World Sleep Industry ENCYCLOPEDIA

BCG mattress-type physiological monitoring devices

1.2.1

Name, Definition, and Device Boundary

A BCG mattress-type physiological monitoring device is a finished system that places one or more mechanical sensors under a sheet, within a mattress, under a mattress, on a bed frame, or beneath bed legs. It uses whole-body micro-motion caused by cardiac ejection, together with respiratory motion and gross movement transmitted through the bed, to estimate occupancy and selected physiological measures over time. A typical system includes the sensing structure, acquisition electronics, firmware or algorithms, a display or application, data storage, and optional networking and alarms. It commonly avoids skin electrodes and worn probes, but it still depends on a mechanical contact chain through the person, bedding, and bed.

“Mattress-type” describes an installation or integration form and does not necessarily mean that the product is itself a mattress. A thin strip may lie above or below the mattress, a sensing panel may attach to the frame, force sensors may sit under the legs, and a smart mattress may incorporate sensors and connections internally. These arrangements have different load paths, bedding compatibility, maintenance, and observable quantities. A shared category name does not imply that every device reports heart rate, respiratory rate, sleep stage, respiratory events, or clinical alarms.

This entry concerns a deployable finished device that delivers results to a user or care system and is distinct from the BCG non-contact monitoring module in Entry 197. A module may provide only raw or intermediate signals, whereas the finished device must also manage installation confirmation, subject attribution, quality decisions, result presentation, fault notification, privacy, and lifecycle support. A conventional pressure alarm pad, a switch mat that detects only bed exit, a therapeutic pressure-relief mattress, radar, a camera, a wearable ECG, and full polysomnography are separate device categories.

Measurement Chain and Installation Compatibility

A mattress BCG system records the composite mechanical waveform produced after heartbeat, breathing, posture, and movement propagate through the body, sheet, topper, mattress, frame, and mounting hardware. Mattress material, thickness, firmness, springs, foam or air-chamber damping, frame rigidity, leg contact, head elevation, body mass, height, sleeping posture, and position relative to the sensor all affect the result. Instructions should identify supported beds, mattress thickness or construction, mounting orientation, effective sensing area, user-weight range, and incompatible combinations rather than describing the sensor as universally compatible.

Installation should associate the device with the correct bed and side, secure the sensor against sliding, folding, or suspension, and protect cables from articulating frame mechanisms. Thick toppers, waterproof nursing pads, heated blankets, alternating-pressure mattresses, adjustable bases, split beds, and clinical-bed accessories may attenuate or transform the signal. Existing performance can be claimed only for combinations that were evaluated. Moving a device to another bed, replacing the mattress, changing a support, or repairing the frame should trigger the specified installation check and any required baseline or functional test.

An empty-bed baseline, entry stabilization, and bed-exit detection cannot depend on one universal amplitude threshold. Placed objects, a child or pet, a caregiver leaning on the bed, vibration from an adjacent bed, ventilation equipment, the floor, or a bed motor can alter load or create periodic motion. On a shared bed, two people’s cardiac and respiratory signals mix through the common structure. Unless zoning and identity attribution have been specifically validated, the device should be restricted to single-person use or label mixed results as unattributable; it should not automatically assign the stronger source to the account holder.

Output Levels and Physiological Meaning

The device may separate candidate mechanical heartbeats, respiratory rhythm, movement, and occupancy from the composite waveform, then calculate instantaneous or window-averaged heart rate, respiratory rate, movement counts, turns, bed exits, and trends. Each output should state its unit, calculation window, update period, latency, validity conditions, and quality flags. A continuously displayed number does not prove that the sensor continuously observed reliable beats. During motion, posture change, clipping, or poor signal-to-noise ratio, the correct behavior may be to mark the measure unavailable and suspend updates.

The BCG mechanical beat follows ECG electrical activation, and its beat intervals are not inherently identical to ECG RR intervals on every beat. Performance for average heart rate does not automatically support heart-rate variability, arrhythmia or atrial-fibrillation detection, stroke volume, cardiac output, or blood pressure. Performance for respiratory rate does not automatically support tidal volume, airflow, respiratory effort, oxygen saturation, apnea type, or apnea-hypopnea index. Each additional measure requires separate validation against a synchronized reference that actually measures the claimed quantity.

Sleep versus wake, sleep onset, wake after sleep onset, total sleep time, sleep efficiency, sleep stage, and a “sleep quality score” are additional model inferences. Quiet wakefulness can be labeled as sleep, and brief arousals can be hidden by movement smoothing. Without EEG, EOG, and chin EMG, the device does not directly observe the neurophysiological signals on which standard sleep staging depends. A validated output may support longitudinal trends or clinical discussion, but accuracy of heart rate and respiratory rate cannot be used as evidence that sleep staging is equally accurate.

Overnight Workflow and Quality Management

An overnight workflow should begin with device self-test, clock synchronization, bed and user association, an empty-bed baseline, entry detection, and initial stabilization. The record should preserve raw or traceable quality states that distinguish valid rest, motion contamination, sensor clipping, installation faults, mixed-person signals, power loss, network loss, and data gaps. A nightly summary should keep recording duration, time in bed, estimated sleep time, and time with valid physiological data as distinct denominators. Missing intervals must not be silently classified as stable sleep.

