Features were extracted from seismocardiogram (SCG) data to correlate with stroke volume changes, estimated from arterial pressure recorded from the finger. Stroke volume was gradually reduced in twenty nine human subjects using lower body negative pressure. Twenty three features were extracted from SCG amplitudes and timings. There was at least one feature in every subject with correlation coefficient of 0.87or greater (P-value <0.01). The ratio of left ventricular pre-ejection interval to left ventricular ejection time (LVPEI/LVET) proved to correlate the most with stroke volume (r=-0.92±0.06) compared to the other features. The second most correlated SCG feature, with stroke volume, was LVPEI with (r=−0.90±0.12). The findings of this paper suggest that with simple processing of SCG it should be possible to detect a sudden drop in stroke volume that could result from a variety of different cardiac abnormalities.
展开▼