Deep learning identifies cardiac coupling between mother and fetus during gestation

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Maternal-Fetal Medicine

Background:

• Stillbirth causes approximately 2 million fetal deaths globally each year.
• Current fetal monitoring methods like ultrasound have limitations including safety concerns, need for skilled personnel, and high costs in developing regions.

Purpose of the Study:

• To introduce "deep coherence," a novel AI approach for non-invasive fetal monitoring using electrocardiography (ECG).
• To assess the association between maternal and fetal heartbeats during pregnancy.

Main Methods:

• Utilized a deep learning tool trained on 941 one-minute maternal-fetal R-peak segments from 172 pregnant women (20-40 weeks gestation).
• Employed non-invasive electrocardiography (ECG) to capture maternal and fetal heart signals.

Main Results:

• The AI tool achieved 90% accuracy in identifying maternal-fetal heartbeat coupling scenarios.
• Demonstrated the interpretability of deep learning in elucidating synchronization mechanisms between maternal and fetal heartbeats.

Conclusions:

• Deep coherence shows significant potential as an AI-driven tool for frequent and continuous fetal development evaluation.
• This AI approach could be integrated into clinical practice to reduce costs and side-effects associated with current fetal monitoring devices.