A customised down-sampling machine learning approach for sepsis prediction
Copyright © 2024 The Author(s). Published by Elsevier B.V. All rights reserved..
OBJECTIVE: Sepsis is a life-threatening condition in the ICU and requires treatment in time. Despite the accuracy of existing sepsis prediction models, insufficient focus on reducing alarms could worsen alarm fatigue and desensitisation in ICUs, potentially compromising patient safety. In this retrospective study, we aim to develop an accurate, robust, and readily deployable method in ICUs, only based on the vital signs and laboratory tests.
METHODS: Our method consists of a customised down-sampling process and a specific dynamic sliding window and XGBoost to offer sepsis prediction. The down-sampling process was applied to the retrospective data for training the XGBoost model. During the testing stage, the dynamic sliding window and the trained XGBoost were used to predict sepsis on the retrospective datasets, PhysioNet and FHC.
RESULTS: With the filtered data from PhysioNet, our method achieved 80.74% accuracy (77.90% sensitivity and 84.42% specificity) and 83.95% (84.82% sensitivity and 82.00% specificity) on the test set of PhysioNet-A and PhysioNet-B, respectively. The AUC score was 0.89 for both datasets. On the FHC dataset, our method achieved 92.38% accuracy (88.37% sensitivity and 95.16% specificity) and 0.98 AUC score on the test set of FHC.
CONCLUSION: Our results indicate that the down-sampling process and the dynamic sliding window with XGBoost brought robust and accurate performance to give sepsis prediction under various hospital settings. The localisation and robustness of our method can assist in sepsis diagnosis in different ICU settings.
| Medienart: |
E-Artikel |
|---|
| Erscheinungsjahr: |
2024 |
|---|---|
| Erschienen: |
2024 |
| Enthalten in: |
Zur Gesamtaufnahme - volume:184 |
|---|---|
| Enthalten in: |
International journal of medical informatics - 184(2024) vom: 05. März, Seite 105365 |
| Sprache: |
Englisch |
|---|
| Beteiligte Personen: |
Wu, Qinhao [VerfasserIn] |
|---|
| Links: |
|---|
| Themen: |
Alarm reduction |
|---|
| Anmerkungen: |
Date Completed 05.03.2024 Date Revised 05.03.2024 published: Print-Electronic Citation Status MEDLINE |
|---|
| doi: |
10.1016/j.ijmedinf.2024.105365 |
|---|
| funding: |
|
|---|---|
| Förderinstitution / Projekttitel: |
|
| PPN (Katalog-ID): |
NLM368402894 |
|---|
| LEADER | 01000caa a22002652 4500 | ||
|---|---|---|---|
| 001 | NLM368402894 | ||
| 003 | DE-627 | ||
| 005 | 20240305232349.0 | ||
| 007 | cr uuu---uuuuu | ||
| 008 | 240214s2024 xx |||||o 00| ||eng c | ||
| 024 | 7 | |a 10.1016/j.ijmedinf.2024.105365 |2 doi | |
| 028 | 5 | 2 | |a pubmed24n1317.xml |
| 035 | |a (DE-627)NLM368402894 | ||
| 035 | |a (NLM)38350181 | ||
| 035 | |a (PII)S1386-5056(24)00028-5 | ||
| 040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
| 041 | |a eng | ||
| 100 | 1 | |a Wu, Qinhao |e verfasserin |4 aut | |
| 245 | 1 | 2 | |a A customised down-sampling machine learning approach for sepsis prediction |
| 264 | 1 | |c 2024 | |
| 336 | |a Text |b txt |2 rdacontent | ||
| 337 | |a ƒaComputermedien |b c |2 rdamedia | ||
| 338 | |a ƒa Online-Ressource |b cr |2 rdacarrier | ||
| 500 | |a Date Completed 05.03.2024 | ||
| 500 | |a Date Revised 05.03.2024 | ||
| 500 | |a published: Print-Electronic | ||
| 500 | |a Citation Status MEDLINE | ||
| 520 | |a Copyright © 2024 The Author(s). Published by Elsevier B.V. All rights reserved. | ||
| 520 | |a OBJECTIVE: Sepsis is a life-threatening condition in the ICU and requires treatment in time. Despite the accuracy of existing sepsis prediction models, insufficient focus on reducing alarms could worsen alarm fatigue and desensitisation in ICUs, potentially compromising patient safety. In this retrospective study, we aim to develop an accurate, robust, and readily deployable method in ICUs, only based on the vital signs and laboratory tests | ||
| 520 | |a METHODS: Our method consists of a customised down-sampling process and a specific dynamic sliding window and XGBoost to offer sepsis prediction. The down-sampling process was applied to the retrospective data for training the XGBoost model. During the testing stage, the dynamic sliding window and the trained XGBoost were used to predict sepsis on the retrospective datasets, PhysioNet and FHC | ||
| 520 | |a RESULTS: With the filtered data from PhysioNet, our method achieved 80.74% accuracy (77.90% sensitivity and 84.42% specificity) and 83.95% (84.82% sensitivity and 82.00% specificity) on the test set of PhysioNet-A and PhysioNet-B, respectively. The AUC score was 0.89 for both datasets. On the FHC dataset, our method achieved 92.38% accuracy (88.37% sensitivity and 95.16% specificity) and 0.98 AUC score on the test set of FHC | ||
| 520 | |a CONCLUSION: Our results indicate that the down-sampling process and the dynamic sliding window with XGBoost brought robust and accurate performance to give sepsis prediction under various hospital settings. The localisation and robustness of our method can assist in sepsis diagnosis in different ICU settings | ||
| 650 | 4 | |a Journal Article | |
| 650 | 4 | |a Alarm reduction | |
| 650 | 4 | |a Early detection | |
| 650 | 4 | |a Intensive care unit | |
| 650 | 4 | |a Machine learning | |
| 650 | 4 | |a Sepsis prediction | |
| 700 | 1 | |a Ye, Fei |e verfasserin |4 aut | |
| 700 | 1 | |a Gu, Qianqian |e verfasserin |4 aut | |
| 700 | 1 | |a Shao, Feng |e verfasserin |4 aut | |
| 700 | 1 | |a Long, Xi |e verfasserin |4 aut | |
| 700 | 1 | |a Zhan, Zhuozhao |e verfasserin |4 aut | |
| 700 | 1 | |a Zhang, Junjie |e verfasserin |4 aut | |
| 700 | 1 | |a He, Jun |e verfasserin |4 aut | |
| 700 | 1 | |a Zhang, Yangzhou |e verfasserin |4 aut | |
| 700 | 1 | |a Xiao, Quan |e verfasserin |4 aut | |
| 773 | 0 | 8 | |i Enthalten in |t International journal of medical informatics |d 1998 |g 184(2024) vom: 05. März, Seite 105365 |w (DE-627)NLM092193730 |x 1872-8243 |7 nnns |
| 773 | 1 | 8 | |g volume:184 |g year:2024 |g day:05 |g month:03 |g pages:105365 |
| 856 | 4 | 0 | |u http://dx.doi.org/10.1016/j.ijmedinf.2024.105365 |3 Volltext |
| 912 | |a GBV_USEFLAG_A | ||
| 912 | |a GBV_NLM | ||
| 951 | |a AR | ||
| 952 | |d 184 |j 2024 |b 05 |c 03 |h 105365 | ||