A multivariate discrete failure time model for the analysis of infant motor development

Area of Science:

• Biostatistics
• Developmental Pediatrics
• Statistical Modeling

Background:

• Infant motor development involves achieving key milestones like sitting, crawling, and walking.
• Analyzing the timing and interdependencies of these milestones is crucial for understanding child development.
• Existing models may not fully capture population heterogeneity or within-subject correlations.

Purpose of the Study:

• To develop and validate a multivariate discrete failure time model for joint analysis of infant motor milestones.
• To account for subject-specific heterogeneity and within-subject dependence in milestone achievement.
• To apply the model to real-world birth cohort data for insights into early development.

Main Methods:

• Development of a multivariate discrete failure time model.
• Inclusion of a subject-specific latent factor for heterogeneity and dependence.
• Flexible modeling of factor loadings, covariate effects, and unique at-risk intervals.
• Bayesian inference using a data-augmented Gibbs sampler.
• Simulation studies and application to the Nurture study cohort data.

Main Results:

• The developed model effectively analyzes the time to achievement of multiple motor milestones simultaneously.
• The model successfully accounts for individual variability and the correlation between milestone achievement times.
• Application to the Nurture study provided insights into factors influencing infant motor development.

Conclusions:

• The proposed multivariate discrete failure time model offers a robust framework for analyzing infant motor development.
• This approach enhances our understanding of developmental trajectories by considering individual differences and milestone interdependencies.
• The model is a valuable tool for researchers studying early childhood health and development.