Hidden Markov model-based heartbeat detector using electrocardiogram and arterial pressure signals

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiovascular Monitoring

Background:

• Electrocardiogram (ECG) alone has limitations for reliable heartbeat detection due to noise and data loss.
• Combining multiple physiological signals offers a more robust approach to accurately detect heartbeats.

Purpose of the Study:

• To develop and evaluate a novel heartbeat detection system integrating information from ECG, arterial blood pressure (ABP), and pulmonary arterial pressure (PAP) signals.
• To leverage hidden Markov models (HMMs) for joint signal processing to enhance heartbeat detection accuracy.

Main Methods:

• Physiological signals (ECG, ABP, PAP) were preprocessed.
• A sliding window approach extracted observation sequences for analysis.
• Two trained hidden Markov models (HMMs) were employed to calculate log-likelihoods for detection based on an adaptive threshold.

Main Results:

• The HMM-based detector achieved high performance across multiple databases (MIT-BIH Arrhythmia, PhysioNet Challenge 2014, MGH/MF Waveform).
• Sensitivity and positive predictivity exceeded 99% on two databases and 95% on another, using ECG and ABP signals.
• Detection performance was found to be comparable to existing state-of-the-art methods.

Conclusions:

• The proposed HMM-based approach effectively integrates multi-signal physiological data for accurate heartbeat detection.
• This method offers a reliable and robust alternative to traditional ECG-only detection, especially in the presence of signal noise or missing data.