An in-development R package and a Bayesian hierarchical model jointly fitting multiple "local" wastewater data streams and "global" case count data to produce nowcasts and forecasts of both observations
We currently specify the unobserved time as an input parameter. This is by default 50. However, the length of our infection to hospital admissions delay pmf is 55. This means that on day 1 of the observed time, we are missing potential hospital admissions from 55 to 50 days ago, since we don't have incident infections for those time points.
Instead, we should have the "unobserved time" just be a function of this delay length (and be greater than or equal to the delay length). That way we are not missing any input hospital admissions.
Goal
We currently specify the unobserved time as an input parameter. This is by default 50. However, the length of our infection to hospital admissions delay pmf is 55. This means that on day 1 of the observed time, we are missing potential hospital admissions from 55 to 50 days ago, since we don't have incident infections for those time points.
Instead, we should have the "unobserved time" just be a function of this delay length (and be greater than or equal to the delay length). That way we are not missing any input hospital admissions.