Piecewise Linear Mixed Models With a Random Change Point

[SJKkeene724]

Hello Bristol University,

I have no idea what to do here.

I have a cohort of patients with repeated measurements of BMI (kg/m^2) and smoking (current, former, never) across 10 years (1 measurement per year). I would like to start with just current smokers at baseline and see how the slope of their BMI (dependent variable) changes across time. I would also like to look at the slope in those patients who quit and become former smokers during follow-up. However, the time that this happens varies by patient and may not happen at all in some patients. Thus, a margin plot might have a single slope at first but then a branching off point representing when the BMI slopes separate between those who continue smoking and those who quit. I believe this is called Piecewise Linear Mixed Models With a Random Change Point (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5625953/)

I would like to stay away from Bayesian statistics for now if possible but I can try to learn. So, is there commands that can support this topic. I am sorry that this is so broad but I am in the beginning stages and have no idea what to do.

Thank you,
Spencer

[samar999]

help
i want a method to do random change point with mixed model without bayesian approach