ORDINAL PROBIT FUNCTIONAL OUTCOME REGRESSION WITH APPLICATION TO COMPUTER-USE BEHAVIOR IN RHESUS MONKEYS

Area of Science:

• Statistics
• Biostatistics
• Functional Data Analysis

Background:

• Functional regression models have advanced for non-Gaussian outcomes.
• Ordinal functional outcomes remain under-explored in statistical research.
• Existing methods may not fully capture complex ordinal functional data patterns.

Purpose of the Study:

• Introduce the Ordinal Probit Functional Outcome Regression (OPFOR) model.
• Address limitations in analyzing ordinal functional outcomes.
• Apply the model to computer-use behavior in rhesus macaques.

Main Methods:

• Developed the OPFOR model, adaptable with various basis functions (penalized B-splines, wavelets, O'Sullivan splines).
• Conducted simulations to assess model performance under diverse covariance structures.
• Utilized Bayesian model selection criteria for functional outcome regression.
• Compared OPFOR with a cumulative-link mixed-effects model.

Main Results:

• OPFOR demonstrated reasonable estimation performance across multiple basis functions.
• Near nominal coverage was achieved for joint credible intervals in simulations.
• OPFOR outperformed the cumulative-link mixed-effects model in simulation studies.
• The model successfully characterized demographic factors influencing macaque computer use.

Conclusions:

• The OPFOR model provides a robust framework for analyzing ordinal functional outcomes.
• OPFOR offers more nuanced insights compared to standard ordinal longitudinal analyses.
• The model is effective in both simulation and application settings, particularly for complex biological data.