Estimación de la Frecuencia Cardíaca Basada en Autocorrelación Adaptativa a partir de Sismocardiograma de Eje Único:...

Área de la Ciencia:

• Biomedical Engineering
• Cardiovascular Physiology
• Signal Processing

Sus antecedentes:

• Seismocardiography (SCG) offers a non-invasive method for monitoring heart rate (HR).
• Existing SCG-based HR estimation methods often require ECG calibration or machine learning, limiting their applicability.
• There is a need for lightweight, automatic, and calibration-free algorithms for HR estimation from SCG.

Objetivo del estudio:

• To introduce and validate the Adaptive Autocorrelation Function Detector (AACFD), a novel pipeline for window-averaged HR estimation from single-axis SCG.
• To demonstrate AACFD's ability to operate without ECG calibration or machine learning, making it broadly accessible.
• To evaluate AACFD's performance across diverse datasets, including healthy subjects and patients with cardiovascular conditions.

Principales métodos:

• AACFD combines YIN-style difference-function analysis of an adaptive SCG envelope with short-window autocorrelation on 3-second segments.
• A feature-aware weighting strategy fuses the two detection branches, incorporating signal-quality indices and multi-scale features.
• Hampel filtering and temporal consistency checks are employed to remove outlier windows, ensuring robustness.

Principales resultados:

• AACFD achieved a mean absolute error (MAE) of less than 1 bpm on resting datasets (MCG, CEBS).
• Clinically acceptable MAEs were observed in valvular heart disease (VHD) and invasive right-heart catheterization (RHC) cohorts (4.38 bpm and 3.58 bpm on 30-s windows, respectively).
• Over 90% of non-overlapping 10-60 second windows passed quality control across all datasets, indicating high usability.

Conclusiones:

• The AACFD algorithm provides accurate window-averaged HR estimation from single-axis SCG, even in challenging clinical and mobile settings.
• Its lightweight and automatic nature allows for real-time implementation on commodity hardware.
• AACFD represents a significant advancement for non-invasive cardiac monitoring, particularly for applications where ECG is unavailable or impractical.