Beyond the wearable

Updated 13 Oct 2026

Expansion should answer a question the current instruments cannot resolve. It can improve measurement of sleep or investigate what sleep affects. The validation investigation develops reference choices; the outcome investigation develops their connection to performance and health.

Select the instrument from the question

Burden is qualitative. No additional instrument or clinical test has been selected from the first night’s record.

Measuring circadian phase

Sleep onset is behaviour. Dim-light melatonin onset, or DLMO, uses serial samples under controlled dim lighting to investigate biological phase and its relationship to the sleep episode. Collection and assay conditions matter.

One evaluated home protocol required seventeen half-hourly saliva samples, 8.5 hours of dim-light conditions and laboratory assay afterward. Thirty-two of 35 participants completed it. Of valid home onsets, 58% were within thirty minutes of the paired laboratory onset on an adjacent night. That difference can include day-to-day variation as well as measurement error. This establishes a feasible measurement approach with substantial evening burden. NIH funded the study; authors reported no financial conflicts. Burgess et al.

Human bright-light experiments show that the direction of a phase shift depends on internal timing. A clock label such as “morning light” leaves part of the specification unresolved. Laboratory pulses are not an everyday exposure prescription. Khalsa et al.

Use phase measurement when it could change a timing decision. Ordinary sleep/light observations answer cheaper preliminary questions; DLMO adds a phase marker rather than a complete circadian-health score.

For light exposure, calibrated spectral measurement can estimate melanopic equivalent daylight illuminance. Conventional lux uses a different weighting; spectrum and exposure at the eye matter. A phone or desk-level reading is insufficient to certify the quantitative references. Measurement framework

Breathing needs events, not only averages

The first Oura respiratory average is approximately 12.4 breaths/min. It cannot show the event pattern, and lower is not inherently better.

Professional guidance distinguishes clinically indicated PSG and technically adequate home sleep apnea testing. A negative or inconclusive home test does not settle persistent suspicion. These assessments concern airflow, effort, oxygen and related events; a wearable average does not replace them. AASM diagnostic guideline

History or findings should establish the question before selecting a respiratory intervention. A preliminary mouth-taping study involved twenty selected mouth-breathers with mild obstructive sleep apnea. Two larger GSK-funded nasal-strip trials failed their primary subjective sleep/restfulness endpoints despite improvement in both active and placebo groups. Neither establishes a generic healthy-sleeper benefit. Mouth-taping pilot, nasal-strip trials

Supplements need an outcome and a population

A randomized magnesium-bisglycinate trial in 155 adults reporting poor sleep found a modest subjective insomnia-score benefit after four weeks: average ISI reductions were 3.9 versus 2.3 points, with a small effect size of 0.2. It did not measure objective stage improvement. The formulation supplied glycine as well as magnesium, so it cannot isolate their contributions. The study was university-funded; one author disclosed nutraceutical-industry research relationships. Schuster et al.

A glycine study during partial sleep restriction reported fatigue and vigilance benefits, but only seven participants were analysed in its single-blind crossover. Authors were affiliated with Ajinomoto. This does not show that broth with unknown glycine content reproduces the intervention or that glycine replaces sufficient sleep. Bannai et al.

In a defined delayed sleep-wake phase disorder trial, melatonin plus behavioural scheduling advanced sleep onset 34 minutes versus placebo plus scheduling. Phase-based selection and the clinical population limit transfer to healthy sleepers. NHMRC funded the trial, with recruitment support and industry relationships disclosed. Sletten et al.

Specify the intended outcome: nutritional correction, sleep initiation, phase shifting or waking performance. These cards describe evidence; no supplement intervention has been completed here.

EEG-guided stimulation

An actual research frontier is stimulation timed to brain activity. A small laboratory study found that correctly phase-timed auditory stimulation enhanced slow oscillations and declarative-memory consolidation. Open-loop clicks in another study did not improve word-pair consolidation. Ngo et al., Weigenand et al.

In a later study, oscillation and spindle activity increased in two experiments, including a double-blind near-replication, without improvement in the tested memory outcomes. Measure the physiological change and its intended functional effect separately. Henin et al.

