Hi,

For vancomycin inputevents_mv row_id=429 gives starttime at 8 and endtime at 8:01. How to determine actual administration duration? In row_id=462 amount is 500. What does it mean? Perhaps this came from half 1g frozen bag or 500mg vial? For row_id=429 the amount=1 dose which is quite ambiguous as well. In both cases rate=0.

Thanks in advance for any suggestions.

L

Prerequisites

• [ X] Put an X between the brackets on this line if you have done all of the following:
  • Checked the online documentation: https://mimic.physionet.org/about/mimic/
  • Checked that your issue isn't already addressed: https://github.com/MIT-LCP/mimic-code/issues?utf8=%E2%9C%93&q=

When I run Postgres_load_data script, first three table are loaded and after that I got message: could not stat file CHARTEVENTS.csv: unknown error. Is anyone has this situation and can help.

Hi !

I'm trying to load the MIMIC-III data into a local postgres database by following instructions from this link: https://mimic.physionet.org/tutorials/install-mimic-locally-ubuntu/

So far despite leaving it overnight, it consistently hangs at this stage:

$ psql -f postgres_load_data.sql -U mimic -v mimic_data_dir='/Documents/MIMIC_III/' SET

COPY 58976

COPY 34499

COPY 7567

Here's the configuration of my machine: MacBook Air (13-inch, Early 2014) Processor: 1.7 GHz Intel Core i7 Memory: 8 GB 1600 MHz DDR3

How long should it take to load this data on a machine with my configuration tentatively? The website states it might take several hours, but I didn't find any explicit benchmarking information.

Should I attempt to load this local instance using a machine with more RAM available?

I look forward to your guidance on this. Thanks!

With the addition of text data (a fix in MIMIC-III v1.4), the partitioning of CHARTEVENTS is no longer effectively distributing the data.

with drb_stats as (
select 1 as bucket, 1 as itemid_min , 210 as itemid_max UNION
select 2 as bucket, 210 as itemid_min     , 250 as itemid_max UNION
select 3 as bucket, 250 as itemid_min     , 614 as itemid_max UNION
select 4 as bucket, 614 as itemid_min     , 640 as itemid_max UNION
select 5 as bucket, 640 as itemid_min     , 742 as itemid_max UNION
select 6 as bucket, 742 as itemid_min     , 1800 as itemid_max UNION
select 7 as bucket, 1800 as itemid_min    , 2700 as itemid_max UNION
select 8 as bucket, 2700 as itemid_min    , 3700 as itemid_max UNION
select 9 as bucket, 3700 as itemid_min    , 4700 as itemid_max UNION
select 10 as bucket, 4700 as itemid_min    , 6000 as itemid_max UNION
select 11 as bucket, 6000 as itemid_min    , 7000 as itemid_max UNION
select 12 as bucket, 7000 as itemid_min    , 8000 as itemid_max UNION
select 13 as bucket, 8000 as itemid_min    , 220074 as itemid_max UNION
select 14 as bucket, 220074 as itemid_min  , 323769 as itemid_max
),
histogram as (
select bucket, itemid_min, itemid_max, count(*) as freq
from drb_stats drb
left join chartevents ce
on ce.itemid >= drb.itemid_min and ce.itemid < drb.itemid_max
group by bucket, itemid_min, itemid_max
order by bucket, itemid_min, itemid_max
)
select bucket, itemid_min, itemid_max,
repeat('*', (freq::float / max(freq) over() * 50)::int) as bar
from histogram;

 bucket | itemid_min | itemid_max |                        bar                         
--------+------------+------------+----------------------------------------------------
      1 |          1 |        210 | *******************
      2 |        210 |        250 | *******
      3 |        250 |        614 | *******************
      4 |        614 |        640 | *****
      5 |        640 |        742 | *********
      6 |        742 |       1800 | **************
      7 |       1800 |       2700 | 
      8 |       2700 |       3700 | *****************
      9 |       3700 |       4700 | *
     10 |       4700 |       6000 | *****
     11 |       6000 |       7000 | 
     12 |       7000 |       8000 | 
     13 |       8000 |     220074 | ********************
     14 |     220074 |     323769 | **************************************************

Essentially, the addition of the text data added ~100 million rows to ITEMIDs > 220000. I've noticed queries are going slower and I'm sure users will start to notice this too. Perhaps we should take this opportunity move to a smarter partitioning strategy - one based on the type of data stored with the ITEMID (e.g. the Metavision text data goes into a few partitions of its own, vital sign data goes into its own partition, etc). I'd welcome thoughts from anyone who has any - @parisni @tompollard

First of all: thanks for the great project !

