Active nursing device with early warning function and nursing method

Abstract

The application discloses a kind of active nursing equipment with early warning function and nursing method, the equipment includes nursing pad main body and control unit, nursing pad main body is equipped with the pressure, temperature and humidity sensor and corresponding position's micro air bag arranged in grid;Control unit is based on each sensor data real-time calculation risk index R including pressure, temperature, humidity and duration, when R value exceeds first threshold value, corresponding air bag is deflated and adjacent air bag is inflated to realize local decompression, when R value exceeds second threshold value, trigger first early warning signal, while using historical R value sequence to predict future risk and trigger second early warning signal when predicted value exceeds third threshold value;The application is through the real-time calculation and hierarchical early warning of multi-factor risk index, combined with targeted local active decompression operation and early warning based on time series prediction, active intervention and prospective nursing to pressure sore risk are realized, and the incidence of pressure sore is greatly reduced, and nursing efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of medical and nursing equipment technology, and in particular to an active nursing device and nursing method with an early warning function. Background Technology

[0002] For the prevention and care of pressure ulcers in long-term bedridden patients, clinical practice mostly involves regular manual turning, the use of static pressure ulcer mattresses, or alternating air mattresses. Static pressure ulcer mattresses distribute pressure on the body surface through the elasticity of the material, but they cannot dynamically adjust for different pressure points. Alternating air mattresses inflate and deflate air chambers at fixed intervals, which can alleviate continuous pressure to some extent, but their operation is an open-loop timed control, not dependent on individualized parameters such as the patient's actual pressure, temperature, and humidity, and cannot identify specific high-risk locations for precise intervention. In addition, most existing nursing equipment lacks risk warning functions, and caregivers usually only discover problems after visible skin redness or breakage, by which time pressure ulcers have often already formed.

[0003] The aforementioned devices, unable to perceive real-time changes in the microenvironment at multiple locations on the patient's body surface (such as the duration of local pressure, the cumulative effects of temperature and humidity), struggle to predict risk trends before pressure ulcers form, and are unable to issue early warnings to nursing staff. Even with alternating air mattresses, their fixed inflation / deflation cycles cannot adapt to different patients' body shapes, weights, activity levels, and changes in their conditions, potentially leading to frequent ineffective or untimely adjustments, resulting in prolonged ischemia and hypoxia in local tissues. Furthermore, existing devices lack active monitoring of patient activity, making it impossible to determine whether a patient remains motionless in the same position for an extended period, or to respond quickly when a patient slides towards the edge of the bed or is at risk of falling.

[0004] Therefore, in response to the problems mentioned above, this invention proposes an active nursing device and nursing method with early warning function. Summary of the Invention

[0005] To overcome the problems of existing nursing equipment lacking real-time risk assessment, unable to provide proactive early warning, and having intervention modes that do not match the actual condition of the patient, this invention proposes a proactive nursing device and method with early warning function, which can realize real-time perception, early warning, and automatic and precise pressure relief care for bedridden patients with pressure ulcer risk.

[0006] The technical solution of the present invention is: an active nursing device with an early warning function, comprising: The nursing pad body has a multi-layered structure, consisting of a breathable layer, a sensor layer, an actuator layer, and a base layer, arranged from the patient contact surface downwards. The sensor layer includes pressure sensor units, temperature sensor units, humidity sensor units, and tilt sensors arranged in an M×N grid. The tilt sensor is set along the edge to detect the tilt angle of the nursing pad body. The control unit determines whether the patient is in a position at risk of falling out of bed based on the tilt angle. When the tilt angle exceeds the angle threshold and the pressure distribution map shows that the patient's center of gravity is biased towards the edge, the fifth warning signal is triggered and the micro airbags at the edge are quickly inflated to form a barrier. The actuator layer includes multiple independently controlled micro-airbags that correspond one-to-one with the positions of the sensor units, and each micro-airbag is connected to an independent inflation / deflation valve. The control unit, electrically connected to all sensor units and all charge / discharge valves, includes: The data acquisition unit is used to acquire pressure, temperature and humidity values ​​from each sensor unit at a predetermined frequency. The risk assessment unit is used to calculate a real-time risk index for the location of each sensor unit, using the following formula: ; Where P is the pressure value, T is the temperature value, H is the humidity value, t is the time during which the pressure exceeds the preset pressure threshold, and w1, w2, w3, and w4 are weighting coefficients. max T max H max The maximum reference value is t0, which is the reference time. Among them, w1=0.5, w2=0.2, w3=0.2, w4=0.1, and the first risk threshold R1=0.4, the second risk threshold R2=0.7, the third risk threshold R3=0.6, and the preset time Δt is 30 minutes. An active intervention unit is used to control the micro airbag corresponding to any sensor unit to perform a decompression operation when the R value of any sensor unit is greater than the first risk threshold R1. The decompression operation includes deflating the micro airbag to a first predetermined air pressure and simultaneously controlling multiple micro airbags adjacent to the sensor unit to inflate to a second predetermined air pressure. The first early warning unit is used to trigger a first early warning signal when the R value of any sensor unit is greater than the second risk threshold R2, where R2>R1. The first early warning signal includes the location information of the high-risk sensor unit and alerts the nursing staff through an audible and visual alarm. The prediction and early warning unit is used to input the historical R value sequence of each sensor unit into a pre-trained time series prediction model (the model is a neural network model based on Long Short-Term Memory Network LSTM or Gated Recurrent Unit GRU, which is trained by collecting the patient's historical sensor data for the previous 72 hours) to predict the R value after a future time Δt. When the R predicted value is greater than the third risk threshold R3, a second early warning signal is triggered, which indicates the location and expected time of the pressure ulcer risk. The body movement monitoring module is connected to the control unit. The body movement monitoring module calculates the body movement frequency by the number of times the pressure change amplitude of the pressure sensor unit exceeds the change threshold per unit time. When the body movement frequency is lower than the first frequency threshold and the duration exceeds the time threshold, a third warning signal is triggered to indicate that the patient is not active enough. The edge detection module identifies the patient's body edge contour based on the gradient changes in the pressure distribution map. When the edge contour moves less than a distance threshold within a predetermined time, it determines that the patient is in a dangerous resting state and triggers the fourth warning signal. The wireless communication module sends the first warning signal, the second warning signal, and the risk index R value distribution map to the remote monitoring terminal or the nurse station. The patient identification module, which is an RFID reader or a QR code scanner, is used to read patient information and associate and store the patient's sensor data and early warning history with the patient's file.

