Refined matrix completion for spectrum estimation of heart rate variability

Area of Science:

• Cardiovascular Health
• Biomedical Signal Processing
• Machine Learning

Background:

• Heart rate variability (HRV) analysis is crucial for assessing cardiovascular health and autonomic nervous system function.
• Spectral analysis of HRV provides key insights but is susceptible to data artefacts that compromise signal quality and reliability.
• Existing methods struggle to accurately quantify uncertainties in HRV spectral estimations.

Purpose of the Study:

• To introduce a novel matrix completion-based approach for estimating uncertainties in HRV spectral analysis.
• To enhance the accuracy and computational efficiency of HRV uncertainty estimation.
• To validate the proposed method against existing deep learning models.

Main Methods:

• Utilized the low-rank properties of HRV spectrum matrices for uncertainty estimation.
• Developed a refined matrix completion technique to improve accuracy and reduce computational load.
• Benchmarked the proposed model on five public HRV datasets.

Main Results:

• The matrix completion model effectively and reliably estimated uncertainties in HRV spectrum.
• Demonstrated superior performance compared to five leading deep learning models.
• Confirmed the robustness of the approach across diverse datasets.

Conclusions:

• The developed matrix completion-based statistical machine learning model offers a reliable solution for HRV spectrum uncertainty estimation.
• This method has the potential to significantly improve the accuracy of cardiovascular health monitoring.
• The findings highlight the utility of matrix completion in biomedical signal processing applications.