Effective monitoring of fetal heart activity during the second and third trimesters remains a vital challenge in perinatal medicine. This study proposes an adaptive algorithm for extracting the fetal electrocardiograms signal from abdominal ECG in pregnant women, considering the physiological characteristics of each trimester. Utilizing modern machine learning methods, independent component analysis, and data from wearable textile electrodes. The goal is to enhance the accuracy and reliability of automatic signal separation. A dataset of 300 recordings will be collected and analyzed. The resulting algorithm will enable rapid and precise detection of fetal heartbeats. To validate the algorithm, 50 patients will be recruited separately.
Research Objective Development and validation of an algorithm for separating maternal and fetal electrocardiographic signals based on non-invasive abdominal ECG in pregnant women during the second and third trimesters of gestation. Research Tasks 1. Perform abdominal ECG recordings in pregnant women using a non-invasive technology, ensuring standardized recording conditions and accounting for gestational age. Each recording should contain at least 5-10 minutes of continuous signals, providing sufficient data volume for analysis and algorithm training. 2. Analyze features of abdominal ECG signals at various gestational stages, including morphology of maternal and fetal rhythms, their degree of overlap, and the influence of physiological factors. Compare findings with clinical history and other diagnostic methods. 3. Develop and adapt an algorithm for separating maternal and fetal electrocardiographic signals, considering the specific features during the second and third trimesters, to enhance the accuracy of fetal cardiac activity diagnosis based on machine learning. 4. Evaluate the diagnostic parameters of the algorithm for assessing the fetal condition
Study Type
INTERVENTIONAL
Allocation
NA
Purpose
OTHER
Masking
NONE
Enrollment
350
Sensors are attached to the pregnant woman's abdomen on pre-prepared sites, and data are recorded for at least 10 minutes. Afterwards, the ECG signals are processed to remove noise.
V.F. Snegirev Clinic of Obstetrics and Gynecology of I.M. Sechenov First Moscow State Medical University
Moscow, Russia
RECRUITINGCorrelation coefficient between automatically extracted fetal heart rates and reference. signals
Сardiotocography (CTG) will be used as a reference.
Time frame: Through study completion, an average of 1 year
Signal processing time and computational complexity of the algorithm.
The signal processing time refers to the duration required for the algorithm to analyze and process the input signals, including steps such as filtering, noise removal, feature extraction, and data alignment.
Time frame: Through study completion, an average of 1 year
Accuracy of R-peak detection: number of correctly identified fetal heartbeats (sensitivity) and number of false positives (specificity).
The accuracy of R-peak detection refers to the algorithm's ability to correctly identify fetal heartbeats within the recorded signals. Sensitivity (true positive rate) indicates the proportion of actual fetal heartbeats that were correctly detected by the algorithm. Specificity (true negative rate or false positive rate) reflects the number of false detections, i.e., instances where non-heartbeat signals were incorrectly identified as fetal heartbeats. High sensitivity and specificity are essential for reliable fetal heart rate monitoring, minimizing missed beats and false alarms.
Time frame: Through study completion, an average of 1 year
Proportion of rejected or invalid segments where the algorithm failed to reliably extract fetal data.
The proportion of rejected or invalid segments refers to the percentage of data segments in which the algorithm was unable to reliably extract fetal heart rate information. These segments are typically excluded from analysis due to poor signal quality, noise, or other artifacts that prevent accurate detection of fetal data.
Time frame: Through study completion, an average of 1 year
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.