PC program for growth prediction in the two-stage polynomial growth curve model

Area of Science:

• Biometrics
• Anthropometry
• Statistical Modeling

Background:

• Growth prediction is crucial in longitudinal studies.
• Polynomial growth curve models offer a framework for analyzing individual growth trajectories.
• Accurate prediction for new individuals requires robust statistical methods.

Purpose of the Study:

• To develop and present a computational program for growth prediction using a one-sample polynomial growth curve model.
• To estimate future measurements for a new individual based on their past data and a normative sample.
• To compare the predictive accuracy of the two-stage model with a less restrictive model.

Main Methods:

• Utilized a two-stage (random coefficients) one-sample polynomial growth curve model.
• Developed a PC program in GAUSS386i for computational implementation.
• Applied the method to mandibular ramus height measurements in rhesus monkeys.
• Compared results with a model having less restrictive assumptions on the covariance matrix.

Main Results:

• The GAUSS386i program facilitates growth prediction for new individuals.
• Prediction accuracies from the two-stage model and the less restrictive model were comparable.
• The study demonstrated the application using a real-world dataset of primate growth.

Conclusions:

• The developed program effectively addresses growth prediction challenges.
• Less restrictive models may offer comparable or superior predictive performance in certain scenarios.
• The findings suggest flexibility in model selection for growth curve analysis.