The Haemoblibin profile is slightly non-linear. However, I'm not allowed to fit a non-linear model. I can either transform the variable or ignore the days/ parts that are non-linear (you can advise or we can speak further about this
statistics
Description
1. For sampling: Hint: change data to the wide form, sample 75% of patients by pid then change back data to long format then work with that)
2. The Haemoblibin profile is slightly non-linear. However, I'm not allowed to fit a non-linear model. I can either transform the variable or ignore the days/ parts that are non-linear (you can advise or we can speak further about this)
3. We can also bring the severity of the disease at baseline i.e. parasite density at baseline (note that the parasite density change over time, so you just want it to be the predictions)
4. Of course, we are going to treat patients as random effects, so are you going to say these patients are nested between sites or not. It depends on whether you want to bring site as a fixed effect to compare sites with one another (you can also advise here).
Please advise on the procedure and how soon can it be done.
