Summarizing data through a piecewise linear growth curve model.

Area of Science:

• Biostatistics
• Clinical Trials Methodology
• Longitudinal Data Analysis

Background:

• Longitudinal designs, involving repeated measurements, are crucial in clinical trials.
• Advanced statistical methods and software now accommodate complex covariance structures for longitudinal data analysis.
• Hathaway et al. previously proposed a fixed four-summary-statistic linear growth curve model for longitudinal data.

Purpose of the Study:

• To develop a procedure for analyzing longitudinal data using a piecewise linear growth curve model.
• To evaluate the proposed model's performance under various covariance structures using Leprosy data.
• To assess the robustness and goodness of fit of the new method through simulation studies.

Main Methods:

• Fitting a piecewise linear growth curve model to longitudinal data.
• Analysis conducted under different covariance structures (unstructured, compound symmetry, auto-regressive, random effects).
• Calculation of residual sum of squares and assessment of goodness of fit for various models.

Main Results:

• The piecewise linear growth curve model was fitted to Leprosy data.
• Residual sum of squares were computed across different covariance structures.
• Simulation studies indicated the proposed method's robustness and superior goodness of fit.

Conclusions:

• The developed piecewise linear growth curve model provides a robust approach for analyzing longitudinal data.
• The method demonstrates improved goodness of fit compared to existing models.
• This technique enhances the statistical and scientific power of analyzing repeated measurements in clinical research.