In sleep laboratories, a technician wires your head with a couple of dozen sensors to create a polysomnogram (PSG), graphing not only brain-wave activity but also eye movements, heart rhythm, and muscle activation. Since this level of detail is not particularly realistic for a consumer device, the Zeo makes use of a clever shortcut. It scans just brainwaves, but analyzes them with a painstakingly devised neural-net algorithm. In a study in which 10 sleepers wore a Zeo headband as well as a full set of polysomnography sensors, the Zeo's results were substantively comparable to the results of the PSG as read by two separate trained technicians. For instance, in monitoring the 10 sleepers, one human technician scored the group's average number of minutes of REM sleep at 62.4; the other put it at 56.6; and the Zeo's algorithm came up with 59.9 minutes. The Zeo isn't nearly as nuanced as a polysomnogram, but it seems able to fake it pretty well for home-use purposes.