Is there a plan to release a database demo for the MIMIC-IV as well? There is already one for the MIMIC-III.

Thanks !

Apologies if this is not the right place to ask this question.

The MIMIC III Documentation points to "A separate, complementary resource named the “MIMIC-III Waveform Database” contains high resolution waveforms and numerics..." https://mimic.physionet.org/mimicdata/waveforms/

I have been able to find the MIMIC II, version 3 waveform DB. https://www.physionet.org/physiobank/database/mimic2wdb/

After searching extensively for a MIMIC III waveform database, I am not having any luck. Does anyone know if this exists yet? Or is the most recent publicly accessible version of waveform data the MIMIC II, version 3 waveform DB?

Thank you!

Hi, I have downloaded and build the MIMIC-III clinical database. I am just wondering how to build the MIMIC-III waveform database as there is no information available on the website. Is there a way I can build it just like the way clinical database is being set up. ?

Thanks

Dear all, I have some doubts regarding the codes that should be used to analyse the urine output of the patient. I am interested in study patients with kidney injuries and I need to verify the level of the urine output volume. I need in the mL/kg/hr units. Can I use 51108/51109 from LABEVENTS or should be use another code from IOEVENTS? Another question, what is the difference between the codes 51108 and 51109? They seem equal, in the value, date and units. If we have to use the LABEVENTS code, how can we have the urine output in mL/kg/hr? To have in mL/kg it is just needed to divide by the weight of the patient, but what is the sample time rate of the measurements from 51108 and 51109? Thank you! Sincerely, Vanessa Cunha

I got an error loading postgres_load_data_pg10.sql.

I got COPY 58976 COPY 34499 COPY 7567 COPY 1048575 COPY 573146 COPY 1048575 COPY 651047 COPY 125557 COPY 134

then

ERROR: duplicate key value violates unique constraint "d_icd_diag_code_unique" DETAIL: Key (icd9_code) = (844) already exists. CONTEXT: COPY d_icd_diagnoses,line 307

ERROR loading postgres_load_data_pg10.sql.docx

How can I fix it please?

I'm not terribly experienced in building databases. I'm running on mac os X 10.10.5 I attempted to download the files using make mimic-download but got the following error:

-- Downloading MIMIC-III from PhysioNet --

wget --user --ask-password -P -A csv.gz -m -p -E -k -K -np -nd "https://physionet.org/works/MIMICIIIClinicalDatabase/files/" --2017-07-19 13:32:39-- http://csv.gz/ Resolving csv.gz (csv.gz)... failed: nodename nor servname provided, or not known. wget: unable to resolve host address ‘csv.gz’ --2017-07-19 13:32:39-- https://physionet.org/works/MIMICIIIClinicalDatabase/files/ Resolving physionet.org (physionet.org)... 128.30.30.88 Connecting to physionet.org (physionet.org)|128.30.30.88|:443... connected. ERROR: cannot verify physionet.org's certificate, issued by ‘CN=Let's Encrypt Authority X3,O=Let's Encrypt,C=US’: Unable to locally verify the issuer's authority. To connect to physionet.org insecurely, use `--no-check-certificate'. Converted links in 0 files in 0 seconds. make[1]: *** [mimic-download] Error 4 make: *** [mimic-download] Error 2

I then manually downloaded the data files and decompressed them. I then downloaded and installed PostgreSQL. I then tried to run: make mimic datadir=/path to data/

from the command line and had issues with the mimic/postgres password. I modified the Makefile script to change the user to postgres so that I could use the password I specified on install. I re-ran: make mimic datadir=/path to data/

and am now getting the following error:

psql "dbname=mimic user=postgres options=--search_path=mimiciii" -f postgres_create_tables.sql psql: FATAL: database "mimic" does not exist make[1]: *** [mimic-build] Error 2 make: *** [mimic-build] Error 2

I'd love some help with this. Thanks,

The current naming convention of calling the .csv files things like ADMISSIONS_DATA_TABLE.csv seems redundant, and also causes extra work to edit the output of my heuristic structure finder (https://github.com/mitmedg/csv2mysql), which thus generates scripts like CREATE TABLE ADMISSIONS_DATA_TABLE ...

