QS: Initial Queries Computation

[jthompson-arcus]

clinsight/R/mod_queries.R

Lines 110 to 113 in 60507ac

	df <- with(r$query_data, r$query_data[n == 1, ] )
	if(identical(nrow(df),0)) return(df)
	df <- df |>
	dplyr::slice_min(timestamp, by = c(subject_id, event_label, query_id)) |>

@LDSamson handful of questions here:

1. Why are we grabbing queries with n == 1? In what scenario would there be query data without an initial query of n == 1? I ask this in part because our EDC queries have an audit ID which is sequential across all items. Does n signify anything outside marking subsequent query records for a query_id?
2. Why slice by minimum timestamp instead of n? This seems to imply that there can be multiple queries with n==1. Is that somehow the case?

I would think that we should be able to slice the data frame at minimum n by query_id and call it a day.

[LDSamson]

When a query is created, it always gets n=1 assigned. the purpose of n is to be able to show a conversation chain (in the form of a chat). It just increments with each written follow-up message. Since it is always 1 for the first message, I used this simple logic, but slicing using minimum n by query id should also work.

Regarding the timestamp slice, I am not sure, I need to check it. It might be related to my idea of introducing an edit option for the queries in the future, since in the original database design any edit will just be a new row but with a newer timestamp. But then it should be a slice_max instead here I think...

[jthompson-arcus]

So in the bit of code I snipped above, we really don't need to grab n == 1, we can just grab the first n. Correct? I would be very useful to us if this n could correspond to an audit ID.

[LDSamson]

Yes that is also possible, as long as you select the first n for each unique ID, as I mentioned in my previous message. Do I understand it correctly that you have two different IDs, an audit ID and the unique query ID?