score_auc.R

score_auc <- function(
  predictions,
  truth
){
  if (!"patientid" %in% names(predictions)) stop("patientid missing from predictions")
  if (!"prediction" %in% names(predictions)) stop("prediction missing from predictions")
  
  if (!"patientid" %in% names(truth)) stop("patientid missing from truth")
  if (!"has_long_covid_diag" %in% names(truth)) stop("has_long_covid_diag missing from truth")
  
  if (sum(is.na(predictions)) != 0) stop("Missing data detected in predictions")
  
  if (min(predictions$prediction) < 0) stop("Predictions can't be less than 0")
  if (max(predictions$prediction) > 1) stop("Predictions can't be more than 1")
  
  merged_data <- truth %>%
    dplyr::select(patientid,has_long_covid_diag) %>%
    dplyr::mutate(
      has_long_covid_diag = factor(has_long_covid_diag, levels = c("1","0"))
      ) %>%
    dplyr::left_join(
      predictions %>%
        dplyr::select(patientid, prediction),
      by = "patientid"
    )
  

  yardstick::roc_auc(
    merged_data,
    has_long_covid_diag,
    prediction,
    estimator = "binary"
    
    )  
}