Automated Quality Assessment for Accelerometer-Based Heart Sounds Recorded with a Novel Subcutaneous Medical Implant

Area of Science:

• Biomedical Engineering
• Cardiology
• Signal Processing

Background:

• Automated assessment of heart sound quality is vital for accurate heart failure monitoring.
• Existing methods may struggle with noisy or complex physiological data.
• High-quality heart sound data ensures the relevance of medical analysis.

Purpose of the Study:

• To develop and validate a feature-based classification technique for heart sound recording quality.
• To assess the performance of the proposed method on data from a novel implantable device.
• To improve the reliability of heart sound analysis for heart failure management.

Main Methods:

• A quality classification technique using a small set of representative features.
• Features characterize periodicity, complexity, statistical nature, and detectability of heart sounds.
• Method applied to data from a subcutaneous implant combining ECG and accelerometric heart sound measurements in pigs.

Main Results:

• The method achieved 82% ± 2% sensitivity and 88% ± 6% accuracy in retrieving good quality heart sounds.
• Performance was evaluated on pre-clinical data from pigs, a challenging environment.
• Manual annotation classified recordings as "good", "unsure", and "bad" for reference.

Conclusions:

• The proposed automated method effectively classifies heart sound quality.
• Accurate recovery of high-quality heart sound sequences enables reliable monitoring of heart failure complications.
• This technique supports the clinical relevance of physiological indicators for conditions like decompensation.