Model-based posturographic sleepiness monitor tested on 20 subjects.

A previous posturographic force platform study verified that human balance deteriorates as a function of time awake (TA). It found that TA can be estimated with +/-2.5 h accuracy using 30 s trial length. For a fast, reliable and convenient sleepiness monitor even better TA estimation accuracy and shorter trial length is needed. We continued this quest by modeling the quiet stance test situation with a single-link inverted pendulum model (SLIPM). The center-of-mass (COM) trace in the AP (anterior-posterior) direction was calculated from the measured center-of-pressure (COP) trace. The sway angle theta, ankle torque tau and the horizontal force F acting on the COM were calculated from the SLIP equations--each analyzed with 13 different sway measures, i.e. variables that correlate with TA. The effect of circadian rhythm was separated from the sway measures. Twenty subjects' sway measures were analyzed. The SLIPM-based posturographic averaged TA estimation accuracy improved to +/-2.3 h. The trial length could be shortened to 21 s.