Table 1 Descriptive of the included papers (n = 20)
Paper | Data Source type and number(location) | Sample (n) | Outcome (positive outcome rate/%*) | Number of input features used | Categories of features | Best performing model | Types of models | Main performing metrics | ||
|---|---|---|---|---|---|---|---|---|---|---|
Recipient (n) | Donor (n) | Operative (n) | ||||||||
Cruz-RamÃrez et al. (2011) [24] | 11 Hospitals (Spain) | 1001 | 12-month graft mortality (16.1) | 42 | 19 | 20 | 3 | MPDENN_C, MPDENN_MS | Classification (Neural Network) | MPDENN-E C = 84.46; MPDENN-MS MS = 45.55 |
Cruz-RamÃrez et al. (2011) [25] | 11 Hospitals (Spain) | 1003 | 3-month graft survival (NA) | 39 | 16 | 20 | 3 | MPDENN-E, MPDENN-MS | Classification (Neural Network) | MPDENN-E C = 89.29, MS = 13.79, RMSE = 0.3212; MPDENN_MS C = 63.89, MS = 62.07, RMSE = 0.3863 |
Cruz-RamÃrez et al. (2012) [26] | 11 Hospital (Spain) | 1001 | 1-year graft survival (16.1) | 41 | 16 | 16 | 9 | MPDENN-E, MPENSGA-2 | Classification (Neural Network) | MPDENN-E C = 83.68, MRSE = 0.3795; MPENSGA2-MS MS = 52.04, AUC = 0.5694 |
Cruz-RamÃrez et al. (2014) [27] | 11 Hospital (Spain) | 1003 | 3-month graft mortality (NA) | 57 | 26 | 19 | 12 | NN | Classification (Neural Network) | NN-CCR (Correct classification rate) = 90.79%, NN-MS(Minimum sensitivity) = 71.42% |
Briceno et al. (2013) [28] | 11 Hospital (Spain) | 1003 | 3-month graft mortality (NA) | 39 | 16 | 20 | 3 | MPENSGA-2 | Classification (Neural Network) | MS = 48.98, AUC = 0.5659 |
Pérez-Ortiz et al. (2017) [29] | 11 Hospital (UK) | 822 | 3- and 12- month graft survival (NA) | 37 | 16 | 17 | 4 | LSVC (for 3- months), CSSVC (for 12-months) | Classification (Linear, non-Linear, Neural Network) | LSVC Acc = 90.15, CSSVC Acc = 90.15 |
Dorado-Moreno et al. (2017) [30] | Hospitals (7 Spain, 1 UK) | 1,406 | < 15-days, 15-90-days, and 90-365-days graft survival (NA) | 38 | 16 | 17 | 5 | DIM-ORNET | Classification (Neural Network) | Acc = 73.57%, geometric mean sensitivity (GMS) = 31.46%, Average mean absolute error (AMAE) = 1.155 |
Guijo-Rubio et al. (2021) [31] | 1 Registry (UNOS, US) | 39,189 | 3-months (7.7), 1-year (15.3), 2-years (22.1), 5-years (76.8) graft survival | 28 | 15 | 11 | 2 | LR | Classification (Linear, Decision Trees) | LR: AUC = 0.654, Acc = 0.614, MS = 0.584 |
Zhang et al. (2022) [32] | 1 Registry (UNOS, US) | 3-month: 478,777, 1-year: 47,401, 3-years: 6,380, 5-years: 45,270, 10-years: 20,751 | 3-month (6.4), 1-year (12.5), 2-years (21.2), 3-years (21.28), 5-years (27.8), 10-years (45.3) recipient mortality | 42 |  |  |  | XGBoost | Classification (Decision Tree) | AUC = of 0.717 for 3 months, 0.681 for 1 year, 0.662 for 3 years, 0.660 for 5 years, and 0.674 for 10 years. |
Andres et al. (2018) [33] | 1 Registry (SRTR, US) | 2,769 | 0.25-year, 1-year, 3-years, 5-years, 10-years recipient survival (NA) | 4 | 4 |  |  | Cox model | Regression (Linear) | C-statistics for 0.25 year = 95.6%, 1 year = 93%, 3 year = 87.6%, 5 year = 84.1%, and 10 year = 72% |
Lau et al. (2017) [34] | 1 Hospital (Australia) | 180 | 30-days (8.8), and 3-months graft failure (6.1) | 15 | 3 | 12 |  | NN | Classification (Neural Network) | AUC = 0.835 |
Farzindar et al. (2019) [35] | 2 Registry (UNOS and SRTR, US) | 87,334 | Precise time of failure (Time to event) (NA) |  |  |  |  | Deep survival model | Regression (Deep survival model) | C-index = 0.82 during development and 0.57 during testing |
Ershoff et al. (2020) [36] | 1 Registry (UNOS, US) | 57,544 | 90 days recipient mortality (5.4) | 202 | 132 | 70 |  | DNN | Classification (Neural Network) | AUC = 0.703 |
Kwong et al. (2021) [37] | 1 Registry (OPTN, US) | 18,920 | Waitlist dropout at 3-months (6.5), 6-months (11.3), 12-months (17.2) | 12 | 12 |  |  | Cox model | Regression (Linear) | C-statistic = 0.74. |
Kantidakis et al. (2020) [38] | 1 Registry (UNOS, US) | 62,294 | Overall graft survival (NA) | 97 | 52 | 45 |  | RF and NN | Regression (Linear) | RF: C-index = 0.622 NN: IBS = 0.180 |
Yu et al. (2022) [39] | 1 Registry (Korea) | 785 | 1-month (8.1), 3-months (11.2), 12-months (17.2) recipient survival |  |  |  |  | RF | Classification (Decision Tree) | AUC = 0.80 for 1 month, 0.85 for 3 months, and 0.81 for 12 months |
Börner et al. (2022) [40] | 1 Hospital (Germany) | 529 | 2-months, 6-months, 9-months, 12-months in-hospital recipient survival (NA) | 48 | 24 | 20 | 4 | DNN | Classification (Neural Network) | Acc = 95.8% and AUC = 0.940 |
Lankarani et al. (2022) [41] | 1 Center (Iran) | 1,947 | 2-years waitlist mortality (18.4) | 25 | 25 |  |  | ANN, SVM | Classification (Neural Network) | MELDNa < 23, age < 53, and ALP < 257 were the best predictors of survival in candidates |
Raju et al. (2023) [42] | 1 Registry (UNOS, US) | 62,556 | 90 days recipient mortality (NA) | 29 | 29 |  |  | FT-Transformer | Classification (Neural Network) | AUC = 0.96–0.98, Acc = 0.89 |
Ivanics et al. (2023) [43] | 3 Registry (UNOS/US, Canadian, UK) | UNOS = 59,558, Canada = 1,214, UK = 5287, | 90-days recipient mortality (NA) | 23 | 15 | 4 | 4 | Ridge-Logistic regression | Classification (Linear) | AUC = 0.74 − 0.71. External model performance across countries overall had poor performed |