Thread by @teddyhla

NEXT:Interfacing ICU data Nicolas Bennet
- Nicolas was very pleased to hear Chris Sauer(earlier speaker) advocating use of 2 -data set at least
@ESICM #criticalcare #ai #ml #icudatasets

@ESICM eth-mds.github.io/ricu/
showcasing this R dataset

starting with
'r'
> library(ricu)
> lact (loaded all lactate data)
'r'

@ESICM NEVER THOUGHT THAT I would be tweeting code in an ICU conference :)) @ESICM #datascience :P #datascience #LIVES2022

@ESICM '''r
> mimic_demo
#display 25 tables
'''
>mimic_demo$chartevents
#provides metadata
Ricu package handles conversion for you.

@ESICM Uses non standard evaluation for row-subsetting, defaulte values for ID and index columns and interval.

To concisely extract data, uses own "DATA CONCEPTS".
e.g.,
> load_concepts(c("hr","glu"),c("mimic_demo",verbose = FALSE)

So more complex concepts e.g., heart rate - stored as JSON object
unit -
min -
max values
- description
-category etc
(imho - this is a neater version than label function which is more appealing to STATA/SPSS )

Concept dictionary
e.g.,
dict <- load_dictionary()
length(dict)

this two line of code gives an example of what dictionary values are used!

HAVING A CONCEPT DICTIONARY !! WOW :) its like basic
Apologises for R centric view :P

its involved. but line 3 selects insulin values for hour 0 and 24.

and line 18, returns a value of insulin value you set.

Next diabetes data, lets create a new concept
uses icd 9 codes as hook

Output

We are not done
- has worked well for us
- available via Github. / CRAN
- would have done many things differently in retrospect
- spot a bug and open an issue on github.

(THIS IS ON MY CHECKOUT LIST DEFINITELY) :P
@ESICM #LIVES2022
#datascience #ai #ml #icudata

@ESICM eth-mds.github.io/ricu/articles/…

@ESICM @threadreaderapp unroll

@ESICM obviously a mention to @hadleywickham wont be complete when talking about learning R and R package :P