Supplementary Material S4. TRIPOD+AI Reporting Checklist The Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) Statement was updated in 2024 to include Artificial Intelligence (TRIPOD+AI). This checklist ensures complete reporting of machine learning-based clinical prediction models. ┌──────────────────────────────────────────────────────────────────────────────┐ │ TITLE/ABSTRACT │ │ □ AI1. Identify the study as developing and/or validating a multivariable │ │ prediction model, the target population, and the outcome to be predicted. │ │ ✓ Done: Title states "Prediction of Cesarean Delivery in Multiparous │ │ Women Following Prior Vaginal Birth: A Machine Learning Study" │ │ │ │ □ AI2. Indicate the machine learning methods used. │ │ ✓ Done: Abstract and Methods specify XGBoost │ └──────────────────────────────────────────────────────────────────────────────┘ ┌──────────────────────────────────────────────────────────────────────────────┐ │ INTRODUCTION │ │ □ AI3a. Explain the medical context (including whether diagnostic or │ │ prognostic) and rationale for developing or validating the prediction │ │ model, including references to existing models. │ │ ✓ Done: Introduction discusses CS rate rise, prior VB as lower-risk │ │ population, and scarcity of ML models with temporal validation. │ │ │ │ □ AI3b. Describe the target population, setting, intended use, and │ │ intended moment of model application. │ │ ✓ Done: Multiparous women with prior VB, antenatal counseling setting. │ └──────────────────────────────────────────────────────────────────────────────┘ ┌──────────────────────────────────────────────────────────────────────────────┐ │ METHODS │ │ │ │ Data sources │ │ □ AI4a. Describe the sources of data (e.g., electronic health records, │ │ prospective cohort). │ │ ✓ Done: Electronic birth records 2013-2025. │ │ │ │ □ AI4b. Specify key dates (start, end, model development, validation). │ │ ✓ Done: Training 2013-2022, validation 2023-2025. │ │ │ │ Participants │ │ □ AI5a. Describe eligibility criteria and flow of participants. │ │ ✓ Done: Figure 1 study profile; first VB required, consecutive deliveries. │ │ │ │ Outcome │ │ □ AI6. Clearly define the outcome, including type and time horizon. │ │ ✓ Done: Intrapartum CS (VB as reference), second delivery. │ │ │ │ Predictors │ │ □ AI7a. Describe all candidate predictors before analysis. │ │ ✓ Done: 49 candidate variables (Table 1); 37 final predictors (S1). │ │ │ │ □ AI7b. Report the rationale for predictor exclusion, including label │ │ leakage, reverse causation, and antenatally unknown variables. │ │ ✓ Done: Table 1 lists 3 exclusion categories (4+6+2=12). │ │ │ │ □ AI7c. Describe any feature engineering or dimension reduction. │ │ ✓ Done: Binary encoding for categorical; continuous kept as-is. │ │ │ │ Sample size │ │ □ AI8. Report the sample size, number of events, and events per variable. │ │ ✓ Done: EPV = 8.2 in training set. │ │ │ │ Model development │ │ □ AI9a. Describe the machine learning algorithm and version. │ │ ✓ Done: XGBoost 3.2.0. │ │ │ │ □ AI9b. Describe hyperparameter search strategy and final configuration. │ │ ✓ Done: S2 provides complete hyperparameter table with rationale. │ │ │ │ □ AI9c. Describe internal validation method. │ │ ✓ Done: Temporal validation (2013-2022 train, 2023-2025 validate). │ │ │ │ □ AI9d. Describe any handling of class imbalance. │ │ ✓ Done: scale_pos_weight adjusted for class imbalance. │ │ │ │ Risk groups / score derivation │ │ □ AI10. Describe any simplification (e.g., simplified score derivation). │ │ ✓ Done: 0-100 simplified score based on Top 8 SHAP features. │ │ │ │ Handling missing data │ │ □ AI11. Describe methods for handling missing data. │ │ ✓ Done: Missing values excluded by design (complete case for 37 features). │ │ │ │ Statistical methods │ │ □ AI12a. Describe performance measures (discrimination, calibration). │ │ ✓ Done: AUC with bootstrap CI; van Houwelingen calibration. │ │ │ │ □ AI12b. Describe calibration method in detail. │ │ ✓ Done: Logistic recalibration on logit(predicted probability). │ │ │ │ □ AI12c. Describe any subgroup or sensitivity analyses. │ │ ✓ Done: S3 provides complete subgroup AUCs by age, interval, │ │ complications, and ART. │ └──────────────────────────────────────────────────────────────────────────────┘ ┌──────────────────────────────────────────────────────────────────────────────┐ │ RESULTS │ │ │ │ Participants │ │ □ AI13a. Report flow of participants. │ │ ✓ Done: Figure 1. │ │ │ │ □ AI13b. Report baseline characteristics. │ │ ✓ Done: Table 3 by delivery outcome. │ │ │ │ Model specification │ │ □ AI14. Report final model and predictor importance. │ │ ✓ Done: Table 2, Figure 3 SHAP, S1 complete ranked list. │ │ │ │ Model performance │ │ □ AI15a. Report discrimination measures with confidence intervals. │ │ ✓ Done: AUC 0.785 (95% CI 0.739-0.827). │ │ │ │ □ AI15b. Report calibration measures. │ │ ✓ Done: Slope = 0.912, intercept = -1.819. │ │ │ │ □ AI15c. Report decision curve analysis. │ │ ✓ Done: Figure 2D. │ │ │ │ Simplified score │ │ □ AI16. Report simplified score performance. │ │ ✓ Done: AUC 0.785 [0.742, 0.826]. │ │ │ │ Sensitivity analyses │ │ □ AI17. Report sensitivity analyses. │ │ ✓ Done: Figure 4 (bootstrap, subgroup, threshold, feature subset). │ └──────────────────────────────────────────────────────────────────────────────┘ ┌──────────────────────────────────────────────────────────────────────────────┐ │ DISCUSSION │ │ │ │ □ AI18a. Discuss limitations, including label leakage safeguards. │ │ ✓ Done: Explicit discussion of second_ga/ga_gap exclusion and AUC impact. │ │ │ │ □ AI18b. Discuss generalizability and need for external validation. │ │ ✓ Done: Single-center limitation, calibration transportability. │ │ │ │ □ AI18c. Discuss potential clinical utility and implementation. │ │ ✓ Done: Recalibration requirement, simplified score for bedside use. │ │ │ │ □ AI18d. Discuss whether predictor effects are consistent with prior │ │ knowledge. │ │ ✓ Done: Macrosomia, breech, and interval findings align with literature. │ └──────────────────────────────────────────────────────────────────────────────┘ ┌──────────────────────────────────────────────────────────────────────────────┐ │ OTHER INFORMATION │ │ │ │ □ AI19. Provide registration details and data/code availability. │ │ N/A: Single-institution study; data sharing requires ethics approval. │ │ │ │ □ AI20. Provide funding and conflict of interest disclosures. │ │ N/A: [To be completed by authors] │ └──────────────────────────────────────────────────────────────────────────────┘ Checklist completion: 28/29 items completed (96.6%)