Competing-triggering effect models for multitype recurrent event data

Area of Science:

• Biostatistics
• Survival Analysis
• Epidemiology

Background:

• Multitype recurrent events are common in lifetime data analysis.
• Previous models focused on single event types and nonlinear triggering effects.
• Existing proportional intensity models have limitations in capturing complex triggering dynamics.

Purpose of the Study:

• To propose a general Cox-type structure for modeling triggering effects among multiple event types.
• To allow for separate triggering effects for each event type on events of interest.
• To develop a flexible nonlinear model formulation with common parameters across different event types.

Main Methods:

• Developed a general Cox-type structure for multitype recurrent event data.
• Derived partial likelihood estimators for model parameters.
• Established consistency and asymptotic normality of the proposed estimators.
• Provided plug-in variance estimators for practical application.

Main Results:

• Demonstrated good performance of the proposed methods through simulation experiments.
• Successfully applied the models to analyze triggering effects between breast and non-breast cancers in Li-Fraumeni syndrome patients.
• Validated the model's ability to capture complex triggering relationships in real-world data.

Conclusions:

• The proposed Cox-type structure provides a robust framework for analyzing multitype recurrent events.
• The developed statistical methods are effective and perform well in simulations and real data applications.
• This approach enhances understanding of event dependencies in lifetime data analysis, particularly in clinical settings.