Real-time monitoring and nursing equipment for vital signs of bedridden patients based on multimodal sensing

Abstract

This invention proposes a real-time monitoring and nursing device for vital signs of bedridden patients based on multimodal sensing. It includes a bed for supporting the patient; a touch screen fixedly installed on the side edge of the bed; and a multimodal sensing pad laid on the bed and located under the patient's torso and limbs. The multimodal sensing pad comprises, from bottom to top, an electromagnetic shielding layer, a flexible pressure sensing array layer, and a dielectric elastomer sensing layer, etc. By directly laying the multimodal sensing pad on the bed and integrating the flexible pressure sensing array layer, dielectric elastomer sensing layer, and temperature sensing fiber layer, this invention can simultaneously collect data on body pressure distribution, chest and abdominal respiratory deformation, cardiac impact vibration, and body surface temperature. Combined with the fusion analysis submodule in the edge computing unit, it achieves time alignment between body movement events and physiological waveforms, dynamically adjusting heart rate when body movement is detected, significantly reducing the false alarm rate caused by the patient turning over or limb movement.

Description

Technical Field

[0001] This invention relates to the field of medical and nursing equipment technology, and in particular to a nursing device for real-time monitoring of vital signs of bedridden patients based on multimodal sensing. Background Technology

[0002] In healthcare settings, monitoring the vital signs of bedridden patients is a core component of preventing complications, assessing changes in their condition, and providing timely medical intervention. Traditional monitoring methods often rely on nurses' regular rounds or the use of discrete monitoring equipment. These methods have several limitations: firstly, frequent manual contact increases the burden on healthcare workers and disrupts patients' rest; secondly, traditional patch electrodes or strap-on sensors can easily cause skin discomfort and allergies in patients with prolonged use, and sensors are prone to detachment or signal interruption when patients turn over or move. More critically, existing monitoring equipment often only measures single parameters such as heart rate, respiration, or body temperature independently, making it difficult to obtain spatiotemporal correlation information between body movement, position, respiratory waveforms, and cardiac impact signals. When patients move, traditional algorithms are prone to misinterpreting motion artifacts as abnormal physiological signals, generating frequent false alarms and severely disrupting clinical workflows. Designing a device that can avoid causing additional restraint and skin irritation to patients, and can collect multimodal signals such as chest and abdominal respiratory movements, cardiac impact, body surface pressure distribution, and body temperature in a long-term, stable, and synchronous manner, and intelligently distinguish between body motion interference and real physiological abnormalities, has become an urgent technical problem to be solved. Summary of the Invention

[0003] In view of this, in order to solve the problems existing in the technical background, the present invention proposes a real-time monitoring and nursing device for vital signs of bedridden patients based on multimodal sensing. Specifically, it includes the following: A real-time monitoring and nursing device for vital signs of bedridden patients based on multimodal sensing, comprising a bed for supporting the patient; A touch screen display is fixedly installed on the side edge of the bed frame; And a multimodal sensing pad laid on the bed and located under the patient's torso and limbs; The multimodal sensing pad includes an electromagnetic shielding bottom layer, a flexible pressure sensing array layer, a dielectric elastomer sensing layer, a temperature sensing fiber layer, and a skin-friendly and breathable surface encapsulation layer, arranged in sequence from bottom to top. The flexible pressure sensing array layer consists of multiple capacitive pressure sensing units arranged in a matrix, used to detect the pressure distribution on the patient's body surface to identify body position and body movement signals. The dielectric elastomer sensing layer is used to collect periodic deformation signals of the chest and abdomen caused by the patient's breathing and weak impact vibration signals caused by the heartbeat. The temperature sensing fiber layer is composed of interwoven flexible thermosensitive fibers and is used to collect data on the temperature distribution of the patient's body surface. The electromagnetic shielding layer integrates an edge computing unit, a wireless communication module, and a local storage module. The edge computing unit is electrically connected to the flexible pressure sensing array layer, the dielectric elastomer sensing layer, and the temperature sensing fiber layer, respectively, and is used to perform time synchronization, feature extraction, and fusion analysis on the acquired multimodal sensing signals, and calculate the patient's real-time heart rate, respiratory rate, body movement rate, body position classification, and average body surface temperature based on the fused signals. The edge computing unit is also electrically connected to the touch screen, for displaying the calculated vital signs parameters in real time and issuing abnormal warning signals. The wireless communication module is electrically connected to the edge computing unit and is used to upload real-time vital signs parameters and raw sensor data to the remote nursing monitoring center.