[0007] Preferably, in the decompression operation, the micro-airbag is deflated to a first predetermined pressure of 0.1-0.3 atmospheres, and the adjacent micro-airbag is inflated to a second predetermined pressure of 0.6-0.8 atmospheres. The decompression operation lasts for 2-5 minutes, after which the pressure is automatically restored to the initial pressure.

[0008] A nursing method using an active nursing device with early warning function includes the following steps: S1, initialize the device and collect pressure, temperature and humidity data of the contact surface between the patient and the nursing pad; S2, calculates the risk index R in real time for each sensor unit location; S3, when any R value is greater than the first risk threshold R1, perform active decompression operation: control the corresponding micro airbag to deflate while controlling the adjacent micro airbag to inflate. S4, when any R value is greater than the second risk threshold R2, the first warning signal is triggered; S5. Input the historical R value sequence into the prediction model to predict the future R prediction value. When the R prediction value is greater than the third risk threshold R3, trigger the second early warning signal. S6. Repeat steps S1-S5 to achieve proactive care and early warning of pressure ulcer risk.

[0009] The beneficial effects of this invention are: 1. This invention solves the problem that existing technologies cannot perform real-time risk quantification assessment for individualized patient microenvironments by constructing a multi-factor risk index R that includes pressure, temperature, humidity and duration, and by calculating the R value of each location in real time based on a gridded sensor, thus realizing dynamic perception of pressure ulcer risk.

[0010] 2. This invention achieves local pressure redistribution rather than overall alternating inflation by actively intervening subunits to precisely deflate micro-airbags at high-risk locations and simultaneously inflate eight adjacent airbags when the R value exceeds a first threshold. This solves the problem of mismatch between intervention and the actual condition of the patient caused by the fixed cycle and open-loop control of existing alternating air mattresses, thus avoiding ineffective adjustment and significantly reducing patient discomfort.

[0011] 3. This invention transforms passive detection into active prediction by using a cascaded design of the first and second early warning subunits, and especially by using a time series prediction model to predict the risk in the next 30 minutes based on historical R values. This solves the problem that existing equipment cannot provide early warning of pressure ulcer formation trends, and allows nursing staff sufficient time for preventive intervention.

[0012] 4. This invention integrates a body movement monitoring module, an edge detection module, and a tilt sensor to actively monitor the patient's status from three dimensions: activity frequency, body posture maintenance, and risk of falling from bed. This solves the problem that traditional nursing equipment only focuses on pressure while ignoring the patient's overall mobility and safety, forming a multi-dimensional and proactive closed-loop nursing system. Attached Figure Description

[0013] Figure 1 The diagram shown is a schematic representation of the multi-layer structure of the nursing pad body of the present invention. Figure 2 The diagram shown is a top view of the grid layout of the sensor unit and the micro airbag of the present invention. Figure 3 The diagram illustrates the calculation and threshold determination process of the risk index R of this invention. Figure 4 The diagram shown illustrates the active decompression operation of the present invention. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 some embodiments of the present invention, but 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.

