Background
Methods
Study setting and data collection
The outcome variable & eligibility criteria
Sign | 0 | 1 | 2 |
---|---|---|---|
Heart rate | Absent | < 100 | ≥100 |
Respiratory effort | Absent | Weak cry, hypoventilation | Good, crying |
Reflex irritability | No response | Grimace | Cry or active withdrawal |
Muscle tone | Limp | Some flexions of extremities | Active motion |
Color | Blue, pale | Body pink, extremities blue | Completely pink |
Feature importance and variable correlation
Imbalanced learning model establishment
Machine learning analysis
Accuracy | \(\frac{TP+ TN}{TP+ TN+ FP+ FN}\) | (1) |
Precision | \(\frac{TP}{TP+ FP}\) | (2) |
Recall | \(\frac{TP}{TP+ FN}\) | (3) |
F1 score | \(\frac{2\ast Precision\ast Recall}{Precision+ Recall}\) | (4) |
MCC | \(\frac{TP\ast TN- FP\ast FN}{\sqrt{\left( TP+ FP\right)\ast \left( TP+ FN\right)\ast \left( TN+ FP\right)\ast \left( TN+ FN\right)}}\) | (5) |
BA | \(\frac{TP}{2\left( TP+ FN\right)}+\frac{TN}{2\left( TN+ FP\right)}\) | (6) |
BM | \(\frac{TP}{TP+ FN}+\frac{TN}{TN+ FP}-1\) | (7) |
MK | \(\frac{TP}{TP+ FP}+\frac{TN}{TN+ FN}-1\) | (8) |
Decision curve analysis (DCA)
Results
Maternal characteristics | Low (< 7) Apgar score | Normal (≥7) Apgar score | χ2 p-value |
---|---|---|---|
Parity | |||
Nulliparous | 409 (55.8) | 3817 (54.66) | |
Multiparous | 324 (44.2) | 3166 (45.34) | 0.556 |
Maternal age | |||
< 25 | 273 (37.24) | 2575 (36.88) | |
25–35 | 361 (49.25) | 3606 (51.64) | |
> 35 | 99 (13.51) | 802 (11.49) | 0.214 |
Gestational age | |||
Term | 463 (63.17) | 5683 (81.38) | |
Preterm | 209 (28.51) | 593 (8.49) | |
Post term | 61 (8.32) | 707 (10.12) | < 0.001 |
PROM | |||
No | 709 (96.73) | 6829 (97.79) | |
Yes | 24 (3.27) | 154 (2.21) | 0.067 |
Gestational diabetes | |||
No | 730 (99.59) | 6974 (99.87) | |
Yes | 3 (0.41) | 9 (0.13) | 0.067 |
Prenatal visits | |||
< 3 | 296 (40.38) | 1796 (25.72) | |
3–6 | 365 (49.80) | 3997 (57.24) | |
> 6 | 72 (9.82) | 1190 (17.04) | < 0.001 |
Induction method | |||
Oxytocin | 591 (80.63) | 6361 (91.09) | |
Prostaglandins | 142 (19.37) | 622 (8.91) | < 0.001 |
Referred for delivery | |||
No | 453 (61.80) | 5573 (79.81) | |
Yes | 280 (38.20) | 1410 (20.19) | < 0.001 |
Ever Use of Family planning | |||
No | 344 (46.93) | 2896 (41.47) | |
Yes | 389 (53.07) | 4087 (58.53) | 0.004 |
Smoking during pregnancy | |||
No | 729 (99.45) | 6966 (99.76) | |
Yes | 4 (0.55) | 17 (0.24) | 0.135 |
Alcohol during pregnancy | |||
No | 550 (75.03) | 4977 (71.27) | |
Yes | 183 (24.97) | 2006 (28.73) | 0.032 |
Child sex | |||
Female | 412 (56.21) | 3563 (51.02) | |
Male | 321 (43.79) | 3420 (48.98) | 0.008 |
Body mass index | |||
Underweight | 2 (0.27) | 27 (0.39) | |
Normal | 109 (14.87) | 1262 (18.07) | |
Overweight | 455 (62.07) | 4133 (59.19) | |
Obese | 167 (22.78) | 1561 (22.35) | 0.169 |
Epilepsy | |||
No | 732 (99.86) | 6961 (99.68) | |