Why not just change them all by eliminating _DATA_TABLE from the name of the file?

The measure

• [X] Put an X between the brackets on this line if you have done all of the following:
  • Checked the online documentation: https://mimic.mit.edu/
  • Checked that your issue isn't already addressed: https://github.com/MIT-LCP/mimic-code/issues?utf8=%E2%9C%93&q=

Description

This is a clinical knowledge. I am doing analysis on MIMIC-IV(2.0) dataset. What I have done is that

SELECT * FROM mimiciv_hosp.d_labitems where label like '%lucose%' and fluid='Blood'

The result is

50809	"Glucose"	"Blood"	"Blood Gas"
50931	"Glucose"	"Blood"	"Chemistry"
52027	"Glucose, Whole Blood"	"Blood"	"Blood Gas"
52569	"Glucose"	"Blood"	"Chemistry"

Then, I run

SELECT * FROM mimiciv_hosp.labevents WHERE ITEMID=50809 LIMIT 1000

I got the glucose value successfully. Now, here is a question. How do you GROUPING the glucose value? There are several way to do blood glucose test(e.g. A1C Test, Fasting Blood Sugar Test, Glucose Tolerance Test, Random Blood Sugar Test ). For different tests, we can see whether the patients get diabetes according to the following tabe. | Result* | A1C Test | Fasting Blood Sugar Test | Glucose Tolerance Test | Random Blood Sugar Test | | :--- | :--- | :--- | :--- | :--- | | Diabetes | 6.5% or above | 126mg//dL or above | 200mg//dL or above | 200mg//dL or above | | Prediabetes | 5.7-6.4% | 100-125mg//dL | 140-199mg//dL | N//A | | Normal | Below 5.7% | 99mg//dL or below | 140mg//dL or below | N//A |

Of course, we can obtain information from the Diagnoses related forms (ICD-9, ICD-10). I am now concerned about Glucose, and I hope to understand the test method and grouping method of this indicator (as shown in the table). For GLUCOSE, what interval is normal, what interval is prediabetes, and what interval is diabetes. This test method and grouping method for my research, look forward to your reply.

Dear MIMIC-Community,

I noticed that the presence of CRRT is missing as a parameter for KDIGO stage calculation. KDIGO states that "presence of CRRT" automatically classifies the patient as AKI stage 3.

I allowed myself to quickly write up a query which I believe would fix the problem.

I am so grateful to be able to contribute to this amazing project and I am looking forward to your input.

Kind regards Christian

Prerequisites

• [x] Put an X between the brackets on this line if you have done all of the following:
  • Checked the online documentation: https://mimic.mit.edu/
  • Checked that your issue isn't already addressed: https://github.com/MIT-LCP/mimic-code/issues?utf8=%E2%9C%93&q= Hi, I find it hard to determine wether someone has non invasive ventilation (NIV) or it that patient has invasive ventilation through an endotracheal tube. Looking at ventilation.sql in mimiciv-derived gives some hints about that, but it still seems fuzzy to me.

Ventilation.sql after defining thracheostomy ventilation mode, defines ventilation_status as InvasiveVent

