Assessing the Effectiveness of Heart Rate Variability as A Diagnostic Tool for Brain Injuries in Infants

Area of Science:

• Neonatal neurology
• Biomedical signal processing
• Machine learning in healthcare

Background:

• Hypoxic-Ischemic Encephalopathy (HIE) is a critical condition in neonates caused by oxygen deprivation to the brain.
• Current diagnostic methods like electroencephalogram (EEG) have limitations.
• Exploring alternative physiological signals like heart rate variability (HRV) is crucial for improved HIE assessment.

Purpose of the Study:

• To investigate the utility of HRV for neonatal seizure detection.
• To evaluate HRV's effectiveness in grading the severity of Hypoxic-Ischemic Encephalopathy (HIE).
• To compare machine learning classifier performance for these tasks.

Main Methods:

• Utilized two annotated clinical datasets of electrocardiogram (ECG) signals from neonates.
• Calculated, denoised, and segmented Heart Rate (HR) from ECG.
• Extracted 16 time and frequency domain HRV features.
• Applied Random Forest (RF), Support Vector Machine (SVM), and Isolation Forest (IF) classifiers.

Main Results:

• Seizure detection using HRV showed promising results in specific patient-independent cases but requires further refinement (AUC 68.54%).
• HIE grading using HRV achieved a higher Area Under the Curve (AUC) of 77.13%.
• The study highlights the potential of HRV as a diagnostic tool for HIE severity.

Conclusions:

• HRV analysis demonstrates significant potential as a non-invasive tool for grading neonatal Hypoxic-Ischemic Encephalopathy severity.
• Further research is needed to enhance HRV-based seizure detection accuracy and explore its correlation with EEG.
• The findings suggest HRV could complement existing methods in neonatal neurological assessment.