[0015] This invention provides an embodiment: Please see Figure 1 This nursing pad is laid on the hospital bed, and its size can be designed as a rectangular pad with a length of 180-200cm and a width of 80-100cm, depending on the bed specifications. The main body of the nursing pad adopts a multi-layer composite structure, consisting of a breathable layer, a sensor layer, an actuator layer, and a base layer from the patient's body contact surface downwards. The breathable layer is made of highly breathable medical-grade polyurethane foam material with a thickness of 5-10mm and a wavy texture on the surface to reduce shear force and maintain adequate airflow between the skin and the sensor layer. The sensor layer has a flexible circuit board structure, embedding pressure sensor units, temperature sensor units, and humidity sensor units arranged in an M×N grid. In this example, M is 16 and N is 12, for a total of 192 sensor units. The pressure sensor unit employs a flexible thin-film piezoresistive sensor with a measurement range of 0-200 mmHg and an accuracy of ±2 mmHg; the temperature sensor unit uses a surface-mount thermistor with a measurement range of 25-45℃ and an accuracy of ±0.1℃; and the humidity sensor unit uses a capacitive humidity sensor with a measurement range of 20-95% RH and an accuracy of ±3% RH. These three sensors are spatially integrated at the same grid node, enabling each grid node to simultaneously acquire pressure, temperature, and humidity data for that point.

[0016] The actuator layer, located below the sensor layer, comprises multiple independently controlled micro-inflatable bladders, each corresponding to a sensor unit. The planar dimensions of each bladder are identical to the sensor unit area, and its inflated thickness is 2-4 cm. Each micro-inflatable bladder has an independent inflation / deflation valve connected to its bottom, which is linked to the air supply system via a micro-pump and an air tank. The control unit is mounted on the side of the nursing pad body and electrically connected to the sensor and actuator layers via a flexible cable. The control unit integrates an embedded microprocessor, memory, and power management module, and provides external interfaces for audible and visual alarms, wireless communication, and patient identification. The control unit operates on 5V DC, with a total power consumption not exceeding 15W, and can operate continuously for over 72 hours.

[0017] During installation, first, lay the main body of the nursing pad flat on the hospital bed, ensuring the base layer is completely in contact with the bed surface. Then, place the patient on the nursing pad, with the patient's body roughly in the center of the pad. Scan the patient's wristband using an RFID reader to automatically record the patient's name, age, weight, and pressure ulcer risk assessment level, and bind this information to the nursing pad. The control unit initiates a self-test program, sequentially checking whether the output of all sensor units is normal, whether the inflation and deflation of all micro-inflators is flexible, and whether the air source pressure is stable within the range of 0.8-1.2 atmospheres. After passing the self-test, the system enters standby mode.

[0018] Please see Figure 2 Furthermore, during data acquisition, the data acquisition unit continuously collects pressure, temperature, and humidity values ​​from each sensor unit at a predetermined frequency of 10 times per second. Because changes in patient position, respiration, and heart rate can cause slight fluctuations in sensor output, a digital low-pass filter is embedded within the data acquisition unit, with a cutoff frequency set to 0.5Hz. This filter removes noise components higher than 0.5Hz, retaining low-frequency signals related to changes in the patient's posture and microenvironment. For each sensor unit, the filtered pressure value is denoted as P(i, j), the temperature value as T(i, j), and the humidity value as H(i, j), where i = 1 - M and j = 1 - N.

[0019] The data acquisition unit is also responsible for recording the time *t* for each sensor unit's pressure to exceed a preset pressure threshold. The preset pressure threshold is set according to clinical pressure ulcer guidelines, typically 32 mmHg, because when local pressure remains above 32 mmHg, capillary blood flow begins to be obstructed. The specific timing method is as follows: for each sensor unit, the control unit monitors its pressure value in real time. If the current pressure value is greater than or equal to 32 mmHg, a timer is started and incremented; once the pressure value drops below 32 mmHg, the timer is reset to zero. *t* represents the duration of this current round of continuous high pressure, in minutes. This timing method effectively distinguishes between intermittent and continuous pressure, because only continuous high pressure truly poses a risk of tissue ischemia.

[0020] At the end of each acquisition cycle, the data acquisition subunit packages each set of (P, T, H, t) data and stores it along with a timestamp into the control unit's circular buffer. This buffer is designed to store 72 hours of all raw data, facilitating subsequent analysis and model training.

[0021] Please see Figure 3 Furthermore, when conducting risk index assessment, the risk assessment unit calculates the real-time risk index R for the location of each sensor unit according to the following formula: ; In this invention, w1, w2, w3, and w4 are weighting coefficients, with fixed values: w1 = 0.5, w2 = 0.2, w3 = 0.2, and w4 = 0.1. These weighting coefficients were determined based on multicenter clinical data using principal component analysis. Pressure contributes the most to pressure ulcer formation and is therefore assigned the highest weight. Temperature and humidity mainly affect the softening of the stratum corneum and bacterial proliferation, contributing moderately. The time factor reflects the cumulative effect and has a slightly lower weight but is still not negligible. max The maximum reference pressure is set at 120 mmHg, which corresponds to the critical pressure above complete vascular occlusion; T max The maximum reference temperature is 42℃; above this temperature, the tissue metabolic rate increases sharply. max The maximum humidity reference value is 95% RH, which is close to full saturation; t0 is the reference time, which is 60 minutes.

