Analyzing Zero-Truncated Recurrent Events by Stratified Regression with Time-Varying Coefficients

2025-07-30

AI Summary

This paper addresses the challenge of modeling event dependence on historical records in zero-truncated recurrent event data—where individuals experiencing no events prior to enrollment remain unobserved. We propose a hierarchical Cox regression model that integrates external census information. Methodologically, we innovatively incorporate time-varying coefficients and random effects to capture both individual heterogeneity and dynamic event dependence. To accommodate zero truncation, we develop a tailored partial likelihood estimation procedure and establish its asymptotic normality and finite-sample robustness. Simulation studies confirm the estimator’s accuracy and favorable statistical properties. Applied to pediatric mental health care (PMHC) data, our approach yields substantially improved predictive performance and enhanced causal interpretability compared to conventional methods. The framework thus bridges theoretical rigor with practical applicability for analyzing truncated recurrent event data in population health research.

Technology Category

• Reasoning under Uncertainty: Causality
• Machine Learning: Causal Learning
• Cognitive Modeling & Cognitive Systems: Other Foundations of Cognitive Modeling & Systems

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Web Mining and Content Analysis: Models for Web evolution

Abstract

This paper presents a strategy for analyzing zero-truncated recurrent events data. Motivated by a pediatric mental health care (PMHC) program, we are particularly concerned with how the event occurrence depends on the occurrences in the past. We consider a stratified Cox regression model with time-varying coefficients and propose a procedure for estimating the model parameters using the zero-truncated data integrated with population census information. We evaluate the finite-sample performance of the proposed estimator through simulation and establish its asymptotic properties. Data from the PMHC program are used throughout the paper to motivate and to illustrate the proposed approach.

Problem

• Analyzing zero-truncated recurrent events data
• Modeling event dependence on past occurrences
• Estimating stratified Cox regression with time-varying coefficients

Innovation

• Stratified Cox regression with time-varying coefficients
• Estimating model using zero-truncated data
• Integrating population census information