Table 4.
Special surgery-related AKI prediction model.
| Reference | Modeling data sources | Data volume |
Model performance AUC value |
Type of surgery | Machine learning methods | Model explanation | Diagnostic criteria | |
|---|---|---|---|---|---|---|---|---|
| Lei et al. (42) | Fuwai Hospital in Beijing, China | 897 | 0.80 | Aortic surgery | Logistic regression Support vector machines Random forest Gradient boosting |
None | KDIGO: Creatinine | |
| Penny-Dimri et al. (43) | The Australian and New Zealand Society of Cardiac and Thoracic Surgeons (ANZSCTS) database | 97,964 | 0.77–0.78 | Aortic surgery, cardiopulmonary diversion surgery, valve surgery, constrictive pericarditis surgery | Logistic regression Gradient boosted machine K-nearest neighbor L neural networks |
Yes | Improvement criteria | |
| Zhang et al. (40) | Nanjing First Hospital, China | 1,457 | 0.857–0.881 | Coronary artery bypass grafting, valve surgery | Extreme gradient boosting Random forest Deep forest Logistic regression |
Yes | KDIGO: Creatinine | |
| Li et al. (41) | Zhongshan Hospital, Fudan University, Shanghai, China | 3,639 | AKI | 0.755 | Valve surgery Coronary artery bypass grafting aorta + valve + CABG, valve + great vessel |
Bayesian networks | None | KDIGO: Creatinine and urine volume |
| Severe AKI | 0.845 | |||||||
| Lee et al. (44) | Seoul National University Hospital | 2010 | 0.78 | Coronary artery bypass grafting valve surgery | Decision trees Random forest Extreme gradient boosting Support vector machines Neural networks Deep learning |
None | KDIGO: Creatinine | |
| Tseng et al. (38) | Far Eastern Memorial Hospital (FEMH), New Taipei City | 671 | 0.839 | Coronary artery bypass grafting valve surgery, combination of both treatments | Logistic regression Support vector machines Random forest Extreme gradient boosting Integration algorithm (RF + XGBoost) |
Yes | KDIGO: Creatinine | |
| Petrosyan et al. (39) | Cardiocore, University of Ottawa Heart Institute |
6,522 | 0.74 | Cardiopulmonary diversion surgery | Hybrid algorithm (Random forest + logistic regression) Logistic regression Enhanced logistic regression |
None | KDIGO: Creatinine | |
| Lee et al. (45) | Seoul National University Bundang Hospital | 4,104 | 0.81 | Unilateral partial or total nephrectomy | Support vector machines Random forest Extreme gradient boosting Light gradient boosting machine |
None | KDIGO: Creatinine | |
| Lazebnik et al. (46) | Cancer Institute, University College London, UK | 723 | 0.75 | Open partial nephrectomy | Random forest | None | RIFLE and AKIN: Creatinine | |
| Zhu et al. (47) | Peking University First Hospital, China | 87 | 0.749 | Isolated partial nephrectomy | Decision trees Random forest Logistic regression Support vector machines Extreme gradient boosting |
None | KDIGO: Creatinine | |
| Bredt et al. (48) | 145 | 0.81 | Deceased donor liver transplantation | Logistic regression Artificial neural networks |
None | KDIGO: Creatinine | ||
| Lee et al. (44) | Seoul National University Hospital, South Korea | 1,211 | 0.90 | Deceased donor/living donor liver transplantation | Decision trees Random forest Gradient boosted machine Support vector machines Naive Bayes Multilayer perceptron Deep belief networks |
None | AKIN: Creatinine | |
| Dong et al. (49) | Changzheng Hospital, Shanghai, China | 2,450 | 0.92 | Liver cancer resection | Logistic regression Support vector machines Random forest Extreme gradient boosting Decision trees |
None | KDIGO: Creatinine | |
| He et al. (50) | The First Affiliated Hospital of Zhejiang University School of Medicine, China |
493 | 0.85 | Cardiac death donor liver transplantation | Random forest Support vector machines Decision trees Conditional reasoning tree Logistic regression |
None | KDIGO: Creatinine and urine volume | |
| Lei et al. (42) | The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China | 1,173 | 0.772 | Liver cancer resection | Gradient boosting decision tree Random forest Decision trees |
None | KDIGO: Creatinine | |
| Ko et al. (51) | Seoul National University College of Medicine, Seoul National University Bundang Hospital, South Korea |
5,302 | 0.78 | Knee arthroplasty | Gradient enhancement | None | KDIGO: Creatinine | |
| Nikkinen et al. (52) | Oulu University Hospital, Oulu, Finland | 648 | 0.91/0.98 | Knee arthroplasty, hip arthroplasty | RUSBoost Naive Bayes Support vector machines |
None | KDIGO: Creatinine and urine output | |
KDIGO, Kidney Disease: Improving Global Outcomes; AKIN, Acute Kidney Injury Network; RIFLE, Risk, Injury, Failure, Loss of renal function and End-stage renal disease.