Functional modeling of recurrent events on time-to-event processes

Area of Science:

• Biostatistics
• Survival Analysis
• Functional Data Analysis

Background:

• Modeling time-to-event outcomes often involves recurrent events.
• Incorporating dynamic, time-varying features into survival models is crucial for accurate predictions.
• Existing methods may not fully capture the complexity of time-varying processes influencing patient outcomes.

Purpose of the Study:

• To develop an innovative methodology for modeling survival data with time-varying covariates.
• To quantify the association between dynamic processes and time-to-event outcomes.
• To provide new insights into personalized treatment strategies by analyzing patient behaviors and therapeutic patterns.

Main Methods:

• Utilized functional data to represent time-varying variables.
• Modeled time-varying variables as compensators of marked point processes.
• Applied Functional Principal Component Analysis (FPCA) for dimensionality reduction.
• Employed a Cox-type functional regression model for overall survival analysis.

Main Results:

• Successfully modeled self-exciting behaviors in recurrent events.
• Quantified the influence of time-varying Heart Failure (HF) hospitalizations and drug consumption on patient survival.
• Demonstrated the utility of functional data in survival analysis for complex patient data.

Conclusions:

• The proposed methodology effectively incorporates time-varying covariates into survival models.
• This approach offers a novel way to understand the impact of dynamic patient factors on survival.
• Findings provide a foundation for more personalized treatment strategies in chronic diseases like Heart Failure.