Enhanced labor pain monitoring using machine learning and ECG waveform analysis for uterine contraction-induced pain

Abstract

Objectives: This study aims to develop an innovative approach for monitoring and assessing labor pain through ECG waveform analysis, utilizing machine learning techniques to monitor pain resulting from uterine contractions.

Methods: The study was conducted at National Taiwan University Hospital between January and July 2020. We collected a dataset of 6010 ECG samples from women preparing for natural spontaneous delivery (NSD). The ECG data was used to develop an ECG waveform-based Nociception Monitoring Index (NoM). The dataset was divided into training (80%) and validation (20%) sets. Multiple machine learning models, including LightGBM, XGBoost, SnapLogisticRegression, and SnapDecisionTree, were developed and evaluated. Hyperparameter optimization was performed using grid search and five-fold cross-validation to enhance model performance.

Results: The LightGBM model demonstrated superior performance with an AUC of 0.96 and an accuracy of 90%, making it the optimal model for monitoring labor pain based on ECG data. Other models, such as XGBoost and SnapLogisticRegression, also showed strong performance, with AUC values ranging from 0.88 to 0.95.

Conclusions: This study demonstrates that the integration of machine learning algorithms with ECG data significantly enhances the accuracy and reliability of labor pain monitoring. Specifically, the LightGBM model exhibits exceptional precision and robustness in continuous pain monitoring during labor, with potential applicability extending to broader healthcare settings.

Trial registration: ClinicalTrials.gov Identifier: NCT04461704.

Keywords: Artificial Intelligence (AI); Clinical Decision-Making; Electrocardiography (ECG); Healthcare Technology; Labor Pain; Machine Learning; Nociception Assessment; Pain Monitoring; Predictive Analytics; Uterine Contractions.

Fig. 1

Workflow Diagram for ECG-Based Labor Pain Monitoring. This figure illustrates the methodological workflow employed in this study for monitoring labor pain using ECG waveform analysis. The process consists of seven key steps: 1) Data Collection: Gathering ECG data from women pre-paring for natural spontaneous delivery (NSD); 2) Feature Selection: Identifying critical ECG features correlated with labor pain; 3) Data Splitting: Dividing the data into training and validation sets; 4) Model Development: Building machine learning models using the selected features; 5) Model Evaluation: Assessing model performance with metrics such as AUC, accuracy, and precision; 6) Model Optimization: Tuning model parameters through grid search and cross-validation; 7) Monitoring and Validation: Conducting final predictions and validating the optimized model's performance to ensure accurate and reliable labor pain monitoring

Fig. 2

A Characteristic curves for receiver operations. and (B) Assessing Machine Learning Models' Performance. For every machine learning model, performance metrics such as AUC (Area Under the Curve), F1 Score, Accura-cy, Specificity, Sensitivity, and Precision were evaluated. An asterisk (*) denotes the LightGBM model's AUC value

Fig. 3

Crucial Clinical Characteristics for Labor Pain Monitoring Based on ECG. A Features' Significance Plot of several clinical variables utilizing the Nociception Monitoring Index (NoM) and Machine Learning for ECG-Based Monitoring of Labor Pain. B The SHAP Summary Plot offers a concise synopsis of the key clinical characteristics that influence the Nociception Monitoring Index (NoM) and ECG-Based Monitoring of Labor Pain

Fig. 4

Effects of Key Elements on ECG-Based Labor Pain Monitoring. This figure displays LIME (Local Interpretable Model-agnostic Explanations) force graphs to illustrate the impact of important aspects on the ECG-Based Monitoring of Labor Pain. Specifically, the Nociception Monitoring Index (NoM) and various frequency bins are shown to influence the probability of labor pain. These plots provide individual-level explanations by visually representing how each variable contributes to the likelihood of labor pain, with red bars indicating a decrease and green bars indicating an increase in the probability