[0004] The edge computing unit includes a signal conditioning submodule, a multi-channel synchronous acquisition submodule, a feature extraction submodule, and a fusion analysis submodule. The signal conditioning submodule receives the capacitance change signal output from the flexible pressure sensing array layer, the charge change signal output from the dielectric elastomer sensing layer, and the resistance change signal output from the temperature sensing fiber layer, and performs filtering and amplification processing. The multi-channel synchronous acquisition submodule performs synchronous analog-to-digital conversion on the three conditioned signals at a sampling rate of no less than 100Hz and adds timestamp tags. The feature extraction submodule extracts features from the synchronized pressure signal. The start and end times of body movement events and the body pressure distribution feature vector are used to separate the respiratory waveform and cardiac impaction waveform from the dielectric elastomer signal through adaptive bandpass filtering, and the respiratory rate and heart rate are calculated respectively. The resistance values ​​of each sensor node are extracted from the temperature signal and converted into body surface temperature values. The fusion analysis submodule aligns the timestamps of body movement events with the respiratory waveform and cardiac impaction waveform in time sequence. When body movement is detected, the signal-to-noise ratio thresholds for respiratory and heart rate calculations are dynamically adjusted. Based on the body pressure distribution feature vector, the preset supine position classification template is matched to output the body position classification results. At the same time, the number of body movements and the average body surface temperature within the preset time window are counted.

[0005] Furthermore, the dielectric elastomer sensing layer includes a dielectric elastomer film with flexible electrodes coated on its upper and lower surfaces. The initial pre-stretch rate of the dielectric elastomer film is set to 10%-20%. A silicone buffer layer with a thickness of 0.5mm-1.5mm is sandwiched between the flexible pressure sensing array layer and the dielectric elastomer sensing layer. The silicone buffer layer is used to diffuse the concentrated pressure on the patient's body surface into a uniformly distributed pressure and transmit it to the dielectric elastomer sensing layer to avoid local pressure overload causing the capacitance output of the dielectric elastomer film to saturate.

[0006] In some embodiments of the present invention, the local storage module built into the electromagnetic shielding layer adopts a circular storage mechanism, retaining only the most recent 72 hours of multimodal raw sensor data and vital sign parameters output by the edge computing unit. When the edge computing unit detects that the connection between the wireless communication module and the remote nursing monitoring center is interrupted, it continuously writes real-time data into the local storage module and automatically uploads all stored data after the connection is restored.

[0007] The above technical solution has the following beneficial effects: This invention achieves non-contact vital sign monitoring for patients by directly laying a multimodal sensing pad on the bed, eliminating the need for any additional devices and avoiding the skin irritation and discomfort associated with traditional electrodes and straps. By integrating a flexible pressure sensing array layer, a dielectric elastomer sensing layer, and a temperature sensing fiber layer, it can simultaneously collect data on body pressure distribution, chest and abdominal respiratory deformation, cardiac impact vibration, and body surface temperature. Combined with the fusion analysis submodule in the edge computing unit, it achieves time alignment between body movement events and physiological waveforms, dynamically adjusting the signal-to-noise ratio thresholds for heart rate and respiration calculations when body movement is detected, significantly reducing the false alarm rate caused by patient turning over or limb movement. It utilizes body pressure distribution feature vectors to automatically match recumbent position classification templates, achieving intelligent position recognition. The silicone buffer layer protects the dielectric elastomer film from local pressure overload, ensuring the stability of cardiac impact signals during long-term use; the local storage module's cyclic storage and network interruption resumption mechanism ensures data integrity, making it particularly suitable for wards or home care environments with unstable signals. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the structure of the real-time monitoring and nursing device for vital signs of bedridden patients based on multimodal sensing according to the present invention. Figure 2 This is a schematic diagram of the structure of the multimodal sensing pad of the real-time monitoring and nursing device for vital signs of bedridden patients based on multimodal sensing according to the present invention. In the diagram: 1-bed; 2-touchscreen display; 3-multimodal sensing pad. Detailed Implementation

