latent states as regression parameters

[hivehat]

Hi,

thanks so much for this amazing toolbox! I was wondering whether/how its possible to use the CountLDS class such that the latent states are the parameters of the regression model i.e. such that

x_t = A * x_t-1 + e
y_t = Bern(inputs * x_t)

where inputs are exogenous pre-determined inputs. I've had a look through the examples and classes but haven't been able to make sense of it.