Definition, core signal, and category boundary
A wrist sleep monitor is a device centred on wrist accelerometry that records activity–rest patterns over consecutive days and applies a defined algorithm to estimate sleep–wake state and sleep timing. Professional actigraphs commonly emphasize long endurance, stable sampling, an event button, ambient light, raw data, and an auditable software workflow. Some models add PPG, temperature, or other sensors. Their initial observation of “sleep” remains an inference from activity rather than a direct brain recording.
The form overlaps a consumer sleep band and smartwatch, but the operational emphasis differs. A consumer product often provides a recovery score and lifestyle coaching, while a professional wrist monitor emphasizes protocol-based continuous wear, rest intervals, a sleep diary, non-wear detection, and batch export. Medical-device identity still depends on the exact model, intended use, functions, labelling, and jurisdiction. The word “monitor” does not establish independent diagnostic capability.
Wrist actigraphy, full polysomnography, a home sleep-apnea test, and an EEG headband are different examinations. Actigraphy is suited to observing sleep timing and circadian regularity over days or weeks. PSG directly records EEG, eye movement, muscle activity, breathing, and other physiology on selected nights. Before choosing a device, define whether the question concerns schedule, sleep quantity, respiratory events, limb movement, or neurophysiology; convenience cannot replace the signal required for that question.
Activity acquisition, wear, and non-wear classification
An accelerometer records wrist movement and forms activity counts over fixed epochs or preserves raw triaxial data. An algorithm commonly assigns sustained low movement to sleep and marked movement to wake, but sampling rate, dynamic range, epoch length, filtering, and sensitivity threshold shape the result. The same raw night processed by another algorithm or a low, medium, or high sensitivity setting can yield different sleep onset, awakening, and total sleep time.
The wrist side, position, and tension should remain fixed under the clinical or research protocol. Switching wrists, removal for bathing, charging at night, band slippage, sharing a bed with a person or animal, pushing a stroller, wheelchair movement, tremor, and periodic movement all alter the trace. Quiet sitting, reading, or lying awake with insomnia can be scored as sleep; frequent turning during sleep can be scored as wake. Actigraphy is therefore often better at finding sleep than quiet wake.
Non-wear is neither wake nor sleep. A sound workflow combines activity with temperature or capacitive input when available, event markers, diary entries, and human review, and preserves missingness. Consecutive zero movement can mean a device left on a table, device failure, or genuine immobility and cannot be decided from one threshold alone. Valid days, daily coverage, gaps, clock drift, and completeness of download should be reported.
Rest intervals, diary, and scoring workflow
An actigraphy algorithm first needs a rest interval representing attempted sleep or time in bed and then assigns sleep and wake within it. The interval may combine an event marker, sleep diary, ambient light, activity change, and predefined rules. One fixed clock interval applied to everyone truncates shift work, naps, split sleep, travel, and irregular schedules incorrectly. Boundary choice directly changes sleep-onset latency, total sleep time, and sleep efficiency.
A diary records bedtime, lights-out, attempted sleep, estimated sleep onset, awakenings, rising, naps, non-wear, caffeine, alcohol, and medication. It is not an inferior subjective version of actigraphy; it is independent context for interpreting immobility and missingness. An event button supplies a time anchor, but missed, accidental, or nocturnal presses must be checked against the diary and signal and cannot be treated mechanically as physiological sleep or wake.
The scoring workflow should preserve initialization, time zone, daylight-saving change, sampling configuration, rest-interval rule, algorithm name and version, sensitivity, and manual edits. Automated batch processing improves consistency but does not remove quality review. Fully manual dragging of intervals can add scorer variation. A reproducible analysis allows another qualified scorer using the same data and rules to obtain a similar result, with a documented reason for exceptions.
Metrics, circadian patterns, and interpretation limits
Common outputs include bedtime and rising, sleep onset and offset, total sleep time, onset latency, wake after sleep onset, sleep efficiency, naps, and 24-hour sleep. Multi-day data can also describe sleep midpoint, regularity, daytime activity, circadian amplitude, and phase trends. Each metric depends on rest-interval and algorithm definitions and is not exchangeable with the identically named value from a consumer app, diary, or PSG solely because the label matches.
Actigraphy does not directly acquire EEG, eye movements, or chin electromyography. By itself it generally supports sleep–wake classification, not precise assignment of N1, N2, N3, and REM. PPG can be added to predict stages, but that creates a new multisensor algorithm requiring separate validation. When sleep sensitivity is high but wake specificity is low, whole-night accuracy can still look impressive while sleep is systematically overestimated and quiet wake and fragmentation are underestimated.