[0022] The following is a specific calculation example to illustrate this. Assume a sensor unit measures a current pressure P = 80 mmHg, temperature T = 36.5℃, humidity H = 80% RH, and the duration of high pressure is t = 25 minutes. The calculations for each item are as follows: ; ; ; ; Substituting the weights, R = 0.5×0.6667 + 0.2×0.8690 + 0.2×0.8421 + 0.1×0.4167 = 0.3333 + 0.1738 + 0.1684 + 0.0417 = 0.7172. This R value is greater than the second risk threshold R2 = 0.7, therefore the device should trigger the first warning signal.

[0023] If the pressure of another sensor unit is P=40mmHg, but the location has been under high pressure for 90 minutes, and all other conditions are the same, then calculate: ; ; Then R = 0.5 × 0.3333 + 0.2 × 0.8690 + 0.2 × 0.8421 + 0.1 × 1.5 = 0.1667 + 0.1738 + 0.1684 + 0.15 = 0.6589, which is less than 0.7 but greater than the first risk threshold R1 = 0.4. At this time, an active decompression operation should be triggered, but the first warning signal will not be issued for the time being.

[0024] The risk assessment unit recalculates the R-values ​​of all sensor units every 10 seconds, generating an M×N R-value distribution map. This map is displayed in pseudo-color on the local display screen of the control unit or on the remote monitoring terminal. Red areas represent high-risk areas with R ≥ 0.7, yellow areas represent medium-risk areas with R ≤ 0.4 < 0.7, and green areas represent low-risk areas with R < 0.4. This visualization allows caregivers to quickly identify which areas require close monitoring.

[0025] Furthermore, during active intervention, the active intervention unit continuously monitors the R value of each sensor unit. When the R value of any sensor unit exceeds the first risk threshold R1 = 0.4, a decompression operation is immediately initiated for that location.

[0026] It should be noted that the setting of R1=0.4 is based on a large number of in vitro tissue perfusion pressure experiments. When the R value exceeds 0.4, the local tissue has already shown a measurable decrease in oxygen partial pressure, but has not yet reached the level of irreversible damage. At this time, the intervention effect is the best.

[0027] Please see Figure 4 The decompression operation is explained in detail below: First, the control unit sends a command to the inflation / deflation valve of the micro-inflator corresponding to the high-risk sensor unit, rapidly deflating the inflator to a first predetermined pressure of 0.2 atmospheres (absolute pressure). At this point, the inflator is almost completely flattened, significantly reducing the support force in that area and partially unloading the surface pressure that was previously concentrated at that point. Simultaneously, the control unit identifies all micro-inflators adjacent to the high-risk unit. Within a rectangular grid, adjacent inflators include eight in total, located in the top, bottom, left, right, and four diagonal directions. The control unit then sends a command to the inflation / deflation valves of these eight adjacent inflators, inflating them to a second predetermined pressure of 0.7 atmospheres. This redistributes the patient's weight, originally borne by the central inflator, to the surrounding eight inflators, creating an artificial "pressure depression" in the high-risk area, providing temporary and complete decompression of the skin.

[0028] It should be noted that this decompression operation is not a simple deflation or inflation of all airbags, but a targeted, localized, and directional pressure redistribution. Compared with traditional alternating air mattresses (where the entire air mattress is completely re-inflated every few minutes), the decompression operation of this invention only changes the state of the high-risk area and its surrounding eight airbags, while the airbags in other areas maintain their original pressure. Therefore, it does not interfere with the support stability of the patient's non-risk areas and avoids the risk of discomfort or even falls caused by frequent and large-scale pressure adjustments.

[0029] The decompression operation is set to last for 3 minutes. After 3 minutes, the control unit automatically restores the central airbag to the pre-decompression pressure level, and simultaneously restores the surrounding eight airbags to the pre-decompression pressure level.

[0030] It should be noted that under complete decompression conditions, capillary blood flow in local tissues can be largely restored within 2-3 minutes. However, if the decompression time is too long, the pressure distribution on the patient's body surface will change significantly, potentially leading to new risk points. If the R-value of the sensor unit remains above 0.4 during the decompression procedure (e.g., the patient is completely unable to move independently and their weight is constantly pressing on the area), the control unit will continuously perform the decompression procedure, repeating it every 3 minutes until the R-value drops below 0.4. A 30-second recovery period is allowed between each repetition to allow the airbag to return to its initial state and the actual pressure to be re-measured.

