Neural Network Predicts Need for Red Blood Cell Transfusion for Patients with Acute Gastrointestinal Bleeding Admitted to the Intensive Care Unit
Structured Summary Background Acute gastrointestinal bleeding is the most common gastrointestinal cause for hospitalization. For high risk patients requiring intensive care unit stay, predicting transfusion needs during the first 24 hours using dynamic risk assessment may improve resuscitation.Aims Provide dynamic risk prediction for red blood cell transfusion in admitted patients with severe acute gastrointestinal bleeding.Methods A patient cohort admitted for acute gastrointestinal bleeding (N = 2,524) was identified from the Medical Information Mart for Intensive Care III (MIMIC-III) critical care database, separated into training (N = 2,032) and validation (N = 492) sets. 74 demographic, clinical, and laboratory test features were consolidated into 4-hour time intervals over the first 24 hours from admission. The outcome measure was the transfusion of red blood cells during each 4-hour time interval. A long short-term memory (LSTM) model, a type of Recurrent Neural Network (RNN), was compared to the Glasgow-Blatchford Score (GBS).Results The LSTM model performed better than GBS overall (AUROC 0.81 vs 0.63;P<0.001)and at each 4-hour interval (P<0.01). At high sensitivity and high specificity cutoffs, the LSTM model outperformed GBS (P<0.001). The LSTM model performed better in patients directly admitted from the ED to ICU (0.82 vs 0.63;P<0.001), upper GIB (0.84 vs 0.68;P<0.001), lower GIB (0.77 vs 0.58;P<0.001), and unspecified GIB (0.85 vs 0.64;P<0.001).Conclusions A LSTM model can be used to predict the need for transfusion of packed red blood cells over the first 24 hours from admission to help personalize the care of high-risk patients with acute gastrointestinal bleeding.Data Access All clinical data from MIMIC-III was approved under the oversight of the Institutional Review Boards of Beth Israel Deaconess Medical Center (Boston, MA) and the Massachusetts Institute of Technology (Cambridge, MA). Requirement for individual patient consent was waived because the project did not impact clinical care and all protected health information was deidentified. The data was available on PhysioNet were derived from protected health information that has been de-identified and not subject to HIPAA Privacy Rule restrictions. All use of the data was performed with credentialed access under the oversight of the data use agreement through PhysioNet and the Massachusetts Institute of Technology..
Medienart: |
Preprint |
---|
Erscheinungsjahr: |
2022 |
---|---|
Erschienen: |
2022 |
Enthalten in: |
bioRxiv.org - (2022) vom: 24. Okt. Zur Gesamtaufnahme - year:2022 |
---|
Sprache: |
Englisch |
---|
Beteiligte Personen: |
Shung, Dennis [VerfasserIn] |
---|
Links: |
---|
Themen: |
---|
doi: |
10.1101/2020.05.19.20096743 |
---|
funding: |
|
---|---|
Förderinstitution / Projekttitel: |
|
PPN (Katalog-ID): |
XBI017926726 |
---|
LEADER | 01000caa a22002652 4500 | ||
---|---|---|---|
001 | XBI017926726 | ||
003 | DE-627 | ||
005 | 20230429080758.0 | ||
007 | cr uuu---uuuuu | ||
008 | 200525s2022 xx |||||o 00| ||eng c | ||
024 | 7 | |a 10.1101/2020.05.19.20096743 |2 doi | |
035 | |a (DE-627)XBI017926726 | ||
035 | |a (biorXiv)10.1101/2020.05.19.20096743 | ||
040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
041 | |a eng | ||
100 | 1 | |a Shung, Dennis |e verfasserin |4 aut | |
245 | 1 | 0 | |a Neural Network Predicts Need for Red Blood Cell Transfusion for Patients with Acute Gastrointestinal Bleeding Admitted to the Intensive Care Unit |
264 | 1 | |c 2022 | |
336 | |a Text |b txt |2 rdacontent | ||
337 | |a Computermedien |b c |2 rdamedia | ||
338 | |a Online-Ressource |b cr |2 rdacarrier | ||
520 | |a Structured Summary Background Acute gastrointestinal bleeding is the most common gastrointestinal cause for hospitalization. For high risk patients requiring intensive care unit stay, predicting transfusion needs during the first 24 hours using dynamic risk assessment may improve resuscitation.Aims Provide dynamic risk prediction for red blood cell transfusion in admitted patients with severe acute gastrointestinal bleeding.Methods A patient cohort admitted for acute gastrointestinal bleeding (N = 2,524) was identified from the Medical Information Mart for Intensive Care III (MIMIC-III) critical care database, separated into training (N = 2,032) and validation (N = 492) sets. 74 demographic, clinical, and laboratory test features were consolidated into 4-hour time intervals over the first 24 hours from admission. The outcome measure was the transfusion of red blood cells during each 4-hour time interval. A long short-term memory (LSTM) model, a type of Recurrent Neural Network (RNN), was compared to the Glasgow-Blatchford Score (GBS).Results The LSTM model performed better than GBS overall (AUROC 0.81 vs 0.63;P<0.001)and at each 4-hour interval (P<0.01). At high sensitivity and high specificity cutoffs, the LSTM model outperformed GBS (P<0.001). The LSTM model performed better in patients directly admitted from the ED to ICU (0.82 vs 0.63;P<0.001), upper GIB (0.84 vs 0.68;P<0.001), lower GIB (0.77 vs 0.58;P<0.001), and unspecified GIB (0.85 vs 0.64;P<0.001).Conclusions A LSTM model can be used to predict the need for transfusion of packed red blood cells over the first 24 hours from admission to help personalize the care of high-risk patients with acute gastrointestinal bleeding.Data Access All clinical data from MIMIC-III was approved under the oversight of the Institutional Review Boards of Beth Israel Deaconess Medical Center (Boston, MA) and the Massachusetts Institute of Technology (Cambridge, MA). Requirement for individual patient consent was waived because the project did not impact clinical care and all protected health information was deidentified. The data was available on PhysioNet were derived from protected health information that has been de-identified and not subject to HIPAA Privacy Rule restrictions. All use of the data was performed with credentialed access under the oversight of the data use agreement through PhysioNet and the Massachusetts Institute of Technology. | ||
650 | 4 | |a Biology |7 (dpeaa)DE-84 | |
650 | 4 | |a 570 |7 (dpeaa)DE-84 | |
700 | 1 | |a Castro, Egbert |e verfasserin |4 aut | |
700 | 1 | |a Huang, Jessie |e verfasserin |4 aut | |
700 | 1 | |a Tay, J. Kenneth |e verfasserin |4 aut | |
700 | 1 | |a Simonov, Michael |e verfasserin |4 aut | |
700 | 1 | |a Laine, Loren |e verfasserin |4 aut | |
700 | 1 | |a Krishnaswamy, Smita |e verfasserin |4 aut | |
773 | 0 | 8 | |i Enthalten in |t bioRxiv.org |g (2022) vom: 24. Okt. |
773 | 1 | 8 | |g year:2022 |g day:24 |g month:10 |
856 | 4 | 0 | |u https://doi.org/10.1038/s41598-021-88226-3 |z lizenzpflichtig |3 Volltext |
856 | 4 | 0 | |u http://dx.doi.org/10.1101/2020.05.19.20096743 |z kostenfrei |3 Volltext |
912 | |a GBV_XBI | ||
912 | |a SSG-OLC-PHA | ||
951 | |a AR | ||
952 | |j 2022 |b 24 |c 10 |