Circadian metrics describe an activity–rest pattern and are not equivalent to endogenous phase markers such as melatonin or core body temperature. A light sensor usually measures illumination near the wrist and can be obscured by a sleeve, bedding, or direction; it does not automatically represent retinal light exposure. Work, caregiving, illness, pain, and social timing shape activity rhythms. An unusual plot is a contextual clue rather than a diagnosis of one circadian rhythm disorder.
Validation, clinical use, and diagnostic boundary
Validation should compare the exact device, configuration, and algorithm against synchronized PSG or an appropriate reference and report epoch-level sensitivity, wake specificity, error distribution, usable coverage, and failed recordings rather than correlation alone. Low activity in chronic illness, insomnia, frailty, or hospital care makes wake more likely to be scored as sleep. Laboratory results in healthy adults are not directly transferable. Evidence also needs rechecking after an algorithm update, rescaling of activity counts, or a change in wrist side.
Professional guidelines support qualified actigraphy as an adjunct in selected assessment of insomnia, circadian rhythm sleep–wake disorders, insufficient sleep, and sleep quantity before hypersomnolence testing. It can also be integrated with a home respiratory study to estimate total sleep time when another objective measure is absent. “Adjunct” and protocol-based multi-day use are essential. An actigram does not replace history and cannot independently diagnose obstructive sleep apnea or explain every cause of daytime sleepiness.
Actigraphy cannot replace leg electromyography for periodic limb movement disorder and does not detect airflow, respiratory effort, oxygen saturation, epileptiform discharges, or cardiac rhythm. Persistent insomnia, marked sleepiness, snoring with gasping, unusual sleep behavior, falls, palpitations, or sleepiness that threatens driving safety warrants professional selection of PSG, home respiratory testing, EEG, ECG, or another examination. A normal actigraphic report does not exclude those problems.
Wear, battery, data, and operational safety
The band should be stable without excessive tension or continuous pressure over the wrist and skin. Remove it after redness, itching, blistering, broken skin, pain, numbness, swelling, or colour change and seek care if severe or unresolved. Band materials, metal, sweat, cleaners, and friction can all irritate. A long recording protocol may allow brief removal for cleaning if the time is documented accurately; completeness should never be pursued through skin injury.
Charging should generally occur after removal on a hard, dry, ventilated surface away from bedding with a compatible supply. Stop use after battery swelling, abnormal heat or odour, cracked housing, corroded contacts, liquid entry, or abnormal behaviour after a fall. A water rating applies only to specified conditions. Whether bathing, swimming, or cleaning is allowed follows the exact instructions, and every removal remains missing data rather than an inferred state.
Several weeks of actigraphy can expose habitual sleep, shift work, naps, absence from home, activity, and light patterns and constitutes sensitive health and behavioural data. Use a unique strong password and multifactor authentication where offered. Restrict unnecessary location and third-party permissions and inspect local and cloud retention, sharing, export, deletion, and end-of-service plans. Research or clinical export should preserve raw data, diary, time zone, quality flags, and version. Unpair and erase data before transfer or disposal.
Selection, acceptance, and claim boundaries
Selection begins with the measurement task, not the screen or number of scores. Professional multi-day observation should prioritize raw triaxial data or activity counts, an event button, a light sensor, offline capacity, sampling configuration, clock stability, non-wear classification, diary integration, analysis software, device-specific validation, and regulatory status. Comfort, sizing, endurance, fleet management, export format, update support, subscription, warranty, and recalls also matter.
Initial acceptance can verify device identity, wrist side, time, and time zone and then test stillness, walking, removal, and event presses before downloading to confirm alignment of events, missingness, and raw signal. During the formal record, maintain the usual routine, complete the diary as directed, do not modify activity to create a good result, and do not fill a gap from memory. Review coverage and intervals for every date before analysis. Retain an atypical night with an annotation instead of deleting it silently.
Claims that wrist monitoring equals PSG, precisely identifies every sleep stage, independently diagnoses insomnia or sleep apnea, directly measures the body's biological clock, or records sleep without error are overstatements. Its defensible value is standardized, multi-day, reviewable description of sleep timing and regularity in ordinary life while preserving diary, missingness, and algorithm. Clinical conclusions still require a qualified professional to combine symptoms with the appropriate examination.