Model-based detection of heart rate turbulence using mean shape information

Area of Science:

• Cardiology
• Signal Processing
• Biomedical Engineering

Background:

• Heart rate turbulence (HRT) is a significant indicator of cardiovascular health.
• Existing methods for HRT detection have limitations in accuracy and sensitivity.
• Understanding the relationship between heart rate variability (HRV) and HRT is crucial for improved analysis.

Purpose of the Study:

• To propose a novel generalized likelihood ratio test (GLRT) statistic for enhanced heart rate turbulence (HRT) detection.
• To improve HRT detection by utilizing a priori information about HRT shape.
• To compare the performance of the new GLRT detector against previous methods and the turbulence slope (TS) parameter.

Main Methods:

• Developed a new GLRT statistic using Karhunen-LoEve basis functions to model HRT.
• Employed an extended integral pulse frequency modulation model to account for ectopic beats and HRT.
• Investigated the spectral relationship between HRV and HRT to model noise, considering a white noise assumption.
• Evaluated detector performance using both simulated and real cardiovascular data.

Main Results:

• The new GLRT detector demonstrated superior performance compared to the original GLRT detector and the turbulence slope (TS) parameter.
• For simulated data with averaged ten ventricular ectopic beats, the new detector achieved a detection probability of 0.83 at a 0.1 false alarm probability.
• The original GLRT detector and TS achieved detection probabilities of 0.35 and 0.41, respectively, under the same conditions.

Conclusions:

• The proposed GLRT statistic offers a significant advancement in HRT detection accuracy.
• Incorporating HRT shape information enhances detector sensitivity and reliability.
• The new detector provides a more robust tool for assessing cardiovascular health through HRT analysis.