Universal Pre-Ejection Period Estimation Using Seismocardiography: Quantifying the Effects of Sensor Placement and...

Area of Science:

• Biomedical Engineering
• Cardiology
• Signal Processing

Background:

• Seismocardiography (SCG) measures chest vibrations for non-invasive cardiac assessment.
• Systolic time intervals, like the pre-ejection period (PEP), derived from SCG can indicate cardiac contractility.
• SCG-based PEP estimation shows potential for classifying heart failure (HF) patients, but signal variability poses challenges.

Purpose of the Study:

• To improve the accuracy of pre-ejection period (PEP) estimation from seismocardiography (SCG) signals.
• To investigate the impact of sensor placement and timing features on PEP estimation accuracy.
• To identify optimal regression models and sensor locations for robust PEP estimation.

Main Methods:

• Extracted timing features from SCG signals recorded at multiple upper body positions.
• Employed global regression models, including XGBoost, to estimate PEP.
• Evaluated model performance using root-mean-square error (RMSE) across different sensor placements and combinations.

Main Results:

• Ensemble regression using XGBoost achieved the best performance with an RMSE of 11.6 ± 0.4 ms.
• A combination of sensors on the sternum and below the left clavicle yielded the most accurate PEP estimates.
• Sensor placement below the clavicle (left or right) outperformed traditional sternum placement for PEP estimation.

Conclusions:

• Optimized sensor placement and advanced regression techniques significantly enhance PEP estimation accuracy using SCG.
• This approach offers a more reliable non-invasive method for assessing cardiac function and potentially diagnosing heart failure.
• Multi-positional SCG sensing provides a promising avenue for personalized cardiac monitoring.