Predictive functional ANOVA models for longitudinal analysis of mandibular shape changes.

Area of Science:

• Biostatistics
• Developmental Biology
• Medical Imaging

Background:

• Mandibular shape analysis aids in predicting growth, but shape changes are less understood than size changes.
• Functional data analysis is crucial for understanding developmental trajectories.
• Existing methods often lack the precision to detail specific shape alterations over time.

Purpose of the Study:

• To introduce a Bayesian functional ANOVA model for analyzing 2D human mandible shape changes.
• To identify specific developmental stages and locations of significant mandibular shape alterations.
• To provide a robust statistical framework for functional data in morphometric studies.

Main Methods:

• Development of a Bayesian functional ANOVA model using predictive process models for Gaussian process (GP) priors.
• Application to functional data of human mandible profiles from developmental stages.
• Utilizing graphical posterior summaries for interpretation of morphometric findings.

Main Results:

• The model effectively analyzes multivariate functional data with unequally spaced and differing observations.
• Identified specific temporal and spatial patterns of mandibular shape changes during development.
• Graphical summaries provided biologically interpretable insights into morphometric findings.

Conclusions:

• The proposed Bayesian model offers a significant methodological advancement for studying 2D mandibular shape changes.
• It provides a computationally appealing and interpretable approach compared to classical cephalometric analysis.
• The model enhances understanding of developmental biology through detailed shape analysis.