Table 1.
Basic information of the included studies
| Study | Sepsis definition | Target | Data sources | Missing data processing | Training data | Testing data | Validation data |
|---|---|---|---|---|---|---|---|
| Delahanty et al. (2019) | sepsis3.0 | Early prediction of sepsis | 49 urban community hospitals operated by Tenet Healthcare | NR | 1,839,503 | 920,026 | NR |
| Barton et al. (2019) | sepsis3.0 | Detection and early prediction of sepsis | UCSF data+BIDMC data | Carry-forward and replacing by mean | NR | NR | NR |
| Taylor et al. (2016) | Infection + SIRS | Mortality prediction of sepsis | Four emergency departments | K-means | 4222 | NR | 1056 |
| Kam and Kim (2017) | ICD-9 | Detection and early prediction of sepsis | MIMIC-II | Replacing by nearest measured value | 252 | 72 | 36 |
| Mao et al. (2018) | SIRS | Sepsis detection | UCSF data+BIDMC data | Carry-forward and replacing by mean | 80% | 20% | NR |
| Taneja et al. (2017) | Clinical adjudication label | Early prediction of sepsis | Carle Foundation Hospital | NR | NR | NR | NR |
| Saqib et al. (2018) | Angus | Early prediction of sepsis | MIMIC-III | Forward-filling | 81% | 10% | 9% |
| Perng et al. (2019) | SIRS + qSOFA | Mortality prediction of sepsis | Chang Gung Research Database | Replacing by medium number of the column | 70% | 30% | NR |
| Thottakkara et al. (2016) | the criteria of the Agency for Healthcare Research and Quality | Severe sepsis prediction | DECLARE data | Replacing by mean value | 70% | NR | 30% |
| Bloch et al. (2019) | Infection +SIRS | Early prediction of sepsis | Israel Rabin Medical Center | NR | 75% | 25% | NR |
| Kwon and Baek, (2020) | Infection + qSOFA | Mortality prediction of sepsis | Four hospitals of Korea | NR | 74% | 18% | 8% |
| Nemati et al. (2018) | sepsis3.0 | Early prediction of sepsis | two hospitals within the Emory Healthcare system and an ICU database | NR | 80% | 20% | NR |
| Lauritsen et al. (2020) | Infection +SIRS | Early detection and prediction of sepsis | Four Danish municipalities data | NR | 80% | 10% | 10% |
| Scherpf et al. (2019) | ICD9+SIRS | Early prediction of sepsis | MIMIC-III | Liner interpolation and “carry forward/backward” extrapolation | NR | NR | NR |
| Hou et al. (2020) | Sepsis3.0 | Mortality prediction of sepsis | MIMIC III v1.4 | Remove the variables with more than 20% observations missing + multiple imputation method | NR | NR | NR |
| Kong et al. (2020) | Sepsis3.0 | Mortality prediction of sepsis | MIMIC III | Remove the patients with more than 30% predictor variable missing + Replace by mean value | NR | NR | NR |
| Bedoya et al.(2020) | SIRS + infection + end organ failure | Early detection of sepsis | ED of a quaternary academic hospital | NR | NR | NR | NR |
| van Doorn et al. (2021) | Infection + SIRS/SOFA | Mortality prediction of sepsis | ED at the Maastricht University Medical Center+ | NR | 1244 | NR | 100 |
| Li et al. (2021) | ICD-9 | Mortality prediction of sepsis | MIMIC-III V1.4 | Remove the patients with data missing more than 30% + Replace by mean value | NR | NR | NR |
| Burdick et al. (2020) | SIRS | Early severe sepsis prediction | The Dascena Analysis Dataset and the Cabell Huntington Hospital Dataset | last-one carry forward | NR | NR | NR |
| Qi et al. (2021) | Sepsis3.0 | Mortality prediction of sepsis | MIMIC-III | Remove the patients with data missing more than 40% + Replace by 21% and mean value | NR | NR | NR |
Abbreviation:SIRS: Systemic Inflammatory Response Syndrome; ICD9:international classification of diseases 9; NR: not reported.