Postural and longitudinal variability in seismocardiographic signals
Executive Summary
This study investigates postural and longitudinal variability in seismocardiographic (SCG) signals using data from 19 healthy subjects over five months. SCG signals were acquired in three postures (supine, 45° tilt, and sitting) and analyzed using unsupervised machine learning (k-medoid clustering with dynamic time warping) to reduce respiratory-induced variability. Key findings include significant postural effects on SCG morphological variability and cardiac timing intervals, while spectral energy distribution remained stable across postures. Longitudinally, SCG features showed minimal variability, suggesting their potential for monitoring true cardiac changes over time.
Answer Machine Insights
Q: How does posture affect SCG signal variability?
Posture significantly impacts SCG morphological variability and cardiac timing intervals, with sitting posture showing higher variability compared to supine and 45° tilt.
Figure 10 suggests that both intra and inter-cluster variability tend to be higher in sitting position compared to supine and 45°.
Q: Are SCG features stable over time for longitudinal monitoring?
Yes, SCG features such as morphological variability, spectral energy distribution, and cardiac timing intervals remained stable over five months.
Table 5 suggests that the morphological variability (intra and inter-cluster variability) values between recording sessions were not significantly different for the study subjects.
Key Results
Postural changes significantly affect SCG morphological variability, with intra-cluster variability ranging from 1 to 2 milli g and inter-cluster variability from 2 to 5 milli g.
Longitudinal SCG features, including cardiac timing intervals and spectral energy distribution, showed no significant variability over five months.
Visual Evidence
Figure 15. Longitudinal SCG beat variation in a representative subject at different postures.
Clinical Snapshot
Evidence Rating
Relevance
high Priority