Discrete-time survival models with long-term survivors

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Discrete-time survival data present unique analytical challenges including discreteness, ties, and concomitant information.
• Existing discrete-time survival models may not adequately address the presence of long-term survivors or 'cured' individuals.

Purpose of the Study:

• To review existing discrete-time survival models.
• To extend these models to discrete-time cure survival models, specifically accounting for long-term survivors.
• To evaluate the performance of these extended models using simulation and real-world data.

Main Methods:

• Review of established discrete-time survival analysis techniques.
• Development and application of discrete-time cure survival models.
• Parameter estimation using maximum likelihood estimation and approximate partial likelihood approaches.
• Simulation studies to assess model suitability.
• Application to bladder tumor recurrence data.

Main Results:

• Simulation results support the effectiveness of the proposed discrete-time cure survival models.
• The models are suitable for analyzing discrete-time survival data with a cured fraction.

Conclusions:

• Discrete-time cure survival models provide a robust framework for analyzing survival data with long-term survivors.
• The proposed methods offer valuable tools for biostatisticians and researchers dealing with such data.