Maximum likelihood estimation for the proportional odds model with mixed interval-censored failure time data

Area of Science:

• Biostatistics
• Survival Analysis
• Medical Informatics

Background:

• Mixed interval-censored failure time data are prevalent in longitudinal biomedical studies, including childhood cancer survivorship.
• Existing regression methods often handle only right-censored or interval-censored data, not mixed types.
• Proportional hazards models, commonly used, may not be suitable for all mixed interval-censored data scenarios.

Purpose of the Study:

• To develop a novel regression model for analyzing mixed interval-censored failure time data.
• To address limitations of existing methods that assume proportional hazards.
• To provide a flexible analytical tool for complex survival data in biomedical research.

Main Methods:

• Developed a maximum likelihood estimation procedure for the proportional odds regression model.
• The model is specifically designed for mixed interval-censored failure time data.
• Validated the method through an extensive simulation study to assess finite-sample properties.

Main Results:

• The proposed maximum likelihood estimators are consistent and asymptotically Gaussian.
• Simulation results indicate the method performs well in practical situations.
• The approach was successfully applied to analyze the impact of cranial radiation therapy on growth hormone deficiency in childhood cancer survivors.

Conclusions:

• The proportional odds regression model provides a viable and effective approach for analyzing mixed interval-censored failure time data.
• This method offers an important alternative when proportional hazards assumptions are violated.
• The findings have significant implications for understanding long-term outcomes in childhood cancer survivors.