Name, Definition, and Category Boundary
A contactless heart-rate and respiration monitoring pad is a pad-shaped device placed above a mattress, between the mattress and its support, or at another specified part of a bed. It estimates heart rate and respiratory rate from small mechanical changes transmitted through bedding and bed structure. “Contactless” generally means that no electrode, optical probe, or respiratory belt is attached directly to the skin. It does not mean that the sensor is mechanically uncoupled from the mattress or that it can work through arbitrary materials. The pad, acquisition unit, algorithm, installation, and host form one measurement chain.
The principal boundary of this category is concurrent or ongoing output of heart rate and respiratory rate, sometimes accompanied by in-bed state, bed exit, movement, or trends. A pressure switch that reports occupancy only, a sensing sheet that exports raw pressure only, a narrow under-mattress sensor strip, a sensor built into an entire smart mattress, and a remote radar are not automatically the same category because their use cases overlap. A component that requires integration and lacks a finished algorithm and user output is closer to a sensing module than to a complete monitoring pad.
A pad may support general wellness records, care assistance, remote observation, or a defined medical use, but its name does not establish accuracy, alarm reliability, or regulatory status. Documentation should identify intended age and weight ranges, supported bed and mattress conditions, single- or multiple-person limitations, output parameters, use environment, and operator. An evaluation board or brief research prototype cannot share the mechanical design, electrical safety, software configuration, or evidence of a production product by assumption.
Sensing Principle and Meaning of Measurement
Common pads use piezoelectric film or ceramic, electromechanical film, strain or pressure elements, load cells, accelerometers, force-sensitive resistors, or microbend optical fibers to convert tiny bed forces, displacement, velocity, or acceleration into an electrical signal. Recoil of the body associated with cardiac ejection can form a ballistocardiographic or bed-mechanical cardiac component. Chest and abdominal excursion creates a lower-frequency respiratory component, while larger movement creates an activity signal. Sensor material, measurement axis, and placement substantially alter waveform morphology and sensitivity.
The pad senses mechanical effects conducted through the body, bedding, mattress, and frame; it does not directly measure cardiac electrical activity or oral and nasal airflow. A “heart rate” estimated from mechanical pulse intervals does not necessarily identify every ECG R wave and cannot by itself diagnose arrhythmia. A respiratory rate estimated from periodic bed movement gives no direct measure of tidal volume, airway patency, oxygen saturation, or gas exchange. Cardiac, respiratory, bed-resonance, and movement components can overlap, so a separated algorithm output needs an accompanying quality and validity state.
Additional outputs may include beat intervals, heart-rate variability, an apnea clue, sleep stage, or sleep score. Each adds inference beyond the underlying mechanical signal. Good performance for mean heart rate does not establish the accuracy of beat-to-beat intervals, and recognition of a respiratory rhythm does not establish clinical respiratory-event scoring or an apnea-hypopnea index. Each derived output needs its own analysis window, reference method, invalid-data rule, population, and intended use. Evidence for heart or respiratory rate should not be borrowed for an extra metric outside its validation scope.
Placement, Bedding, and Mechanical Coupling
The pad must be installed as specified: under the thoracoabdominal region, below or above the mattress, or at a defined frame location, with correct orientation, flatness, fixation, and cable exit. Lateral displacement, rotation, curling, folding, partial suspension, and migration during the night change the load path. A small sensing area may also lose coverage after a person turns. Commissioning should check supine, lateral, prone, head-elevated, and customary sleep positions rather than ending after one motionless supine trial.
Mattress thickness, firmness, composition, layers, latex or memory-foam damping, springs, protectors, toppers, heated bedding, and sheets all alter mechanical transmission. Slats, solid decks, adjustable bases, motors, air or water mattresses, and separately supported sides of a double bed add different support and resonance behavior. Accuracy shown for one pad-mattress combination cannot be transferred directly to another. Mattress replacement, a new thick topper, or bed-frame adjustment calls for a new installation check and any required baseline or calibration procedure.
Sensing elements usually have a useful preload and dynamic range. A very light load may provide inadequate micromotion, whereas concentrated weight or frame clamping may saturate the element. A soft mattress can attenuate higher-frequency cardiac components while preserving stronger respiration. Temperature, humidity, material aging, adhesive release, and long-term compression can also shift sensitivity. The system should distinguish open or short circuit, overrange, saturation, and abnormal baseline from a genuinely empty bed, a low heart rate, or slow breathing.
Motion, Multiple Occupants, and Failure States
Turning, sitting up, coughing, speaking, limb movement, and bed adjustment are generally much larger than the mechanical cardiac signal. They can saturate a waveform and leave bed resonance after movement stops. A reliable system detects movement, suspends untrustworthy estimation, waits for mechanical settling, and reacquires before marking data valid. Continuously displayed smoothed numbers do not prove continued measurement through movement. Performance reports need valid coverage, the longest gap, initial lock time, and recovery time after motion.
