Converting medical device XML data to data frame in R

Question

I have XML data generated by a medical device that I am trying to parse and I cannot for the life of me figure it out. Below is a sample of the XML data:



  BLANK
  LP15
  false
  2020-04-18T20:57:55.00000
  2L355RRB0200UR
  PT1S
  
    
      96.3298058454348
      9.34405124270701
      99.8261838893409
      6.51749943203332
      0
      1277.72862806776
      0
      1298.04384154134
      0
      1339.14938235198
      1339.14938235198

I am looking to generate two data frames from for and .

I am using the XML package to parse the data, and here is what I have done so far:

df <- xmlParse(file = "file.xml", useInternalNodes = TRUE)

df_1 <- xmlToDataFrame(df, nodes = getNodeSet(df, "//CPRStatistics"))

And this is where I get stuck...

user_12500315 · Answer

Usually, package xml2 is easier to use, though package XML will at least have a go at making a data frame where this is sensible.

The node set obtained from is easier to make into a data frame:

library(XML)
df <- xmlParse(file = "file.xml", useInternalNodes = TRUE)
overall <- xmlToDataFrame(df, nodes = getNodeSet(df, "//OverallStatistic"))

There is only a single row in this data frame:

overall
#>   AverageCompressionRatePerMin AverageVentilationRatePerMin
#> 1             96.3298058454348             9.34405124270701
#>   MedianCompressionRatePerMin MedianVentilationRatePerMin
#> 1            99.8261838893409            6.51749943203332
#>   TotalTimeCompressionsDuringPromptedCprSec TotalTimeCompressionsDuringValidSec
#> 1                                         0                    1277.72862806776
#>   TotalTimeCprDuringPromptedCprSec TotalTimeCprDuringValidSec TotalTimePromptedCprSec
#> 1                                0           1298.04384154134                       0
#>   TotalTimeValidSec TotalTimeValidSecEx
#> 1  1339.14938235198    1339.14938235198

The second node, is more difficult to parse because the values are stored in attributes rather than as text nodes. That requires you to find all the nodes, strip all their attributes into a list and rbind them together into a data frame. In this case, because the data frame has so many columns, I have converted it into a tibble instead of a data frame just for ease of printing to screen:

dplyr::as_tibble(do.call(rbind, lapply(getNodeSet(df, "//CPRStatisticsItem")[-1], xmlAttrs)))
#> # A tibble: 4 x 19
#>   Interval ResponsibleForC~ CPRStartTime CPREndTime CPRDurationSec PauseDurationSec
#>                                                      
#> 1 2        ""               2020-04-18T~ 2020-04-1~ 90.321         12.337          
#> 2 3        ""               2020-04-18T~ 2020-04-1~ 239.785        18.89           
#> 3 4        ""               2020-04-18T~ 2020-04-1~ 764.77         9.322           
#> 4 5        ""               2020-04-18T~ 2020-04-1~ 203.169        0.4910000000000~
#> # ... with 13 more variables: ReasonForPause , IntervalComments ,
#> #   AverageCompressionRatePerMin , AverageVentilationRatePerMin ,
#> #   MedianCompressionRatePerMin , MedianVentilationRatePerMin ,
#> #   TotalTimeCompressionsDuringPromptedCprSec ,
#> #   TotalTimeCompressionsDuringValidSec ,
#> #   TotalTimeCprDuringPromptedCprSec , TotalTimeCprDuringValidSec ,
#> #   TotalTimePromptedCprSec , TotalTimeValidSec , TotalTimeValidSecEx

Converting medical device XML data to data frame in R

1 Answers