A 2026 randomized five-group nap study retained 102 of 126 recruited participants. Stimulation altered evoked slow-wave/spindle activity without a detectable condition-by-time advantage on the declarative, motor-sequence or piano memory tasks. Behavioural analyses were exploratory; power was planned for physiology. Post-allocation sleep/technical exclusions and the two-hour nap opportunity limit transfer to overnight enhancement. FRQNT, NSERC and university support were reported, with no competing interests declared. Jourde et al.

A September 2026 preprint reports a selective benefit in biomarker-confirmed early symptomatic Alzheimer’s disease. In a blinded crossover, 32 participants contributed to the word-pair endpoint: after one stimulation night, overnight retention improved by an estimated 0.88 of thirty pairs versus sham. Benefits were not detected for image recognition, working memory or seven-day retention; follow-up recording failures reduced the latter comparison. Sleep architecture and subjective sleep gains were not detected. Bitbrain was the registered sponsor, many authors had company relationships, and public grants funded the work. The registry was submitted before the trial but became public after completion, limiting prospective verification. Healthy-sleeper transfer, durability and clinical importance remain unestablished. López-Larraz et al., trial registry

A 2024 alpha-phase stimulation trial found about ten minutes’ improvement in its first-visible-spindle latency measure among 21 analysed participants. Of 100 enrolled, 54 completed; data-quality and symptom exclusions narrowed the analysis further. Registration followed the experimental period. Its silent sham does not isolate phase locking from sound or expectation, and all authors were employees of the funder, Elemind. The operational definition of onset also matters. This is preliminary evidence for a selected phenotype and device protocol. Full trial

An experiment needs appropriate EEG, the tested timing logic and a functional outcome. Generic audio or a larger wearable deep-sleep estimate would test a different intervention.

Internal biomarkers: an informative example

A randomized suvorexant study in 38 cognitively unimpaired adults aged 45–65 found acute cerebrospinal-fluid amyloid-beta and one tau-phosphorylation ratio changes. The measured sleep outcomes were not significantly improved between groups. It used repeated sampling through a lumbar catheter. The lead author disclosed Merck consulting; diagnostics-related interests were also disclosed. Lucey et al.

The result concerns acute biomarkers rather than demonstrated dementia prevention or lifespan extension. A changed marker needs validation against the outcome it is meant to improve. The brain-clearance investigation develops that distinction and newer measurements.

Predictive models of sleep physiology

SleepFM learns from raw overnight brain/eye, cardiac, muscle and respiratory recordings. Its 2026 retrospective study discriminated subsequently recorded diagnoses in held-out patients. Predictive information can help define new measurement questions; it does not establish that altering sleep prevents those diagnoses. SleepFM primary study

The October version of the SleepFM-2 preprint extends the research to research-grade EEG and optical pulse recordings (PPG), plus wrist accelerometry. New cohorts used locally trained predictors on the shared model; Ring 5 applicability remains unverified. The primary PSG disease-prediction cohorts consist of sleep-clinic referrals; wrist-accelerometry prediction was also evaluated in UK Biobank. Calibration and useful clinical decisions remain to be established. NIH/philanthropic support and device-company relationships were disclosed. SleepFM-2 v2

Using a predicted risk score as an optimization target would require evidence that changing it through an intervention improves the health outcome. Agreement with future diagnoses alone cannot establish that relationship.

Experimental peptides and cannabinoids

In a sixteen-person double-blind chronic-insomnia study of delta sleep-inducing peptide, DSIP, authors judged effects weak and major therapeutic benefit unlikely. The older result provides little support for healthy optimization; funding remains unverified. Bes et al.

A two-week CBD/placebo insomnia pilot randomized 34 people and analysed 30 completers. Most sleep outcomes were similar; secondary actigraphy efficiency favoured CBD by about 6.9 percentage points at study end. Higher well-being scores already existed at screening, so that difference was not newly produced improvement. Cannvalate and the Barbara Dicker Brain Sciences Foundation funded the study; Brains Bioceuticals supplied treatments. Its short, selected clinical sample limits healthy-sleeper transfer. Narayan et al.

Retain negative studies alongside interesting mechanisms. Population, endpoint, comparator, duration, adverse effects and conflicts determine what a trial supports. No drug or peptide regimen has been selected.

The next useful expansion

Begin with a plausible constraint and a repeatable comparison. Add instrumentation when it resolves a consequential uncertainty, or an outcome when it tests transfer to life while awake. Experiment design connects those questions to a retain/reject decision.

All sleep investigations