Projects

Project 4: Temporal Association Analysis Using Functional Models for Sparse Longitudinal Microbiome Data

In this project, we are introducing functional linear modeling tool to analyze the sparse longitudinal microbiome data. The goal of our analysis is to reveal the temporal association effects between functional predictors and functional/scalar outcomes. We use the functional PCA-based method to recover the trajectory of the microbiome data and fit the penalized functional model to estimate the temporal effects of microbiome on outcome of interest. In a simulation study, we evaluated the performance of the functional models for different proportions of the missing data (caused by irregular time) and different lengths of the time. Furthermore, we analyzed a real longitudinal data set to investigate the effect of gut microbiome on the faecal calprotectin.