A multimodal deep learning-based algorithm for specific fetal heart rate events detection.

Area of Science:

• Medical technology
• Artificial intelligence in healthcare
• Fetal monitoring

Background:

• Fetal heart rate (FHR) monitoring is crucial for assessing fetal well-being.
• Accurate detection of specific FHR events and deceleration patterns is vital for timely clinical intervention.
• Current monitoring methods can be labor-intensive and require expert interpretation.

Purpose of the Study:

• To develop a multimodal deep learning algorithm for automated detection of specific FHR events.
• To enhance the intelligent assessment of fetal well-being through advanced signal analysis.
• To improve the accuracy and efficiency of classifying various FHR abnormalities and deceleration types.

Main Methods:

• Analysis of FHR and uterine contraction signals using combined feature extraction techniques.
• Integration of morphological, heart rate variability, and nonlinear domain features with deep learning models.
• Development of a multi-model deep neural network and a pre-fusion deep learning model for signal classification.

Main Results:

• High classification accuracy for FHR events: 96.2% for acceleration, 94.4% for deceleration, 90.9% for tachycardia, and 85.8% for bradycardia.
• Successful classification of four distinct deceleration patterns with 67.0% overall accuracy.
• Specific accuracies for deceleration patterns include 80.9% for late deceleration and 98.9% for prolonged deceleration.

Conclusions:

• The developed multimodal deep learning algorithm provides reliable decision support for clinicians.
• The algorithm significantly improves the detection and assessment of critical FHR events.
• This technology is crucial for advancing fetal health monitoring and improving patient outcomes.