A discrete Weibull proportional odds survival model

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Discrete time-to-event data presents unique challenges in survival analysis.
• Existing models may not adequately capture the nuances of discrete survival outcomes.
• The proportional odds assumption is critical but requires robust checking procedures.

Purpose of the Study:

• To develop and present the Proportional Odds Model specifically for discrete time-to-event data.
• To incorporate the discrete Weibull distribution as a baseline for model inference.
• To propose and evaluate methods for assessing the proportional odds assumption.

Main Methods:

• Development of the Proportional Odds Model for discrete survival data.
• Inference on model parameters using point and interval estimates.
• Application of the discrete Weibull distribution as the baseline hazard.
• Proposal of diagnostic procedures for the proportional odds assumption.
• Simulation studies to assess estimator properties.

Main Results:

• The Proportional Odds Model for discrete data was successfully developed.
• Point and interval estimates for model parameters were obtained.
• Procedures for checking the proportional odds assumption were established.
• Simulation studies demonstrated the asymptotic properties of the estimators.
• The model's utility was shown in analyzing leukemia patient survival data.

Conclusions:

• The developed Proportional Odds Model offers a valuable tool for analyzing discrete time-to-event data.
• The discrete Weibull baseline provides a flexible framework for survival analysis.
• The proposed methods enhance the reliability of survival data modeling and assumption checking.