From: Predicting sepsis onset in ICU using machine learning models: a systematic review and meta-analysis
Studies | First Author | Nation | Study Type | Sample source | Disease background | Diagnosis of sepsis | Missing data | Model type |
---|---|---|---|---|---|---|---|---|
SSP: Early prediction of sepsis using fully connected LSTM-CNN model | Alireza Rafiei2020 | Iran | Retrospective Cohort | Retrospectively | ≥ 14y patients and ICU LOS > 10d | Sepsis 3.0 | Multiple imputation | SSP-LSTM, SSP-GRU, InSight, AISE |
Evaluation of a machine learning algorithm for up to 48-h advance prediction of sepsis using six vital signs | Christopher Barton2019 | U.S.A | Prospectively | Prospectively | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | MLA、SIRS、MEWS、SOFA |
Predicting 30-days mortality for MIMIC-III patients with sepsis-3: a machine learning approach using Xgboost | Nianzong Hou2020 | P.R.C | Prospectively | Prospectively | ≥ 18y patients and ICU LOS > 24d | Sepsis 3.0 | Delete | XGBoost, LR,SAPS-II |
A machine learning-based model for 1-year mortality prediction in patients admitted to an Intensive Care Unit with a diagnosis of sepsis | J E García-Gallo2018 | Colombia | Retrospectively | Retrospectively | ≥ 16y patients | Sepsis 3.0 | Multiple imputation | SGB、OASIS、SOFA、SAPS2 |
Interpretable Machine Learning for Early Prediction of Prognosis in Sepsis: A Discovery and Validation Study | Chang Hu2022 | P.R.C | Prospectively | Prospectively | ≥ 18y patients and ICU LOS > 24d | Sepsis 3.0 | Multiple imputation | SVM、KNN、XGBoost、DT、RF、NB、LR |
Machine Learning Model to Identify Sepsis Patients in the Emergency Department: Algorithm Development and Validation | Lin PC2021 | P.R.C | Retrospective Cohort | Retrospective Cohort | ≥ 20y patients | Sepsis 3.0 | Multiple imputation | XGBoost、SIRS、SOFA |
Dynamic Sepsis Prediction for Intensive Care Unit Patients Using XGBoost-Based Model With Novel Time-Dependent Features | Shuhui Liu2017 | P.R.C | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | RF、GRU、CNNLSTM、EASP |
Effect of a machine learning-based severe sepsis prediction algorithm on patient survival and hospital length of stay: a randomised clinical trial | Shimabukuro DW2018 | U.S.A | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | MLA、SIRS、MEWS、SOFA |
A Predictive Model Based on Machine Learning for the Early Detection of Late-Onset Neonatal Sepsis: Development and Observational Study | Wongeun Song2022 | KOREA | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | RF、LR、SVM、NB、XGBOOST |
Machine learning approach for the prediction of 30-day mortality in patients with sepsis-associated encephalopathy | Liwei Peng2022 | P.R.C | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | CS、MIG、LLI、ET、RF、GB |
Development and validation of a novel blending machine learning model for hospital mortality prediction in ICU patients with Sepsis | Zhixuan Zeng2021 | P.R.C | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | SAPS II、SOFA、LR、LDA、CART、NB、KNN、MLP、SVM、RF、XGB |
Machine learning predicts mortality in septic patients using only routinely available ABG variables: a multi-centre evaluation | Bernhard Wernly2021 | Austria | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | LR、LSTM、SOFA |
A Machine Learning Model for Accurate Prediction of Sepsis in ICU Patients | Dong Wang2021 | P.R.C | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | RF |
Early Prediction of Mortality, Severity, and Length of Stay in the Intensive Care Unit of Sepsis Patients Based on Sepsis 3.0 by Machine Learning Models | Longxiang Su2021 | Germany | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | LR、RF、XGBoost |
Supervised classification techniques for prediction of mortality in adult patients with sepsis | Rodríguez A2021 | Colombia | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | DT、RF、NN、SVM |
A Machine Learning Sepsis Prediction Algorithm for Intended Intensive Care Unit Use (NAVOY Sepsis): Proof-of-Concept Study | Inger Persson2021 | Sweden | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | NAVOY |
Development of a Nomogram to Predict 28-Day Mortality of Patients With Sepsis-Induced Coagulopathy: An Analysis of the MIMIC-III Database | Zongqing Lu2021 | P.R.C | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | Nomogram 、SOFA、LODS、SAPS II、SIC score |
A Simple Weaning Model Based on Interpretable Machine Learning Algorithm for Patients With Sepsis: A Research of MIMIC-IV and eICU Databases | Wanjun Liu2022 | P.R.C | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | XGBOOST 、MLP、RF 、SVM 、LR 、KNN |
The development an artificial intelligence algorithm for early sepsis diagnosis in the intensive care unit | Yuan KC2020 | P.R.C | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | XGBoost、SOFA |
Early diagnosis of bloodstream infections in the intensive care unit using machine-learning algorithms | Michael Roimi2019 | Israel | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | RF |
Predicting sepsis with a recurrent neural network using the MIMIC III database | Matthieu Scherpf2019 | Germany | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | RNN、InSight |
