Skip to content

Commit

Permalink
Merge pull request MIT-LCP#458 from JackieMium/master
Browse files Browse the repository at this point in the history
 hadm_id written as icustay_id
  • Loading branch information
alistairewj authored Aug 13, 2018
2 parents 9e71e6b + e6abac6 commit ea168b5
Showing 1 changed file with 1 addition and 1 deletion.
2 changes: 1 addition & 1 deletion tutorials/cohort-selection.ipynb
Original file line number Diff line number Diff line change
Expand Up @@ -2242,7 +2242,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"Note how `icustay_id` = 100006 no longer has an entry for OMED above: this is because this service was given after their ICU admission, so we do not want to consider it. Also note that our join clause has `+ interval '12' hour` - this adds a bit of \"fuzziness\" to our criteria. As these times are entered asynchronously by different people in varying locations in the hospital, there can be some minor inconsistencies in the order. For example, a patient may be transferred to the surgical service for ICU admission, but the `transfertime` in *services* occurs after the `intime` in *icustays* by an hour or so. This is administrative \"noise\" - and a fuzzy interval can be useful in these cases. Again, this is a cohort selection decision - you may not want to use an interval as large as 12 hours - perhaps only 2 or 4 - though in this case there is likely to be very minor differences as 80% of patients never have a change in hospital service.\n",
"Note how `icustay_id` = 291788 no longer has an entry for OMED above: this is because this service was given after their ICU admission, so we do not want to consider it. Also note that our join clause has `+ interval '12' hour` - this adds a bit of \"fuzziness\" to our criteria. As these times are entered asynchronously by different people in varying locations in the hospital, there can be some minor inconsistencies in the order. For example, a patient may be transferred to the surgical service for ICU admission, but the `transfertime` in *services* occurs after the `intime` in *icustays* by an hour or so. This is administrative \"noise\" - and a fuzzy interval can be useful in these cases. Again, this is a cohort selection decision - you may not want to use an interval as large as 12 hours - perhaps only 2 or 4 - though in this case there is likely to be very minor differences as 80% of patients never have a change in hospital service.\n",
"\n",
"Finally, we want to collapse this down so we only have one service for a given ICU admission. As done earlier, we will use RANK() to do this."
]
Expand Down

0 comments on commit ea168b5

Please sign in to comment.