Create 00_api_queries.R

code/00_api_queries.R

@@ -0,0 +1,108 @@
+#' @name 00_api_queries.R
+#' @author Tim Fraser
+#' @description 
+#' This is a script one of the project teams and I cooked up together.
+#' I include it here as an example of how you could use an API to gather data.
+#' We're using the FDA's medical devices database 
+#' to query data about Medtronic Pacemakers.
+
+# Load packages
+library(dplyr)
+library(readr)
+library(stringr)
+library(lubridate)
+library(httr) # sends http requests
+library(jsonlite) # work with json data
+
+# Write a simplication function to handle API results
+get_result = function(result){
+  result2 = rawToChar(result$content)
+  output = jsonlite::fromJSON(result2)
+  return(output)
+}
+
+
+
+# CHUNK 1
+result1 = paste0(
+  "https://api.fda.gov/device/event.json?",
+  'search=device.model_number.exact:"ADDRL1"',
+  "&skip=0",
+  "&limit=1000"
+) %>%
+  GET() %>%
+  get_result()
+
+# CHUNK 1
+result2 = paste0(
+  "https://api.fda.gov/device/event.json?",
+  'search=device.model_number.exact:"ADDRL1"',
+  "&skip=1000",
+  "&limit=1000"
+) %>%
+  GET() %>%
+  get_result()
+
+# CHUNK 1
+result3 = paste0(
+  "https://api.fda.gov/device/event.json?",
+  'search=device.model_number.exact:"ADDRL1"',
+  "&skip=2000",
+  "&limit=1000"
+) %>%
+  GET() %>%
+  get_result()
+
+# Bundle results into a data.frame
+output = bind_rows(
+  result1$results,
+  result2$results,
+  result3$results
+)
+
+# Investigate variables
+output %>%
+  names()
+
+# tibble(table = output %>% names() ) %>%
+#   filter(str_detect(table, "problem"))
+
+# output %>%
+#   select(contains("date")) %>%
+#   glimpse()
+
+output2 = output %>%
+  select(product_problems, 
+         patient,
+         device_date_of_manufacturer,
+         date_of_event)
+
+output2$device_date_of_manufacturer %>% class()
+
+output2 %>%
+  filter(!is.na(device_date_of_manufacturer), !is.na(date_of_event))
+
+# Handle dates and calculate time intervals
+output2 %>%
+  filter(!is.na(device_date_of_manufacturer), !is.na(date_of_event)) %>%
+  mutate(
+    device_date_of_manufacturer = lubridate::make_date(
+      year = str_sub(device_date_of_manufacturer, 1,4),
+      month = str_sub(device_date_of_manufacturer, 5,6),
+      day = str_sub(device_date_of_manufacturer, 7,8)
+    ),
+    date_of_event = lubridate::make_date(
+      year = str_sub(date_of_event, 1,4),
+      month = str_sub(date_of_event, 5,6),
+      day = str_sub(date_of_event, 7,8)
+    )
+  ) %>%
+  mutate(days = as.numeric(difftime(date_of_event, device_date_of_manufacturer, units = "days")),
+         hours = as.numeric(difftime(date_of_event, device_date_of_manufacturer, units = "hours"))) %>%
+  write_csv("dataset.csv")
+
+# Cleanup!
+rm(list())
+
+
+