Predicting central line-associated bloodstream infections and mortality using supervised machine learning | Joshua P. Parreco20182018 | U.S.A | Retrospective Cohort | Retrospective Cohort | ≥ 19y patients | Sepsis 3.1 | Multiple imputation | LR、GBT、DL |
Multicentre validation of a sepsis prediction algorithm using only vital sign data in the emergency department, general ward and ICU | Qingqing Mao2018 | U.S.A | Retrospective Cohort | Retrospective Cohort | ≥ 18y patients | Sepsis 3.0 | Multiple imputation | InSight、 MEWS 、SOFA、 SIRS |
Studies | Train set sepsis number | Train set number | Testing set | Method of testing | Test set sepsis number | Test set number | Number of variables | Outcome indicators |
---|---|---|---|---|---|---|---|---|
SSP: Early prediction of sepsis using fully connected LSTM-CNN model | 2542 | 20336 | 1 | Multicenter | 2500 | 20000 | 14 | AUROC, Sensitivity Specificity |
Evaluation of a machine learning algorithm for up to 48-h advance prediction of sepsis using six vital signs | 2649 | 91445 | 1 | Multicenter | 1024 | 21507 | 4 | AUROC Sensitivity Specificity |
Predicting 30-days mortality for MIMIC-III patients with sepsis-3: a machine learning approach using Xgboost | 10704 | 46520 | 1 | Multicenter | 889 | 4559 | 12 | AUROC |
A machine learning-based model for 1-year mortality prediction in patients admitted to an Intensive Care Unit with a diagnosis of sepsis | 46520 | 58977 | 1 | Multicenter | 5650 | 15254 | 18 | AUROC |
Interpretable Machine Learning for Early Prediction of Prognosis in Sepsis: A Discovery and Validation Study | 12292 | 76540 | 1 | Multicenter | 8817 | 12292 | 15 | AUROC Sensitivity Specificity |
Machine Learning Model to Identify Sepsis Patients in the Emergency Department: Algorithm Development and Validation | 6637 | 8296 | 1 | Random sampling | 506 | 1744 | 26 | AUROC Sensitivity Specificity |
Dynamic Sepsis Prediction for Intensive Care Unit Patients Using XGBoost-Based Model With Novel Time-Dependent Features | 3526 | 34472 | 1 | Random sampling | 4526 | 34472 | 30 | AUROC Sensitivity Specificity |
Effect of a machine learning-based severe sepsis prediction algorithm on patient survival and hospital length of stay: a randomised clinical trial | 67 | 142 | 0 | Single center | —— | —— | 3 | AUROC Sensitivity Specificity |
A Predictive Model Based on Machine Learning for the Early Detection of Late-Onset Neonatal Sepsis: Development and Observational Study | 1572 | 40366 | 1 | Multicenter | 315 | 1257 | 21 | AUROC Sensitivity Specificity |
Machine learning approach for the prediction of 30-day mortality in patients with sepsis-associated encephalopathy | 4897 | 382278 | 1 | Multicenter | 2097 | 382278 | 15 | AUROC Sensitivity Specificity |
Development and validation of a novel blending machine learning model for hospital mortality prediction in ICU patients with Sepsis | 12558 | 200859 | 1 | Multicenter | 12095 | 61532 | 6 | AUROC |
Machine learning predicts mortality in septic patients using only routinely available ABG variables: a multi-centre evaluation | 8061 | 61532 | 1 | Multicenter | 3853 | 200859 | 23 | Sensitivity Specificity |
A Machine Learning Model for Accurate Prediction of Sepsis in ICU Patients | 3539 | 17005 | 1 | Multicenter | 910 | 17005 | 55 | AUROC, Sensitivity Specificity |
Early Prediction of Mortality, Severity, and Length of Stay in the Intensive Care Unit of Sepsis Patients Based on Sepsis 3.0 by Machine Learning Models | 2436 | 11512 | 0 | Single center | —— | —— | 26 | AUROC, Sensitivity Specificity |
Supervised classification techniques for prediction of mortality in adult patients with sepsis | 2510 | 5022 | 1 | Multicenter | 2510 | 5022 | 27 | AUROC, Sensitivity Specificity |
A Machine Learning Sepsis Prediction Algorithm for Intended Intensive Care Unit Use (NAVOY Sepsis): Proof-of-Concept Study | 2893 | 61532 | 0 | Single center | —— | —— | 6 | AUROC Sensitivity Specificity |
Development of a Nomogram to Predict 28-Day Mortality of Patients With Sepsis-Induced Coagulopathy: An Analysis of the MIMIC-III Database | 3280 | 9432 | 1 | Multicenter | 987 | 3280 | 17 | AUROC |
A Simple Weaning Model Based on Interpretable Machine Learning Algorithm for Patients With Sepsis: A Research of MIMIC-IV and eICU Databases | 5020 | 10832 | 1 | Multicenter | 7081 | 33790 | 20 | AUROC Sensitivity Specificity |
The development an artificial intelligence algorithm for early sepsis diagnosis in the intensive care unit | 319 | 1588 | 0 | Single center | —— | —— | 19 | AUROC Sensitivity Specificity |
Early diagnosis of bloodstream infections in the intensive care unit using machine-learning algorithms | 1021 | 1812 | 1 | Multicenter | 2351 | 7419 | 29 | AUROC Sensitivity Specificity |
Predicting sepsis with a recurrent neural network using the MIMIC III database | 4278 | 58976 | 0 | Single center | —— | —— | 10 | AUROC |
Predicting central line-associated bloodstream infections and mortality using supervised machine learning | 22201 | 57786 | 0 | Single center | —— | —— | 37 | AUROC Sensitivity Specificity |
Multicentre validation of a sepsis prediction algorithm using only vital sign data in the emergency department, general ward and ICU | 1179 | 21604 | 0 | Single center | —— | —— | 7 | AUROC Sensitivity Specificity |