (GF) distribution, which includes the GG and log logistic distributions, offers even greater flexibility.

Purpose of the Study:

• To characterize the hazard functions of the generalized F (GF) distribution.
• To explore the GF distribution's utility in parametric survival modeling.
• To apply the GF distribution to analyze survival data in different HIV therapy eras.

Main Methods:

• Characterization of GF distribution hazard functions.
• Exploration of GF distribution properties and parameterizations.
• Application of GF models to HIV/AIDS survival data from four distinct eras.

Main Results:

• The GF distribution's hazard functions are characterized, showing limitations to decreasing and arc-shaped forms, and can be decreasing but not monotone.
• The GF distribution provides additional flexibility beyond the GG distribution for parametric modeling.
• The GF distribution was applied to refine the description of hazard functions for death after AIDS diagnosis across different HIV therapy eras.

Conclusions:

• The GF distribution offers advanced flexibility for parametric survival analysis.
• Understanding GF hazard function characteristics is essential for appropriate model selection.
• The GF distribution provides a valuable tool for analyzing complex survival data, such as that from HIV/AIDS clinical trials.