[0031] Furthermore, the various warning signals are explained in detail below: The first early warning subunit is responsible for handling situations where real-time risks reach a severe level. When the R value of any sensor unit exceeds the second risk threshold R² = 0.7, it indicates that the tissue at that location is in a highly dangerous state. Active decompression alone may not be sufficient to quickly reverse the risk, and nursing staff must be immediately notified to perform manual intervention at the bedside, such as helping the patient change position or examining the skin. The first early warning signal consists of two parts: first, a 90-decibel intermittent alarm sound is emitted by the buzzer on the control unit; second, the control unit sends an early warning message to the nurse station via the wireless communication module. The message includes the patient ID, high-risk location coordinates, current R value, pressure value, temperature value, and humidity value. A pressure distribution map of the patient's nursing pad will pop up on the central monitoring screen at the nurse station, with high-risk areas marked with a prominent border.

[0032] Once triggered, the first warning signal will persist until one of the following conditions is met: 1. The patient's R value drops below 0.6 and remains below 0.6 for more than 1 minute; 2. The caregiver presses the "Confirm" button on the control unit; 3. The caregiver manually cancels the alarm after completing the manual repositioning operation. During the duration of the alarm, the active intervention unit continues to perform decompression operations but will not trigger a new first warning signal to avoid excessive alarm repetition.

[0033] The second early warning signal is generated by the prediction and early warning subunit, which operates based on a time series prediction model. First, for each sensor unit, the risk assessment unit stores the historical R-value sequence over a past period in a circular buffer. Every 5 minutes, the prediction and early warning subunit extracts the R-value sequence for each sensor unit over the most recent 4 hours (i.e., 240 data points, each representing a 10-second average R-value) and inputs it into a pre-trained time series prediction model. This model is a neural network based on Long Short-Term Memory (LSTM).

[0034] Specifically, the network structure is as follows: the input layer receives a time series of length 240, with each time step featuring a 1-dimensional feature (i.e., the R-value). This is followed by two LSTM layers: the first layer has 64 neurons, and the second layer has 32 neurons, both using the tanh activation function. Then, a fully connected layer outputs a scalar (i.e., the predicted R-value Δt = 30 minutes later). The model was trained using historical sensor data from the patient's first 72 hours of admission before deployment. The training process employed the Adam optimizer with mean squared error as the loss function, 100 training epochs, and early stopping to prevent overfitting. Because each patient has different body types and mobility levels, the model is personalized. Specifically, each nursing pad device collects R-value data from the first 72 hours of use as a training set to train a dedicated LSTM model. For emergency situations or patients who cannot wait 72 hours, a pre-trained general model on a public dataset can be used as the initial model, which is then fine-tuned through online learning.

[0035] When the predicted R value output by the model exceeds the third risk threshold R3 = 0.6, the prediction and warning unit triggers a second warning signal. R3 is set to 0.6 because if the predicted risk index reaches 0.6 or higher after 30 minutes, it indicates a rapidly deteriorating risk trend. Even if the current R value is less than 0.4, without early intervention, it will enter a medium-to-high risk zone after 30 minutes. The second warning signal differs from the first warning signal in that it uses a lower frequency of audible and visual alarms, prompting caregivers to allocate time for preventative turning within the next 30 minutes, rather than requiring the same emergency response as the first warning. The second warning signal also includes location information and the estimated time of occurrence. This warning information can also help caregivers develop individualized turning plans, focusing on areas predicted to be high-risk but not yet actually at risk. This second warning signal allows caregivers to detect risk trends 20-30 minutes in advance.

[0036] Furthermore, the motion monitoring module will be explained in detail below: The motion monitoring module runs in the background of the control unit, using signal changes from existing pressure sensor units to infer the patient's spontaneous activity. For each pressure sensor unit, the module calculates the magnitude of pressure change within a unit of time (1 minute). The magnitude of change is defined as the difference between the maximum and minimum pressure values ​​within that minute. When the magnitude of change exceeds a threshold (set at 10 mmHg), a significant motion event is considered to have occurred at that location. Since a patient turning over, moving their limbs, or even just taking a deep breath can cause pressure changes in multiple sensor units simultaneously, the motion monitoring module counts the total number of motion events across all sensor units on the entire nursing pad, obtaining the motion frequency per unit of time, expressed in "times / minute".

[0037] The body movement monitoring module has a first frequency threshold of 0.2 times / minute and a time threshold of 2 hours. When the body movement frequency is below 0.2 times / minute and this state lasts for more than 2 hours, a third warning signal is triggered. The third warning signal indicates "Insufficient patient activity; passive movement assistance recommended." This warning is sent to the nurses' station as a text message, and the control unit simultaneously broadcasts it in a low voice (50 decibels to avoid startling the patient). This is because if the patient's voluntary movement frequency is too low, even if the R value of each sensor unit is not exceeded, the skin tissue will suffer from lymphatic drainage obstruction and metabolic waste accumulation due to prolonged lack of micro-deformation, increasing susceptibility to pressure ulcers. After receiving the third warning, nursing staff can perform passive joint movements or adjust the patient's position to promote blood circulation.

