Sequential ordinal modeling with applications to survival data

Area of Science:

• Statistics
• Biostatistics
• Health Informatics

Background:

• Ordinal response data is common in various fields, including healthcare.
• Existing models may not fully capture the complexities of sequential ordinal outcomes.
• Accurate modeling is crucial for understanding patient pathways and outcomes, such as hospital stay duration.

Purpose of the Study:

• To introduce and develop Markov chain Monte Carlo (MCMC) algorithms for fitting sequential ordinal models.
• To compare sequential ordinal models with other non-nested models using real-world data.
• To provide a robust statistical framework for analyzing ordered categorical data.

Main Methods:

• Development of MCMC algorithms based on Albert and Chib (1993).
• Application of these methods to a dataset on length of hospital stay post-heart surgery.
• Comparison of sequential, cumulative ordinal, Weibull, and log-logistic models using marginal likelihoods and training sample priors.

Main Results:

• The study demonstrates the feasibility and utility of MCMC for fitting sequential ordinal models.
• The analysis provides a detailed comparison of different ordinal models on a practical healthcare dataset.
• The findings highlight the strengths of sequential ordinal models in specific contexts.

Conclusions:

• Sequential ordinal models offer a valuable approach for analyzing ordered response data.
• MCMC methods provide an effective computational tool for fitting these complex models.
• The comparative analysis aids in selecting appropriate statistical models for healthcare research and similar fields.