[0009] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0010] like Figure 1-2The illustrated multimodal sensing-based real-time monitoring and nursing device for bedridden patients includes a bed 1 for supporting the patient, a touch screen 2 fixedly installed on the side edge of the bed 1, and a multimodal sensing pad 3 laid on the bed 1 and located under the patient's torso and limbs. In actual use, the patient can lie directly on the multimodal sensing pad 3 without attaching any electrodes or wearing any restraint devices. The multimodal sensing pad 3 adopts a multi-layer composite structure, consisting of, from bottom to top, an electromagnetic shielding layer, a flexible pressure sensing array layer, a dielectric elastomer sensing layer, a temperature sensing fiber layer, and a skin-friendly and breathable surface encapsulation layer.

[0011] The electromagnetic shielding layer is composed of conductive fabric or a metal-plated thin film, effectively isolating the bed's metal frame and external power frequency electromagnetic interference from affecting weak physiological signals. The flexible pressure sensing array layer uses matrix-type capacitive pressure sensing units. When different parts of the patient's body press on the corresponding units, the capacitance value changes, thus reconstructing a surface pressure distribution cloud map. The dielectric elastomer sensing layer uses a pre-stretched dielectric elastomer film, with flexible electrodes formed on its upper and lower surfaces through printing or vapor deposition. When the patient experiences periodic chest and abdominal movements due to breathing or weak surface impacts due to heartbeats, the film undergoes slight deformation, outputting a charge or capacitance change signal proportional to the deformation. The temperature sensing fiber layer is woven from thermosensitive fibers with a negative temperature coefficient. The fiber resistance decreases as temperature increases, and multi-point temperature data can be obtained by measuring the resistance between the interlacing nodes. The top surface encapsulation layer is made of skin-friendly and breathable textile material, ensuring patient comfort during long-term contact. The electromagnetic shielding layer also integrates an edge computing unit, a wireless communication module, and a local storage module.

[0012] The input terminals of the edge computing unit are connected to the signal output terminals of the flexible pressure sensing array layer, the dielectric elastomer sensing layer, and the temperature sensing fiber layer, respectively, and the output terminals are connected to the touch screen 2 and the wireless communication module.

[0013] The algorithm of this invention adapts to signals from each layer of the multimodal sensing pad and achieves high-precision monitoring by combining edge computing units. Its core is signal conditioning and fusion analysis, specifically as follows: the capacitance signal of the flexible pressure sensing array layer. RC low-pass filtering suppresses power frequency interference; the formula is as follows: ; in, Let be the original capacitance signal at position (x, y) at time t. The signal is the filtered signal, R is a resistor of 10kΩ to 100kΩ, C is a capacitor of 10nF to 100nF, and T is the filtering time window. The integral variable is the charge signal of the dielectric elastomer sensing layer. After conditioning with a charge amplifier, the formula is: ; in The original charge signal at time t, Conditioned voltage signal, For feedback capacitor, The feedback resistor is used. The two signals are sampled synchronously at ≥100Hz and timestamped. The fusion analysis submodule aligns the body movement and physiological signals, dynamically adjusts the signal-to-noise ratio threshold, matches the body position template, reduces false alarms of body movement, and achieves accurate monitoring of heart rate and respiratory rate.