[0038] It should be noted that the body movement monitoring module does not interfere with normal sleep. If the patient is in deep sleep, the frequency of body movement will naturally decrease, but the 2-hour threshold can avoid false alarms, because it is unlikely that a normal person will remain completely still for two hours during sleep, and the body will unconsciously turn over or make minor adjustments to its limbs.

[0039] Table 1 provides the activity status evaluation criteria corresponding to different body movement frequencies: Table 1 Evaluation Criteria for Activity Status Corresponding to Different Body Movement Frequencies

[0040] Furthermore, the edge detection module identifies the patient's body edge contours and determines whether the patient has maintained the same posture for an extended period based on edge movement. This module first extracts the effective pressure area from the pressure distribution map. Due to the incomplete fit between the air mattress or the patient's body surface and the nursing pad, the pressure of some sensor units may be close to zero. The edge detection module uses a gradient threshold method to calculate the absolute value of the pressure difference between each sensor unit and its surrounding units. When the gradient value is greater than 5 mmHg / cm, the location is considered to be at the body edge. Connecting all edge points yields a closed contour line, within which lies the patient's projected body area.

[0041] The edge detection module tracks the centroid coordinates and shape parameters of the contour line. Within a continuous 30-minute time window, if the centroid moves less than a distance threshold (set to 2cm) and the shape parameters change by less than 5%, the patient is determined to be in a "dangerous resting state." This dangerous resting state means the patient has not changed their basic posture, and the area of ​​pressure on their body remains almost unchanged, which greatly increases the risk of pressure ulcers. At this point, a fourth warning signal is triggered. The fourth warning signal includes the message "The patient has not changed position for an extended period; the current pressure area is [mapped to its anatomical location via coordinates]," and recommends that caregivers turn the patient over.

[0042] Unlike the motion monitoring module, the edge detection module focuses on maintaining overall posture rigidity, while motion monitoring focuses on the frequency of small, localized movements. They complement each other. A patient might frequently move their arms (high frequency of movement) but keep their trunk stationary (small edge movement distance). In this case, the fourth alert can still remind caregivers to pay attention to areas of pressure on the trunk. Conversely, a patient might turn over completely (large edge movement distance), but if a certain localized area is continuously compressed during the turn, motion monitoring can also detect this.

[0043] Furthermore, falls from bed are another major safety incident for bedridden patients, especially elderly patients, those with impaired consciousness, or those still recovering from anesthesia after surgery. These patients may slip towards the edge of the bed or even fall out while restless or turning over. This invention incorporates six tilt sensors—MEMS triaxial accelerometers—at the four corners and the middle of the long side of the nursing pad. These sensors measure the tilt angle of the nursing pad relative to the direction of gravity. The control unit reads the output of each tilt sensor 10 times per second and calculates the overall tilt angle of the nursing pad in the roll and pitch directions through fusion calculations.

[0044] In addition, pressure distribution maps can provide information about the patient's center of gravity location. Center of gravity coordinates (X...) cg Y cg The pressure values ​​from all sensor units are weighted and averaged to obtain: ; ; Where x j and y j These represent the horizontal and vertical coordinates of the sensor unit, respectively.

[0045] A risk assessment for falling from a bed is considered high when both of the following conditions are met: 1. If the lateral tilt angle of the nursing pad (towards the patient) exceeds the angle threshold (set to 15°), it indicates that the bed or the nursing pad itself has tilted, or that the patient's center of gravity has shifted, causing the pad to be pressed to one side. Second, the horizontal coordinate X of the patient's center of gravity. cg If the distance from the edge of the nursing pad is less than the safe distance (set to 8 cm), it means that the patient is very close to the edge of the bed.

[0046] When the aforementioned high-risk conditions are met, the control unit triggers the fifth warning signal. Upon triggering the fifth warning signal, the audible and visual alarm is activated with the highest priority (a 100-decibel high-frequency buzzer), and simultaneously, the control unit immediately sends a rapid inflation command to the micro-inflators closest to the patient's body edge. Specifically, if the center of gravity is closer to the right edge, all micro-inflators in both rows on the right edge are inflated to 0.8 atmospheres within 0.5 seconds, forming an airbag barrier. This barrier effectively prevents the patient from sliding further towards the edge; even if the patient is agitated, the sudden inflating barrier will change their sliding direction or reduce their speed, buying time for nursing staff to arrive.

