Need to Generalize add_timevars_to_data()

[jthompson-arcus]

clinsight/R/fct_data_helpers.R

Lines 116 to 159 in 0eec2fd

	add_timevars_to_data <- function(
	data
	){
	stopifnot("[data] should be a data frame" = is.data.frame(data))
	missing_new_cols <- required_col_names[!required_col_names %in% names(data)] |>
	paste0(collapse = ", ")
	if(nchar(missing_new_cols) > 0) stop(
	paste0("The following columns are missing while they are required:\n",
	missing_new_cols, ".")
	)
	df <- data |>
	dplyr::mutate(
	edit_date_time = as.POSIXct(edit_date_time, tz = "UTC"),
	event_date = as.Date(event_date),
	day = event_date - min(event_date, na.rm = TRUE),
	vis_day = ifelse(event_id %in% c("SCR", "VIS", "VISEXT", "VISVAR", "FU1", "FU2"), day, NA),
	vis_num = as.numeric(factor(vis_day))-1,
	event_name = dplyr::case_when(
	event_id == "SCR" ~ "Screening",
	event_id %in% c("VIS", "VISEXT", "VISVAR") ~ paste0("Visit ", vis_num),
	grepl("^FU[[:digit:]]+", event_id) ~ paste0("Visit ", vis_num, "(FU)"),
	event_id == "UN" ~ paste0("Unscheduled visit ", event_repeat),
	event_id == "EOT" ~ "EoT",
	event_id == "EXIT" ~ "Exit",
	form_id %in% c("AE", "CM", "CP", "MH", "MH", "MHTR", "PR", "ST", "CMTR", "CMHMA") ~ "Any visit",
	TRUE ~ paste0("Other (", event_name, ")")
	),
	event_label = dplyr::case_when(
	!is.na(vis_num) ~ paste0("V", vis_num),
	event_id == "UN" ~ paste0("UV", event_repeat),
	TRUE ~ event_name
	),
	.by = subject_id
	) |>
	dplyr::arrange(
	factor(site_code, levels = order_string(site_code)),
	factor(subject_id, levels = order_string(subject_id))
	)
	if(any(grepl("^Other ", df$event_name))) warning(
	"Undefined Events detected. Please verify data before proceeding."
	)
	df
	}

We need to find some way to generalize this function and remove all the hard coding. I don't know if this means we need another sheet/item in the metadata or not.

[jthompson-arcus]

@LDSamson what is your process for creating metadata$events?

I think I'm most interested in how you know how many visits to create for a study since the code above is imputing the visit number for each subject.

[LDSamson]

The length of metadata$events is currently not data-driven but just defined based on the study protocol. The reason for that is so that we can show a consistent small visit timeline in the header widgets, even if there is not data yet:

I think that is the only place where the tab metadata$events is used. For all other figures, the events are created data-driven (the code that you show in your post above).

Did that answer your question?