[0014] A 65-year-old bedridden patient, weighing 75 kg and measuring 168 cm in height, with a baseline heart rate of 72 bpm and a respiratory rate of 18 breaths / min, was selected for 72-hour continuous monitoring. A multimodal sensing pad was placed in the center of the bed. The flexible pressure sensing array layer used a 32×32 matrix capacitive sensing unit. The RC low-pass filter parameters were set to R=50kΩ, C=50nF, and T=0.1s. The filtering formula was... ; The original capacitance signal at position (x, y) at time t is filtered. The fluctuation amplitude was reduced to ±2pF, and the power frequency interference amplitude was reduced from 0.8mV to below 0.05mV; the feedback parameters of the charge amplifier in the dielectric elastomer sensing layer were set to... =100PF, =1MΩ, conditioning formula ; After conditioning The output voltage ranges from 0.1V to 0.8V, the peak difference of the respiratory waveform is 0.3V to 0.5V, and the amplitude of the cardiac impact signal is stable at 0.02V to 0.05V. Both signals are sampled synchronously at 200Hz with timestamps added. The fusion analysis submodule aligns body movement and physiological signals. When the patient turns over, the signal-to-noise ratio threshold increases from 10dB to 18dB. At this time, the heart rate calculation temporarily uses the stable data from 1 second prior to the body movement, and no false alarms are generated. Under resting conditions, the heart rate monitoring error is ≤±1 beats / minute, the respiratory rate error is ≤±1 beats / minute, and the body surface temperature monitoring error is ≤±0.2℃. The amplitude attenuation of the cardiac impact signal is only 3.2% over 72 hours, and the data integrity reaches 99.8%, verifying the stability and accuracy of the algorithm and equipment.

[0015] During operation, the signal conditioning submodule inside the edge computing unit first preprocesses the three signals: for the flexible pressure sensing array layer, a capacitor-to-digital converter circuit is typically used to convert the capacitance value of each unit into voltage or direct digital quantity; for the dielectric elastomer sensing layer, a high input impedance charge amplifier is used to convert the weak charge changes it generates into voltage signals; for the temperature sensing fiber layer, a voltage divider resistor method is used to convert the fiber resistance changes into voltage changes, and low-pass filtering and differential amplification are applied to eliminate common-mode noise.

[0016] The multi-channel synchronous acquisition submodule performs synchronous analog-to-digital conversion on the three conditioned signals at a sampling rate of 100Hz to 500Hz, and adds a unified timestamp label to each frame of data to ensure the timing accuracy of subsequent fusion analysis. The feature extraction submodule obtains primary parameters in the following ways: from the pressure signal, it detects transient changes in the capacitance value of each unit; from the dielectric elastomer signal, since the respiratory component has a lower frequency and the cardiac impact component has a slightly higher frequency, an adaptive bandpass filter is used to separate the respiratory waveform and cardiac impact waveform, and then the respiratory rate and heart rate are obtained by peak detection or zero-crossing rate calculation respectively; from the temperature signal, according to the pre-calibrated temperature-resistance curve of the thermal fiber, the measured resistance value of each sensing node is converted into a body surface temperature value. The fusion analysis submodule is the core of the entire edge computing unit. It aligns the timestamps of body movement events with the respiratory and cardiac waveforms frame by frame. When a body movement event is detected within a waveform window, it automatically increases the signal-to-noise ratio threshold of the signals required for heart rate and respiratory rate calculations. Within this window, it prioritizes using the most recent stable waveform data before the body movement occurred for parameter calculation. If the signal quality is too low during the body movement, it is directly marked as "signal interference" and no physiological parameters are output, thus significantly reducing false heart rate / respiratory abnormality alarms caused by patient turning over, coughing, or limb twitching. Simultaneously, the fusion analysis submodule matches the extracted body pressure distribution feature vector with a pre-trained supine position classification template and outputs the current position classification result.

[0017] The fusion analysis submodule also counts the number of body movements and the average body surface temperature value of all temperature nodes within a preset time window. Finally, the edge computing unit sends the calculated heart rate, respiratory rate, body movement frequency, body position classification, and average body surface temperature to the touch screen 2 for visualization in real time, and issues an abnormal warning signal by flashing the screen and sounding when any parameter exceeds a preset threshold; at the same time, the wireless communication module packages and uploads the above vital signs parameters and the compressed raw sensor data to the remote nursing monitoring center for viewing by medical staff on the central workstation or mobile terminal.