A partner on a double bed transmits cardiac, respiratory, and movement forces through the shared mattress and frame. Zoned pads, multiple channels, or spatial models can improve attribution, but crosstalk remains possible when people are close, cross a zone, move together, or have similar rates. Visitors, children, pets, a caregiver leaning on the bed, and objects placed on the mattress can change load or be classified as a target. The product should state whether it supports one person or a defined two-person arrangement and suppress personalized output when identity is unresolved.
Fans, washing machines, traffic, loudspeakers, phone vibration, motorized beds, building machinery, and motion in an adjacent bed can conduct periodic and transient vibration through floor or frame. An algorithm should not call a periodic peak in a plausible frequency band a vital sign without other evidence. Empty-bed background, pad displacement, loose cable, acquisition-unit restart, clock error, and network loss also need distinct technical states. “No data,” “no person in bed,” and “zero breaths detected” have fundamentally different safety meanings.
Algorithms, Validation, and Performance Reporting
Processing commonly includes detrending, band-pass filtering, motion detection, time-frequency analysis, template matching, or channel selection, followed by rate estimation over windows lasting seconds to tens of seconds. A longer window may stabilize an estimate but slows the response to change, while strong smoothing can erase a short event. The algorithm should retain the actual input quality, selected channel, window, update period, and version. It should not create false continuity during low quality by silently interpolating or carrying forward the last value.
Heart-rate validation should use synchronized ECG and assess beat intervals, short-window rate, and overnight average separately. Respiratory-rate validation needs a synchronized reference appropriate to the claim, such as airflow, thoracoabdominal effort, carbon dioxide, or an explicitly scored reference. High correlation with a reference does not establish agreement for an individual, and a small average bias can hide large errors during rapid rates, movement, or in a subset of subjects. Time alignment, error tolerance, exclusions, and invalid-window rules should be set before analysis.
The validation sample should cover intended ages, body sizes, sleep positions, mattress combinations, normal and abnormal rates, weak signals, overnight movement, and real home conditions. Training, tuning, and independent test data should be separated. Alongside rate error, reports need coverage, lost intervals, false estimates, reacquisition time, and completely failed nights. The production pad, acquisition unit, firmware, host, and mounting accessories require system-level validation. Changes in material, dimensions, sensor position, filtering, or model should trigger testing proportionate to their impact.
Use, Maintenance, Electrical Safety, and Connectivity
Installation should follow load, flatness, bend-radius, frame-clearance, and cable-retention limits. The pad should not be folded, punctured, soaked, or trapped in a moving mechanism. Its power adapter, acquisition box, and cable should not create trip, entanglement, pressure-injury, overheating, or fire hazards or obstruct emergency bed adjustment. Compatibility needs confirmation with heated bedding, adjustable beds, medical electrical equipment, and cleaning agents. An intact outer surface does not prove that internal sensing layers and insulation remain intact.
Routine checks should cover position, flatness, connections, jacket damage, moisture, abnormal heat, baseline, and self-test status. Cleaning must match the material and ingress rating; being under a mattress does not make a product maintenance-free. Moving the bed, replacing the mattress, repairing the frame, updating firmware, or returning after long storage calls for reinspection. Service-life evidence should address repeated loading, flexing, temperature and humidity cycles, and cable fatigue. A damaged pad should be removed from service, not concealed by software recalibration.
A connected pad can create sensitive records of occupancy, nightly routines, movement, and cardiac and respiratory trends. The system should minimize collection and implement device and account authentication, protection in transit and at rest, access control, logs, controlled updates, deletion, and an end-of-support plan. Offline caching and retransmission must preserve time and identity and remove duplicates. When cloud service or phone notifications fail, documentation should state whether local measurement and alarms continue and how users learn that the system is degraded.
Alerts, Sleep Interpretation, and Medical Limits
If the device offers high or low heart-rate, abnormal-breathing, bed-exit, or inactivity alerts, it should separate physiological thresholds from technical conditions such as pad displacement, inadequate signal, low battery, or network failure. Delay, hysteresis, pause, escalation, and recipient need definition. Bed-conducted mechanical signals may not support a safety-critical alert while a person is out of bed, moving strongly, or mixed with another occupant. A wellness notification is not hospital surveillance and must not imply continuous staffing; delivery of a remote message does not prove that anyone saw or acted on it.
Heart-rate and respiratory-rate trends can help describe overnight patterns, but they cannot alone stage sleep, distinguish obstructive from central respiratory events, measure oxygen saturation, or diagnose sleep apnea. Mechanical cardiac motion cannot replace a diagnostic ECG or reliably classify arrhythmia without separate evidence. Consumer products, research prototypes, and devices authorized for a defined medical setting should be described by their own evidence. Results from one named system do not prove that all monitoring pads perform alike.
A pad without validation and authorization for the purpose must not serve as an infant monitor, life-support system, nurse call, fall-prevention system, or sole monitor for a high-risk patient. It also does not replace adult supervision, safe-sleep practices, or prescribed monitoring. Chest pain, severe breathing difficulty, cyanosis, fainting, repeated gasping awakenings, or another danger sign requires the local emergency or clinical pathway. Help should not be delayed because the pad did not alert or showed a normal trend.