[0047] It should be noted that the rapid inflation action of the fifth warning signal is physically identical to the inflation of adjacent airbags in the active decompression operation; only the triggering conditions and inflation range differ. The active decompression operation inflates only eight airbags, while the fall protection operation inflates an entire row. The two do not conflict; the control unit uses priority arbitration, with the fall protection command having the highest priority. Once triggered, it temporarily suspends other ongoing decompression operations until the fall risk is eliminated, at which point it resumes.

[0048] Furthermore, the wireless communication module uses low-power Wi-Fi or Bluetooth protocols in the 2.4GHz band to connect to the hospital's internal local area network or the central server at the nurses' station. This module continuously uploads the following data: First, the raw data of pressure, temperature, and humidity for each sensor unit, along with the corresponding R value, are uploaded once per minute, compressed, and packaged before being sent. Second, detailed information on all warning events, including warning type (first to fifth), trigger location, trigger time, R value at the time of triggering, and other environmental parameters, is sent in real time; Third, the execution record of proactive intervention operations, including the time, location, duration and effect of each decompression operation (change in R value before and after decompression). Fourth, summaries of auxiliary monitoring information such as body movement frequency, edge movement distance, and tilt sensor data are sent every 5 minutes.

[0049] The nurse station monitoring terminal software can simultaneously display the status of nursing pads for multiple patients. The interface is designed as a ward map, with each bed corresponding to a color block. The color of the block dynamically changes according to the patient's current maximum R value: green (maximum R < 0.4), yellow (0.4 ≤ maximum R < 0.7), and red (maximum R ≥ 0.7). Clicking on a color block expands to display a detailed pressure distribution cloud map, historical trend curves, and warning records. When a warning occurs, the corresponding bed's color block will flash.

[0050] The wireless communication module also supports two-way communication. Nursing staff can send commands to the nursing pad through the nurse station software, such as remotely deactivating alarms, manually inflating or deflating airbags in a specific area, modifying risk thresholds, or exporting historical data for research analysis.

[0051] Furthermore, the coordinated operation of the above modules is illustrated in this example to demonstrate the active care method of the present invention. Specifically: S1. After the device is powered on, the control unit performs a self-test, including air source pressure detection, zero-point calibration of all sensors, and movement test of all micro-inflators. After the self-test passes, the nursing staff identifies the patient via RFID, and the device begins to collect pressure, temperature, and humidity data from all sensor units at a frequency of 10 Hz, while simultaneously updating the pressure duration exceeding the threshold t at each location.

[0052] S2, every 10 data points collected (i.e., data for 1 second), the risk assessment unit calculates the average pressure, average temperature, and average humidity for each sensor unit, and then calculates the R value. Simultaneously, the R value distribution map is updated and saved to the circular buffer.

[0053] S3, determine if any sensor unit has an R value > 0.4. If so, the active intervention subunit immediately deflates the corresponding micro-inflator to 0.2 atm, while simultaneously inflating its eight adjacent inflators to 0.7 atm, maintaining this state for 3 minutes before returning to normal. If the R value of this unit drops below 0.4 within 3 minutes, the decompression operation is terminated prematurely. During the decompression operation, the risk assessment and early warning functions for other locations on the same patient remain unaffected.

[0054] S4, determine if any sensor unit has an R value > 0.7. If so, trigger the first warning signal, including audible and visual alarms, location indication, and remote message push. This warning continues until confirmed by caregivers or the R value drops below 0.6.

[0055] S5, every 5 minutes, the prediction and early warning subunit extracts the R value sequence of each sensor unit for the most recent 4 hours and inputs it into the LSTM model to predict the R value 30 minutes later. If the predicted R value > 0.6, a second early warning signal is triggered to prompt nursing staff to intervene in advance.

[0056] S6, the body movement monitoring module calculates the body movement frequency in 1-minute windows and accumulates the duration of movements below 0.2 movements / minute. If the cumulative duration exceeds 2 hours, a third warning signal is triggered.

[0057] S7, the edge detection module calculates the body edge contour and center of mass position every 30 seconds. If the center of mass moves less than 2cm and the contour shape changes less than 5% within 30 consecutive minutes, the fourth warning signal is triggered.

[0058] S8, the tilt sensor reads the tilt angle at 10Hz and calculates the center of gravity coordinates simultaneously. If the lateral tilt angle is > 15° and the center of gravity is < 8cm from the edge, the fifth warning signal is immediately triggered and the edge airbag is rapidly inflated.

[0059] In step S9, all data generated in each step is recorded in real time and packaged and sent to the nurses' station every minute, while also being saved locally. The process then loops back to step S1 to continue the data collection and processing for the next cycle.

[0060] The above steps do not execute all judgments in every loop. For example, the bed fall risk detection can be performed every 100ms at a high frequency, while the prediction and early warning can be performed every 5 minutes. The two run in parallel on different time scales.