[0018] To further improve the stability of the dielectric elastomer sensing layer in detecting cardiac impact signals, this invention sandwiches a silicone buffer layer with a thickness of 0.5mm-1.5mm between the flexible pressure sensing array layer and the dielectric elastomer sensing layer. This silicone buffer layer possesses flexible mechanical properties; when localized high pressure occurs at a point on the patient's body surface, the silicone buffer layer can diffuse this concentrated pressure into a more uniformly distributed pressure over a larger area before transmitting it to the underlying dielectric elastomer sensing layer. This effectively prevents the dielectric elastomer film from experiencing capacitive output saturation or mechanical damage due to localized overload. Simultaneously, the initial pre-stretch rate of the dielectric elastomer film is set to 10%-20%. This moderate pre-stretch ensures that it operates within the linear deformation region, maintaining high sensitivity to weak cardiac impacts without excessive relaxation due to the patient's resting body weight.

[0019] Experiments show that this structural design ensures that even under prolonged bed rest conditions for a 90kg patient, the amplitude attenuation of the cardiac impact signal in the dielectric elastomer sensing layer is less than 5% within 72 hours.

[0020] To address nursing scenarios with unstable network connections, this invention also incorporates a data protection mechanism. The local storage module built into the electromagnetic shielding layer employs a circular storage mechanism, retaining only the most recent 72 hours of multimodal raw sensor data and vital sign parameters output by the edge computing unit. Data older than 72 hours is automatically overwritten by new data to prevent storage space exhaustion. The edge computing unit continuously monitors the connection status between the wireless communication module and the remote nursing monitoring center. Upon detecting a connection interruption, it immediately writes all real-time generated raw sensor data and parameter results to the out-of-network cache in the local storage module. When the connection is restored, the edge computing unit automatically reads the unuploaded data from the cache and resumes uploading, releasing the cache space after the upload is complete. This mechanism ensures zero data loss in ward signal blind spots or during nighttime network maintenance, making it particularly suitable for home-based bedridden care or primary healthcare institutions.

[0021] The basic principles and main features of the present invention have been described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are only illustrative of the principles of the present invention. Various changes and modifications can be made to the present invention without departing from the spirit and scope of the present invention. All such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the invention is defined by the appended claims and their equivalents.

Claims

1. A real-time monitoring and nursing device for vital signs of bedridden patients based on multimodal sensing, characterized in that, It includes a bed for carrying patients; A touch screen display is fixedly installed on the side edge of the bed frame; And a multimodal sensing pad laid on the bed and located under the patient's torso and limbs; The multimodal sensing pad includes an electromagnetic shielding bottom layer, a flexible pressure sensing array layer, a dielectric elastomer sensing layer, a temperature sensing fiber layer, and a skin-friendly and breathable surface encapsulation layer, arranged in sequence from bottom to top. The flexible pressure sensing array layer consists of multiple capacitive pressure sensing units arranged in a matrix, used to detect the pressure distribution on the patient's body surface to identify body position and body movement signals. The dielectric elastomer sensing layer is used to collect periodic deformation signals of the chest and abdomen caused by the patient's breathing and weak impact vibration signals caused by the heartbeat. The temperature sensing fiber layer is composed of interwoven flexible thermosensitive fibers and is used to collect data on the temperature distribution of the patient's body surface. The electromagnetic shielding layer integrates an edge computing unit, a wireless communication module, and a local storage module. The edge computing unit is electrically connected to the flexible pressure sensing array layer, the dielectric elastomer sensing layer, and the temperature sensing fiber layer, respectively, and is used to perform time synchronization, feature extraction, and fusion analysis on the acquired multimodal sensing signals, and calculate the patient's real-time heart rate, respiratory rate, body movement rate, body position classification, and average body surface temperature based on the fused signals. The edge computing unit is also electrically connected to the touch screen, for displaying the calculated vital signs parameters in real time and issuing abnormal warning signals. The wireless communication module is electrically connected to the edge computing unit and is used to upload real-time vital signs parameters and raw sensor data to the remote nursing monitoring center. The edge computing unit includes a signal conditioning submodule, a multi-channel synchronous acquisition submodule, a feature extraction submodule, and a fusion analysis submodule. The signal conditioning submodule receives the capacitance change signal output by the flexible pressure sensing array layer, the charge change signal output by the dielectric elastomer sensing layer, and the resistance change signal output by the temperature sensing fiber layer, and performs filtering and amplification processing on them.

