I’m a neuroscientist and just read their “phase 2b protocol” paper. As far as I can tell, there are no actual control groups. Registering a trial with clinicaltrials.gov is a nice step, but they don’t compare at all to patient improvements in people that don’t receive treatment let alone a sham. Thus, the study does not appear designed to prove efficacy even though they will surely claim it.
The trial was divided into three groups. Group 1 got simultaneous tongue stimulation and audio tone and each tone was correlated to a separate electric lead on the device. Groups 2 and 3 had no association between the tones and which leads were activated (I think they were just random, or they used them all? Can't tell) Group 2 had a small delay between tone and stimulation, group 3 had a longer one (0.5s+) and also used a much narrower subset of tones. All 3 groups showed improvement and any differences were well within the error bands.
Definitely a sloppy design, changing so many variables between the groups. No active control (maybe vibration on the tongue or something). To me the conclusion is that the general method probably works, but almost everything the actual device is overkill, from the 32 different leads to the "proprietary algorithm" and you could do the same thing with just one lead and probably some random tone played over airpods.
Maybe it does something, but tinnitus tends to simple improve over time, too: [1]. Again not a great study as unclear what treatment options people sought, but they have zero evidence of efficacy as I understand it. It’s inaccurate to call these “double-blind” because there is no control group.
