A multiphase non-linear mixed effects model: An application to spirometry after lung transplantation.

Area of Science:

• Medical Sciences
• Biostatistics
• Longitudinal Data Analysis

Background:

• Longitudinal studies in medicine frequently exhibit complex, non-linear temporal relationships.
• The impact of risk factors can evolve significantly over the course of patient follow-up.

Purpose of the Study:

• To present a novel multiphase non-linear mixed effects model for analyzing longitudinal continuous measurements.
• To incorporate temporal decomposition for identifying distinct phases and associated risk factors.
• To demonstrate the model's utility in handling complex temporal patterns and time-varying coefficients.

Main Methods:

• Development of a multiphase non-linear mixed effects modeling system.
• Application of temporal decomposition techniques to segment longitudinal data into distinct phases.
• Utilizing readily available statistical software for model implementation and analysis.

Main Results:

• The proposed model effectively captures non-linear temporal trends in longitudinal data.
• Temporal decomposition successfully identified distinct phases within the follow-up period.
• Risk factor influences were shown to vary across identified phases, demonstrating time-varying effects.

Conclusions:

• The multiphase non-linear mixed effects model offers a flexible and powerful approach for analyzing complex longitudinal medical data.
• The model's ability to identify phases and time-varying coefficients enhances the understanding of disease progression and treatment effects.
• This methodology is particularly useful for analyzing data such as post-transplant spirometry, where complex temporal dynamics are expected.