Table 2.
ML applications in PE research.
| Reference | ML application | Input | ML technique | Main output |
|---|---|---|---|---|
| (41) | Disease prediction | Metabolomics data | LR | AUC = 0.868, Se = 75.1% and Sp = 83.0% |
| (42) | Disease prediction | Proteomics data | LDA | AUC = 0.96, Se = 0.90 and Sp = 0.90 for early-onset cases with maternal vascular malperfusion |
| (26) | Disease prediction | Genomics data | LR | AUC = 0.825, Ac = 83.0%, Se = 81.7% and Sp = 83.3% |
| (43) | Disease prediction | Transcriptomics data and biochemical markers | LR | AUC = 0.940, Se = 86.67% and Sp = 96.67% |
| (44) | Disease prediction | Biochemical markers | BPNN | Ac = 79.8% |
| (45) | Disease prediction | Electronic health records | Stochastic GB | AUC = 0.924, Ac = 0.973, Se = 0.603, Sp = 0.991 and DR = 0.771 for late-onset cases |
| (46) | Disease prediction | Electronic health records | EN | AUC = 0.89, Se = 72.3% and Sp = 91.2% for early-onset cases |
| (47) | Disease prediction | Clinical and biochemical factors | LR | AUC = 0.962, Se = 79.3%, Sp = 97.7%, PPV = 92% and NPV = 93.4% |
| (48) | Disease prediction | Clinical and biochemical factors | LR | AUC = 0.68, Se = 30.6% and Sp = 90% for early-onset cases |
| (49) | Disease prediction | Clinical and biochemical factors | RF | AUC = 0.976, AUPR = 0.958, Ac = 92.6%, Se = 91% and Sp = 93% for placental dysfunction-related disorders |
| (50) | Disease prediction | Clinical parameters | RF | AUC = 0.90, Se = 0.70, Sp = 0.89 and Pr = 0.88 |
| (51) | Disease prediction | Ultrasound images | CNN | Se = 70.6% and Sp = 76.6% for hypertension disorders of pregnancy |
| (52) | Biomarker discovery | Genomics data | SVM | IL7R, IL18, CCL2, HLA-DRA, CD247, ITK, CD2, IRF8, CD48, GZMK, CCR7, HLA-DPA1, LEP, IL1B, CD8A, CD3D and GZMA as novel biomarkers |
| (53) | Biomarker discovery | Transcriptomics data | C4.5, AB and MLP | HTRA4, PROCR, MYCN, ERO1A, EAF1, PPP1R16B, CRH, FLNB, PIK3CB, PLAAT3, FBN2, RFLNB, and TKT as novel biomarkers |
| (54) | Risk estimation | Food frequency questionnaire data | SL | 3.2 and 4.0 fewer cases of PE per 100 births for high density fruit and vegetable intake |
| (55) | Drug screening | Drug databases information | TPOTC | Estradiol, estriol, vitamins E and D, lynestrenol, mifrepristone, simvastatin, ambroxol, and some antibiotics and antiparasitics as potential drugs for PE |
ML, machine learning; PE, preeclampsia; LR, logistic regression; LDA, linear discriminant analysis; BPNN, back-propagation neural networks; GB, gradient boosting; EN, elastic net; RF, random forest; CNN, convolutional neural networks; SVM, support vector machines; AB, adaptative boosting; MLP, multilayer perceptron; SL, SuperLearner; TPOTC, tree-based pipeline optimization tool classifier; AUC, area under the receiver operating characteristic curve; Se, sensitivity; Sp, specificity; Ac, accuracy; DR, detection rate; PPV, positive predictive value; NPV, negative predictive value; AUPR, area under the precision-recall curve; Pr, precision.