2. The real-time monitoring and nursing device for vital signs of bedridden patients based on multimodal sensing according to claim 1, characterized in that, The multi-channel synchronous acquisition submodule performs synchronous analog-to-digital conversion on the three conditioned signals at a sampling rate of not less than 100Hz and adds timestamp tags; the feature extraction submodule extracts the start and end times of body motion events and body pressure distribution feature vectors from the synchronized pressure signal, separates the respiratory waveform and cardiac impact waveform from the dielectric elastomer signal through adaptive bandpass filtering and calculates the respiratory rate and heart rate respectively, and extracts the resistance values ​​of each sensing node from the temperature signal and converts them into body surface temperature values; The fusion analysis submodule aligns the timestamps of body movement events with the respiratory waveform and cardiac impact waveform in time sequence. When body movement is detected, it dynamically adjusts the signal-to-noise ratio thresholds for respiratory and heart rate calculations. Based on the body pressure distribution feature vector, it matches a preset supine classification template and outputs the body position classification results. At the same time, it counts the number of body movements and the average body surface temperature within a preset time window.

3. The real-time monitoring and nursing device for vital signs of bedridden patients based on multimodal sensing according to claim 1, characterized in that, The dielectric elastomer sensing layer includes a dielectric elastomer film with flexible electrodes coated on its upper and lower surfaces. The initial pre-stretch rate of the dielectric elastomer film is set to 10%-20%. A silicone buffer layer with a thickness of 0.5mm-1.5mm is sandwiched between the flexible pressure sensing array layer and the dielectric elastomer sensing layer. The silicone buffer layer is used to diffuse the concentrated pressure on the patient's body surface into a uniformly distributed pressure and transmit it to the dielectric elastomer sensing layer to avoid local pressure overload causing the capacitance output of the dielectric elastomer film to saturate.

4. The real-time monitoring and nursing device for vital signs of bedridden patients based on multimodal sensing according to claim 1, characterized in that, The local storage module built into the electromagnetic shielding layer adopts a circular storage mechanism, retaining only the most recent 72 hours of multimodal raw sensor data and vital sign parameters output by the edge computing unit. When the edge computing unit detects that the connection between the wireless communication module and the remote nursing monitoring center is interrupted, it continuously writes real-time data into the local storage module and automatically uploads all stored data after the connection is restored. The input terminals of the edge computing unit are connected to the signal output terminals of the flexible pressure sensing array layer, the dielectric elastomer sensing layer, and the temperature sensing fiber layer, respectively, and the output terminals are connected to the touch screen 2 and the wireless communication module. High-precision monitoring is achieved by combining edge computing units, with the core being signal conditioning and fusion analysis, specifically as follows: capacitive signal of the flexible pressure sensing array layer. RC low-pass filtering suppresses power frequency interference; the formula is as follows: ; in, Let be the original capacitance signal at position (x, y) at time t. The signal is the filtered signal, R is a resistor of 10kΩ to 100kΩ, C is a capacitor of 10nF to 100nF, and T is the filtering time window. For integration variables; Charge signal of dielectric elastomer sensing layer After conditioning with a charge amplifier, the formula is: ; in The original charge signal at time t, Conditioned voltage signal, For feedback capacitor, For feedback resistor; The two signals are sampled synchronously at ≥100Hz and timestamped. The fusion analysis submodule aligns the body movement and physiological signals, dynamically adjusts the signal-to-noise ratio threshold, matches the body position template, reduces false alarms of body movement, and achieves accurate monitoring of heart rate and respiratory rate.

Images