Yes | 1 (0.14) | 22 (0.32) | 0.399 |
Preeclampsia | |||
No | 717 (97.82) | 6873 (98.42) | |
Yes | 16 (2.18) | 110 (1.58) | 0.217 |
Feature importance
Correlation matrix for predictors of low Apgar score following as successful labor induction (IOL)
Algorithm | Metrics | Before resampling | SMOTE | ROSE (Oversampling) | ROSE (undersampling) | ROSE (Hybrid) |
---|---|---|---|---|---|---|
Logistic regression | Accuracy | 0.91 | 0.80 | 0.72 | 0.80 | 0.73 |
AUC | 0.69 | 0.66 | 0.69 | 0.66 | 0.70 | |
Recall | 0.12 | 0.43 | 0.53 | 0.43 | 0.51 | |
Precision | 0.79 | 0.22 | 0.18 | 0.22 | 0.18 | |
F1-score | 0.21 | 0.29 | 0.27 | 0.29 | 0.27 | |
MCC | 0.29 | 0.20 | 0.18 | 0.20 | 0.17 | |
BA | 0.56 | 0.64 | 0.63 | 0.64 | 0.63 | |
BM | 0.11 | 0.27 | 0.27 | 0.27 | 0.27 | |
MK | 0.71 | 0.15 | 0.12 | 0.15 | 0.12 | |
Neural networks | Accuracy | 0.92 | 0.79 | 0.80 | 0.79 | 0.73 |
AUC | 0.70 | 0.67 | 0.70 | 0.70 | 0.69 | |
Recall | 0.16 | 0.43 | 0.42 | 0.47 | 0.53 | |
Precision | 0.77 | 0.21 | 0.22 | 0.22 | 0.18 | |
F1-score | 0.26 | 0.28 | 0.29 | 0.30 | 0.27 | |
MCC | 0.33 | 0.20 | 0.20 | 0.21 | 0.18 | |
BA | 0.58 | 0.63 | 0.63 | 0.65 | 0.64 | |
BM | 0.15 | 0.26 | 0.26 | 0.29 | 0.28 | |
MK | 0.70 | 0.14 | 0.15 | 0.16 | 0.12 | |
Random forest | Accuracy | 0.91 | 0.85 | 0.90 | 0.81 | 0.88 |
AUC | 0.68 | 0.66 | 0.69 | 0.69 | 0.70 | |
Recall | 0.12 | 0.34 | 0.22 | 0.46 | 0.30 | |
Precision | 0.84 | 0.26 | 0.46 | 0.24 | 0.33 | |
F1-score | 0.21 | 0.29 | 0.30 | 0.32 | 0.31 | |
MCC | 0.30 | 0.21 | 0.27 | 0.23 | 0.24 | |
BA | 0.56 | 0.62 | 0.69 | 0.65 | 0.62 | |
BM | 0.11 | 0.24 | 0.19 | 0.31 | 0.24 | |
MK | 0.76 | 0.19 | 0.38 | 0.17 | 0.26 | |
Naïve Bayes | Accuracy | 0.91 | 0.84 | 0.79 | 0.78 | 0.79 |
AUC | 0.69 | 0.67 | 0.71 | 0.69 | 0.70 | |
Recall | 0.25 | 0.40 | 0.47 | 0.48 | 0.47 | |
Precision | 0.56 | 0.26 | 0.21 | 0.21 | 0.22 | |
F1-score | 0.35 | 0.29 | 0.20 | 0.29 | 0.30 | |
MCC | 0.33 | 0.23 | 0.21 | 0.21 | 0.22 | |
BA | 0.61 | 0.64 | 0.64 | 0.65 | 0.65 | |
BM | 0.23 | 0.28 | 0.29 | 0.26 | 0.30 | |
MK | 0.49 | 0.19 | 0.15 | 0.15 | 0.16 | |
Boosting | Accuracy | 0.92 | 0.86 | 0.79 | 0.75 | 0.78 |
AUC | 0.73 | 0.70 | 0.74 | 0.71 | 0.74 | |
Recall | 0.17 | 0.36 | 0.54 | 0.52 | 0.53 | |
Precision | 0.78 | 0.29 | 0.23 | 0.19 | 0.22 | |
F1-score | 0.28 | 0.32 | 0.32 | 0.28 | 0.31 | |
MCC | 0.34 | 0.25 | 0.25 | 0.20 | 0.24 | |
BA | 0.58 | 0.64 | 0.68 | 0.65 | 0.67 | |
BM | 0.16 | 0.27 | 0.35 | 0.29 | 0.33 | |
MK | 0.70 | 0.22 | 0.17 | 0.13 | 0.16 | |
Bagging | Accuracy | 0.91 | 0.79 | 0.88 | 0.66 | 0.84 |
AUC | 0.68 | 0.67 | 0.67 | 0.67 | 0.67 | |
Recall | 0.19 | 0.37 | 0.22 | 0.58 | 0.30 | |
Precision | 0.52 | 0.19 | 0.33 | 0.16 | 0.23 | |
F1-score | 0.28 | 0.25 | 0.26 | 0.25 | 0.26 | |
MCC | 0.27 | 0.16 | 0.21 | 0.15 | 0.17 | |
BA | 0.58 | 0.61 | 0.59 | 0.63 | 0.60 | |
BM | 0.17 | 0.21 | 0.17 | 0.25 | 0.19 | |
MK | 0.44 | 0.12 | 0.25 | 0.10 | 0.15 |