Hi, I am running a mediation which the mediator is zero-inflated poisson and the outcome variable is binary. The output summary is attached ('zip1' and 'zip2'). I am a bit confused about the two classes in the output. Here is my understanding. Please correct me if I make any misinterpretation.

variables: Predictor = disc_pfreq (count) ; mediator = disc_sfreq (zero-inflated poisson); outcome = letterv_occur (binary)

total n = 13740

From the output, there are 2 classes. Class 1 includes those whose observed mediator count is 0 (i.e.,13559 cases). Class 2 includes all obs cases (i.e., 13740 cases).
However, the zip model entails 2 parts: obs count = 0 vs. obs count > 0 (please see 'zip model' attached to the msg).
Does class 1 from the output match the first equation of the model (obs count = 0)?
Does class 2 from the output match the second equation of the model (obs count >0)? If so, then the obs number for class 2 should be 13740-13559 cases rather than 13740 cases.

In short, I am not sure how to interpret the zip (mediator) model in this case and how it is related to the outcome model.

So far, I can only find one paper on zip mediators by Zhigang Li et al. (2019). https://arxiv.org/abs/1906.09175v1

Any comment or help is highly appreciated!!!