Body movement can obscure cardiac and respiratory components simultaneously, and the mattress or frame may continue ringing after movement stops. The algorithm must allow time to relock. Quality reporting should include valid coverage, longest continuous gap, time to first estimate, recovery after movement, and reasons for failure. An overnight average should be calculated only from qualifying segments and disclose their coverage. Interpolation, carrying forward a prior value, or using a person's history may make a graph look continuous, but it is not a measurement of the missing interval and must not drive a high-risk alarm.

A device for hospitals or care facilities must also handle shift changes, bed swaps, patient transport, bed vacancy, reinstallation after cleaning, and server outages. Heart-rate and respiratory-rate limits, bed-exit delays, and notification routes should be configurable by authorized personnel for the intended workflow, with controls that prevent prior-patient settings or data from following a bed assignment. Alarms should distinguish physiological limit violations, technical faults, and unavailable signals; log occurrence, delivery, acknowledgment, and resolution; and define local degradation when the network or central station is unavailable.

Performance Validation and Clinical Interpretation

Heart rate should be validated against synchronized ECG or an appropriate clinical monitor, respiratory rate against a synchronized reference suitable for the claimed respiratory quantity, and sleep or respiratory-event claims generally against polysomnography and its relevant channels. Evaluation should include interval-level error, Bland–Altman agreement or an appropriate equivalent, performance across rate ranges, valid coverage, missingness, and recovery rather than correlation or an all-night mean alone. The mattress device and reference must be time-aligned, and intervals rejected by the algorithm must still be counted in usability analysis.

Validation samples should represent intended ages, body sizes, bedding, postures, movement, diseases, rhythms, and respiratory patterns, with development and independent test sets reported separately. Results from a small group of motionless healthy volunteers, one mattress, or one laboratory night do not establish long-term home use, inpatient monitoring, abnormal-rhythm performance, respiratory-disorder performance, shared-bed operation, or compatibility with other frames. An aggregate mean can hide participants who produced no usable night; failure rates and exclusion conditions are therefore essential results.

Sleep findings cannot be transferred between products merely because both use BCG. Comparisons of consumer devices with polysomnography have found combinations of acceptable mean heart rate, high data loss, missed wakefulness, overestimated total sleep time, and failed stage discrimination, while other specific smart beds have performed better in particular samples. Evidence must remain linked to the exact hardware, software version, installation, algorithm, and population. A trend or screening output does not become a diagnosis by itself, and a negative output does not exclude a sleep, cardiac, or respiratory disorder.

Safety, Operations, and Reliability

A finished device requires assessment of power, leakage current, heating, electromagnetic compatibility, liquid ingress, flammability, mechanical strength, sharp edges, pinch points, cable entanglement, bed stability, and cleaning or disinfection. A sensing strip should not create a harmful ridge or pressure point, and frame or leg components must not impair load capacity or braking. Cables and sensors on an adjustable bed should not break, become entrapped, or generate false alarms during articulation. Products used around hospital patients, older adults, or people with limited mobility should also be tested with routine care and assistive equipment.

Reliability testing should address long-term compression and creep, repeated entry and exit, turning impacts, mattress flexion, temperature and humidity cycling, liquids and cleaning agents, power variation, radio congestion, full storage, and software restart. Verification after exposure should include sensitivity, noise, frequency response, clock behavior, bed-exit detection, overload recovery, and fault recognition rather than startup alone. Changes to the sensor, mattress material, mounting, front end, firmware, filters, model, or cloud algorithm require impact assessment and proportionate regression validation.

Safety of an alarmed product depends on the whole response loop, not sensor accuracy alone. False alarms, missed alarms, latency, repeated notification, silencing, escalation, staff response, and alarm fatigue must be evaluated, with explicit behavior during network loss, power failure, and sensor unavailability. The device must not obstruct direct observation or established monitoring. If frequent alerts lead users to pad over, move, disconnect, or disable a sensor, the system should detect that state or reduce the risk through better installation and alarm design.

Compliance, Privacy, and Use Limits

Regulatory obligations follow the final intended use, setting, outputs, and claims. A general-wellness product that records trends and an inpatient system that continuously measures heart and respiratory rates and generates limit or bed-exit alarms have different responsibilities even if both sit under a mattress. Authorization of one named device does not establish equivalence for a category. Each manufacturer must support its own labeling, alarms, population, hardware, software, and bed combination with evidence, while a consumer label must not imply medical-grade accuracy.

Occupancy, bed exits, routines, movement, heart rate, respiration, and inferred sleep states can create sensitive behavioral and health records. A connected product should minimize collection and protect device and account identity, transmission, storage, access, export, deletion, logs, and software updates, while stating how local functions depend on cloud service, subscription, and connectivity. A bed partner, visitor, or person without an account may also be passively recorded. Notice, permissions, data attribution, retention, and deletion should therefore be understandable to every affected user.

Without validation and authorization for the particular purpose, a mattress BCG device cannot diagnose arrhythmia, myocardial infarction, sleep apnea, or sleep stage and cannot serve as the sole means of infant safety, SIDS prevention, life support, emergency response, or nurse call. Chest pain, severe breathing difficulty, cyanosis, fainting, unresponsiveness, or another danger sign requires immediate action through the applicable emergency pathway; a person should not wait for an alarm or use a “normal night” to rule out disease. Persistent abnormal trends should be interpreted with symptoms and professional assessment.