How to implement Multiphase Mixed-Effects Models for Repeated Measures Data?

2 次查看（过去 30 天）

rullator 2017-9-8

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/355699-how-to-implement-multiphase-mixed-effects-models-for-repeated-measures-data

编辑： rullator 2017-9-8

Hi,

I try to model longitudinal data, where a change process divides the data in several phases. So a single (non)linear regression will not fit my data.

In my example,

(1) I want a model to fit the data in three phases. So, the model could be approximate by a combination of three submodels, each associated with a distinct (individual) subinterval of time. Each phase could probably be estimated by a linear model.

(2) the change points, where each phase turns to another phase, are not known and need to be estimated individually.

(3) I'm interested in the difference of two groups: difference of linear model for each phase and difference in change points.

A very excellent description of the problem is given by Cudeck and Klebe 2002 here: https://www.ncbi.nlm.nih.gov/pubmed/11928890

Now, I would love to set up such models in Matlab. Has anyone any experience with multiphase models for repeated measures in Matlab and is willing to share his/her approach?