Related Experiment Video
Updated: Jul 19, 2026
09:54
A Model to Simulate Clinically Relevant Hypoxia in Humans
Published on: December 22, 2016
8.9K
Prediction of breath-holding spells based on electrocardiographic parameters using machine-learning model
Mohammad Reza Khalilian1, Saeed Tofighi2, Elham Zohur Attar3
1Department of Pediatrics, School of Medicine, Shahid Beheshti University of Medical Sciences, Tehran, Iran.
Summary
Breath-holding spells (BHS) in children show distinct ECG repolarization changes. Machine learning models accurately predict BHS using these electrocardiogram (ECG) characteristics, aiding early diagnosis.
Area of Science:
- Pediatric Cardiology
- Computational Biology
Background:
- Breath-holding spells (BHS) are common in infants and children, often mimicking seizures.
- Autonomic dysfunction and iron deficiency anemia are potential contributing factors to BHS.
- Electrocardiographic (ECG) parameters in BHS patients require further investigation.
Purpose of the Study:
- To compare ECG parameters between children with BHS and healthy controls.
- To develop and evaluate machine learning (ML) models for predicting BHS using ECG data.
Main Methods:
- A case-control study involving 52 BHS patients and 150 healthy children.
- ECG recordings and clinical examinations were performed on all participants.
- Multivariate logistic regression and a Gradient-Boosting ML model were employed for analysis.
Main Results:
- No significant differences in mean heart rate, PR, or QRS intervals between groups.
- BHS patients exhibited significantly higher QTc, QTd, TpTe, and TpTe/QT ratios (p < .001).
- The ML model achieved an AUC of 0.94, with 90% specificity and 94% sensitivity for BHS prediction.
Conclusions:
- Repolarization abnormalities (elevated QTc, QTd, TpTe, TpTe/QT) are present in BHS patients.
- These ECG findings may indicate future arrhythmia risk.
- A successful ML model can predict BHS in suspected individuals.
