using nlparci with weighted error function

Hello.

I have the same problem than: https://fr.mathworks.com/matlabcentral/answers/265273-how-to-correctly-use-lsqnonlin-in-conjunction-with-nlparci-and-nlpredci-by-fitting-to-a-weighted-fu

As this post is really old and not really answered,I ask the question again.

How to calculate parameters confidence interval with for exemple nlparci when the function is weighted. In general case one can write the error to minimize as:

w is the wheight, y the data, f(x) the function to approximate the data.

In my case I use a relative error:

I used the lsqnonlin function and fitted during 5 days since my function is slow and since I have 7 parameters. I recorded the output of lsqnonlin