Building a model for discrete data


I have a question regarding a model with discrete response and continuous predictors.
I am trying NLME model for exposure and response. The raw data for the response is continuous, but the data I received are rounded off value, resulting in discrete response. But, predictors I am exploring are continuous, so predicted response is continuous. The problem is that the observed response is discrete, but the predicted is continuous, so the GOF is distorted. I was wondering if there is a way to solve this issue. I don’t think it’s a perfect way, but at least I tried the below using derived block. However, I encountered the error like no method matching. Could you please advise me on this issue? Thank you!

    response_raw         = @. Normal(mresp, sqrt(mresp^2 * σ²_prop + σ²_add))
    response_rounded      = @. round(response_raw; digits = 1)

How many levels do you have in your response, i.e. what is the range of values in your response?

They are like 0.1, 0.2, 0.3, …, 1.6 (a total of 13 levels). But the raw data might be something like 0.12, 0.24…


Is there any way to deal with this issue? Thank you so much!