Automatic detection of artifacts in heart period data

Area of Science:

• Biomedical Engineering
• Cardiovascular Physiology
• Data Science in Healthcare

Background:

• Accurate analysis of heart period (RR interval) data is crucial for physiological studies.
• Manual artifact removal is impractical for large datasets, necessitating automated solutions.
• Existing artifact detection algorithms have limitations in performance and specificity.

Purpose of the Study:

• To develop and evaluate an efficient, automatic algorithm for artifact rejection in RR interval data.
• To compare the performance of the new algorithm against five existing artifact detection methods.
• To assess the algorithm's effectiveness in detecting real and simulated artifacts in infant data.

Main Methods:

• Developed a novel automatic artifact rejection algorithm using a +/-20% criterion and a median of 25 surrounding RR intervals.
• Tested the algorithm on 30-minute electrocardiogram and RR interval recordings from 10 infants.
• Quantified performance using sensitivity, specificity, and accuracy, employing both real and simulated artifacts.

Main Results:

• The new algorithm demonstrated superior performance in detecting real artifacts compared to existing methods.
• Achieved a high balance of sensitivity (73.46%) and specificity (99.17%), significantly improving specificity.
• Showed excellent sensitivity for detecting missing beats (100%) and extra beats (97.44%), even with noisy data.

Conclusions:

• The developed automatic artifact rejection algorithm is highly effective for RR interval analysis, particularly in noisy infant data.
• It offers a significant improvement in specificity while maintaining good sensitivity, crucial for large-scale data analysis.
• This algorithm enhances the reliability of physiological data analysis in research settings like the Collaborative Home Infant Monitoring Evaluation (CHIME) study.