20/20 Vision · Evidence

Science behind the brake

Every number on the landing page has a named source. The kill-test is written before the first night.

The problem, measured

Nocturnal scratch in atopic dermatitis is not a vague complaint. Cesnakova et al. mapped a patient-centred conceptual model: about 65% report nocturnal scratch at least one night per week, and 1–1.4 hours of sleep are reported lost on an affected night. Sugiyama et al. measured the sleep architecture cost: 35.5% of scratch episodes shift toward lighter sleep or wake.

The mechanism, measured

Scratch has a rhythmic biomechanical signature that wearables can separate from ordinary sleep motion. Yang, Patel, Chun et al. (JAMA Dermatology, 1 Apr 2025) closed an AI-enabled wearable loop with a haptic cue in mild AD. Mean total scratch per night fell 28%; mean duration 40%; events per hour of sleep opportunity 39%; duration per hour of sleep opportunity 50%. That is the existence proof for detect + interrupt. It is not yet an accessible consumer-watch loop with an adaptive, rate-limited cue and a clinician morning receipt — that is the wedge 20/20 Vision occupies.

Must fall

Scratch events / night, total scratch seconds, seconds per sleep hour.

Must not worsen

Awakenings / arousals, subjective sleep quality, sleep opportunity.

Named sources

20/20 Vision does not re-claim those percentages as our trial. They are the reason a 7-night home pilot is a sane next step rather than a slogan.