Beat-to-beat estimation of LVET and QS2 indices of cardiac mechanics from wearable seismocardiography in ambulant subjects
Executive Summary
This study evaluates an improved algorithm for heart beat detection using smartphone-based seismocardiography (SCG) signals. The methodology involves preprocessing, RMS envelope calculation, and peak detection, achieving an average sensitivity (Se) of 0.994 and positive predictive value (PPV) of 0.966 across signals from four subjects. The findings demonstrate significant performance improvements over previous studies, highlighting the potential of smartphone SCG for accurate heart monitoring.
Answer Machine Insights
Q: What is the sensitivity and positive predictive value achieved by the algorithm?
The algorithm achieved an average sensitivity (Se) of 0.994 and positive predictive value (PPV) of 0.966.
We achieved average Se = 0.994 and PPV = 0.966 and in the best case Se = 0.995 and PPV = 0.970.
Q: What factors contributed to lower performance on female subjects?
Lower sensitivity on female subjects may be caused by smartphone size and position.
Lower sensitivity of beat detection on signals from female subject may be caused by smartphone size and position, as indicated by Landreani et al. [11].
Key Results
Average sensitivity (Se) of 0.994 and positive predictive value (PPV) of 0.966.
Best-case performance: Se = 0.995, PPV = 0.970.
Visual Evidence
Figure 3: Heart beat detection algorithm flowchart.
Clinical Snapshot
Evidence Rating
Relevance
high Priority