Table 1.
Summary of review.
| ID | Method | Sample Size | Data Type | Performance | Important Predictor |
|---|---|---|---|---|---|
| [33] | ANN DT LR * NB RF * SVM |
731 | Numeric | Accuracy 0.79–0.87 AUC 0.54–0.76 |
RFVI for the prediction of preterm birth, which has a strong association with GERD: Age, education, upper gastrointestinal tract symptom, Helicobacter pylori, region |
| [34] |
APACHE XGB * |
5691 | Numeric | Sensitivity 1.00 Specificity 0.04–0.27 AUC 0.80–0.85 |
SHAP for the prediction of mortality from gastrointestinal bleeding in the intensive care unit: mean arterial pressure (max), bicarbonate (min), creatinine (max), PMN, heart rate (mean), Glasgow Coma Scale, age, respiratory rate (mean), prothrombin time (max), aminotransferase aspartate (max), albumin (min), oxygen saturation (mean), white blood cell, AlkPhos (max), platelet (min), lactate (max), intubation, bilirubin (max), international normalized ratio (max), vasopressor, glucose (max), blood urea nitrogen (max), PTT (max), hemoglobin (min), potassium |
| [35] | RF * | 340 | Genomic | Accuracy 0.70 AUC 0.92 |
RFVI for the prediction of food intake (almond, avocado, broccoli, walnut, whole-grain barley, whole-grain oat): Roseburia undefined, Lachnospira spp., Oscillibacter undefined, Subdoligranulum spp., Streptococcus salivarius subsp. thermophiles, Parabacteroides distasonis, Roseburia spp., Anaerostipes spp., Lachnospiraceae ND3007 group undefined, Ruminiclostridium spp. |
| [36] | CB * | 337 | Genomic | AUC 0.81–0.84 |
SHAP for the prediction of early intestinal resection with Crohn’s disease: age, disease behavior (clinical predictors), rs28785174, rs60532570, rs13056955, rs7660164 (single nucleotide polymorphisms) |
| [37] | RF * | 71 | Radiomic | Accuracy 0.78–0.94 |
RFVI for the prediction of pneumatosis: dissecting gas in the bowel wall, intramural gas beyond a gas-fluid/fecal level, a circumferential gas pattern |
| [38] | ANN * LR * RF * |
405,586 | Numeric | Accuracy 0.93–0.98 |
RFVI for the prediction of preterm birth, which has a strong association with GERD: socioeconomic status, age, region (city) |
| [39] | RF * | 710 | Numeric | AUC 0.76–0.80 |
RFVI for the prediction of COVID-19 hospitalization based on gastrointestinal factors: aspartate transaminase, diabetes mellitus, chronic liver disease, alanine transaminase, diarrhea, age, bloating |
| [40] | RF * | 590 | Numeric | AUC 0.68 |
RFVI for the prediction of gastrointestinal sequelae months after COVID-19 infection: acute diarrhea, antibiotics administration |
ANN—Artificial Neural Network, AUC—Area under the Receiver Operating Characteristic Curve, CB—CatBoost, DT—Decision Tree, LR—Logistic Regression, NB—Naïve Bayes, RF—Random Forest, RFVI—Random Forest Variable Importance, SHAP—Shapley Additive Explanations, SVM—Support Vector Machine, XGB—XGBoost, * Method with the best performance.