Power spectrum entropy of acceleration time-series during movement as an indicator of smoothness of movement.

We propose a novel indicator for smoothness of movement, i.e., the power spectrum entropy of the acceleration time-series, and compare it with conventional indices of smoothness. For this purpose, nineteen healthy adults (21.3+/-2.5 years old) performed the task of raising and lowering a beaker between the level of the umbilicus and eye level under the two following conditions: one with the beaker containing water and the other with the beaker containing a weight of the same mass as the water. Moving the beaker up and down when it contained water required extra control to prevent the water from being spilled. This means that movement was not as smooth as when the beaker contained a weight. Under these two conditions, entropy was measured along with a traditional indicator of smoothness of movement, the jerk index. The entropy could distinguish just as well as the jerk index (p<0.01) between when water was used and when the weight was used. The entropy correlated highly with the jerk index, with Spearman's rho at 0.88 (p<0.01). These results showed that the entropy derived from the spectrum of the acceleration time-series during movement is useful as an indicator of the smoothness of that movement.