Related Experiment Video
Updated: Oct 9, 2025
06:56
Human Fetal Blood Flow Quantification with Magnetic Resonance Imaging and Motion Compensation
Published on: January 7, 2021
2.5K
Passive Fetal Movement Recognition Approaches Using Hyperparameter Tuned LightGBM Model and Bayesian Optimization.
Sensong Liang1, Jiansheng Peng1,2, Yong Xu2
1College of Electronic Engineering, Guangxi Normal University, Guilin 541004, China.
Computational Intelligence and Neuroscience
|December 20, 2021
Summary
This study introduces a novel system for accurate fetal movement detection using Kalman filtering and machine learning. The advanced approach enhances prenatal monitoring for high-risk pregnancies, improving fetal health assessment.
Area of Science:
- Biomedical Engineering
- Wearable Health Monitoring
- Signal Processing
Background:
- Fetal movement is a critical indicator of fetal well-being during pregnancy.
- Noninvasive fetal movement detection systems are crucial for remote monitoring of high-risk pregnancies.
- Accurate recognition of fetal movements amidst noise is a significant challenge in wearable health monitoring.
Purpose of the Study:
- To develop an efficient and accurate system for fetal movement recognition.
- To address the challenges of signal recovery and noise reduction in fetal movement detection.
- To enhance the reliability of wearable smart sensing systems for prenatal health monitoring.
Main Methods:
- Utilized Kalman filtering (KF) for fetal movement signal recovery from noisy backgrounds.
- Extracted features from time, frequency, and wavelet domains (TFWD) of the preprocessed signals.
- Employed a hyperparameter-tuned Light Gradient Boosting Machine (LightGBM) model optimized via Bayesian Optimization Algorithm (BOA).
Main Results:
- The proposed KF+TFWD+BOA-LGBM approach achieved high recognition accuracy (94.06%) and F1-Score (96.85%) on the Zenodo fetal movement dataset.
- Demonstrated superior accuracy and robustness compared to eight existing advanced methods.
- Successfully enabled accurate prediction and recognition of fetal movements.
Conclusions:
- The developed system offers a robust solution for fetal movement recognition in wearable health monitoring.
- The integration of Kalman filtering, TFWD features, and optimized LightGBM shows significant potential for clinical application in prenatal care.
- This technology can improve the monitoring of high-risk pregnancies and contribute to better fetal health outcomes.