WHEN o2_delivery_device_1 IN
(
    'Endotracheal tube'
)
OR ventilator_mode IN
(
    '(S) CMV',
    'APRV',
    'APRV/Biphasic+ApnPress',
    'APRV/Biphasic+ApnVol',
    'APV (cmv)',
    'Ambient',
    'Apnea Ventilation',
    'CMV',
    'CMV/ASSIST',
    'CMV/ASSIST/AutoFlow',
    'CMV/AutoFlow',
    'CPAP/PPS',
    'CPAP/PSV+Apn TCPL',
    'CPAP/PSV+ApnPres',
    'CPAP/PSV+ApnVol',
    'MMV',
    'MMV/AutoFlow',
    'MMV/PSV',
    'MMV/PSV/AutoFlow',
    'P-CMV',
    'PCV+',
    'PCV+/PSV',
    'PCV+Assist',
    'PRES/AC',
    'PRVC/AC',
    'PRVC/SIMV',
    'PSV/SBT',
    'SIMV',
    'SIMV/AutoFlow',
    'SIMV/PRES',
    'SIMV/PSV',
    'SIMV/PSV/AutoFlow',
    'SIMV/VOL',
    'SYNCHRON MASTER',
    'SYNCHRON SLAVE',
    'VOL/AC'
)
OR ventilator_mode_hamilton IN
(
    'APRV',
    'APV (cmv)',
    'Ambient',
    '(S) CMV',
    'P-CMV',
    'SIMV',
    'APV (simv)',
    'P-SIMV',
    'VS',
    'ASV'
)

and NonInvasiveVent by:

WHEN o2_delivery_device_1 IN
(
    'Bipap mask ', -- 8997 observations
    'CPAP mask ' -- 5568 observations
)
OR ventilator_mode_hamilton IN
(
    'DuoPaP',
    'NIV',
    'NIV-ST'
)
    THEN 'NonInvasiveVent'

In oxygen_delivery table, there are oxygen_delivery_device_1, o2_delivery_device_2, o2_delivery_device_3, o2_delivery_device_4 that are filled by alphabetical order with o2 delivery device names if there are more than once at the same time. In case of several simultaneous oxygen delivery devices connected, as Bipap mask, or CPAP mask may not be the first to appear in the alphabetical order, I would say that Non Invasive ventilation detection could be missed just focusing on o2_delivery_device_1 without considering o2_delivery_device_2, o2_delivery_device_3 and o2_delivery_device_4.

For patients, I am interested in, no one has in oxygen_delivery table records of 'Bipap mask', some have records with 'CPAP mask'

Looking at ventilator_setting records for these patients , I don't find any records with charttimes closely related to oxygen_delivery charttime records when oxygen_delivery_device_1 or 2 or3 or 4 is set to 'CPAP mask'. I find quite odd, not to find even a PEEP or fiO2 setting for a non Invasive ventilation with 'CPAP mask'. These patients are connected to a ventilator for NIV, but there are no ventilator settings.

What puzzles me too is that Non Invasive ventilation is defined by the interface ('Bipap mask' or 'CPAP mask') or by ventilation_mode_hamilton set to 'DuoPaP', 'NIV', 'NIV-ST' . Thus when a ventilator ,that is not a hamilton, is used , the determination of Non Invasive ventilation doesn't rely at all on the ventilator mode, probably because they may not have a fully specific NIV mode. I guess then that the NIV modes for these non hamilton ventilator are mixed-use and can be also used in Invasive ventilation. Therefore, I am wondering if these mixed Invasive Ventilation/Non Invasive ventilation modes for non Hamilton ventilators, may be part of those defining ventilation_status as InvasiveVent as defined in ventilation.sql.

If all that is clear to someone, thanks very much in advance for your help and explanations ! Cedric

Description

Description of the issue, including:

• what you have tried
• references to similar issues
• queries demonstrating your question (if applicable)

99% of papers use 512x512 or maximum 1024x1024 size images in chest x-ray dataset. 512x512 after resize would hardly be 6-7 GB after zipping.

it is criminal waste of resources to have people download data of 600GB and preprocess it to 6 GB. Can you please resize and upload..

Prerequisites

• [x] Put an X between the brackets on this line if you have done all of the following:
  • Checked the online documentation: https://mimic.mit.edu/
  • Checked that your issue isn't already addressed: https://github.com/MIT-LCP/mimic-code/issues?utf8=%E2%9C%93&q=

Description

Hi! I am wondering if the SAPS-II scores for ICU visits are calculated in the MIMIC-IV dataset, and which module/table are they included in? If they are not currently included, will future versions of MIMIC-IV include mortality probability scores (e.g., SAPS, APACHE, SOFA) for ICU visits? Thank you so much for the help!