[0061] The effects of the present invention are illustrated in Tables 1 and 2 below. These data are derived from a 6-month clinical trial conducted on the prototype in the rehabilitation department of a tertiary-level Class A hospital (sample size: 120 cases, of which 60 cases in the intervention group used the device of the present invention and 60 cases in the control group used a conventional alternating air mattress).

[0062] Table 2 Performance Comparison of the Invention Device and Existing Technologies

[0063] Table 3 Summary of Trigger Thresholds and Response Actions for Each Warning Type

[0064] As can be seen from the table, this invention has achieved excellent results in all aspects. Specifically, the second early warning function introduces machine learning time-series prediction technology into pressure ulcer care, thereby transforming passive detection into proactive prediction. Simultaneously, precise local decompression replaces the traditional alternating inflation of the entire pad, reducing energy consumption and avoiding the insecurity caused to patients by frequent overall pressure changes.

[0065] In summary, those skilled in the art can adjust the grid density, threshold parameters, or model structure details according to actual clinical needs, but such adjustments, if they do not depart from the core features defined in the claims of this invention, still fall within the protection scope of this invention.

[0066] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. An active nursing device with an early warning function, characterized in that, include: The nursing pad body has a multi-layered structure, consisting of a breathable layer, a sensor layer, an actuator layer, and a base layer, arranged from the patient contact surface downwards. The sensor layer includes pressure sensor units, temperature sensor units, and humidity sensor units arranged in an M×N grid. The actuator layer includes multiple independently controlled micro airbags that correspond one-to-one with the positions of the sensor units, and each micro airbag is connected to an independent inflation / deflation valve. The control unit, electrically connected to all sensor units and all charge / discharge valves, includes: The data acquisition unit is used to acquire pressure, temperature and humidity values ​​from each sensor unit at a predetermined frequency. The risk assessment unit is used to calculate a real-time risk index for the location of each sensor unit, using the following formula: ， Where P is the pressure value, T is the temperature value, H is the humidity value, t is the time during which the pressure exceeds the preset pressure threshold, and w1, w2, w3, and w4 are weighting coefficients. max T max H max The maximum reference value is t0, and the reference time is t0. An active intervention unit is used to control the micro airbag corresponding to any sensor unit to perform a decompression operation when the R value of any sensor unit is greater than the first risk threshold R1. The decompression operation includes deflating the micro airbag to a first predetermined air pressure and simultaneously controlling multiple micro airbags adjacent to the sensor unit to inflate to a second predetermined air pressure. The first early warning unit is used to trigger a first early warning signal when the R value of any sensor unit is greater than the second risk threshold R2, where R2>R1. The first early warning signal includes the location information of the high-risk sensor unit and alerts the nursing staff through an audible and visual alarm. The prediction and early warning unit is used to input the historical R value sequence of each sensor unit into the pre-trained time series prediction model to predict the R value after a future time Δt. When the R prediction value is greater than the third risk threshold R3, a second early warning signal is triggered, which indicates the location and expected time of the pressure ulcer risk.

2. The active nursing device with early warning function according to claim 1, characterized in that, The weighting coefficients w1, w2, w3, and w4 satisfy the following conditions: w1=0.5, w2=0.2, w3=0.2, w4=0.1, and the first risk threshold R1=0.4, the second risk threshold R2=0.7, the third risk threshold R3=0.6, and the preset time Δt is 30 minutes.

3. The active nursing device with early warning function according to claim 1, characterized in that: The time series prediction model is a neural network model based on a long short-term memory network (LSTM) or a gated recurrent unit (GRU). This neural network model is trained by collecting historical sensor data from the patient over the previous 72 hours.

4. The active nursing device with early warning function according to claim 1, characterized in that: The device also includes a body movement monitoring module connected to the control unit. The body movement monitoring module calculates the body movement frequency by the number of times the pressure change amplitude of the pressure sensor unit exceeds the change threshold per unit time. When the body movement frequency is lower than the first frequency threshold and the duration exceeds the time threshold, a third warning signal is triggered to indicate that the patient is not active enough.

5. The active nursing device with early warning function according to claim 1, characterized in that: The device also includes an edge detection module, which identifies the patient's body edge contour based on the gradient change of the pressure distribution map. When the edge contour moves less than a distance threshold within a predetermined time, it is determined that the patient is in a dangerous resting state and triggers a fourth warning signal.

6. The active nursing device with early warning function according to claim 1, characterized in that: The nursing pad body also includes multiple tilt sensors set along the edge to detect the tilt angle of the nursing pad body. The control unit determines whether the patient is in a position at risk of falling out of bed based on the tilt angle. When the tilt angle exceeds the angle threshold and the pressure distribution map shows that the patient's center of gravity is biased towards the edge, the fifth warning signal is triggered and the micro airbags on the edge are quickly inflated to form a barrier.

7. The active nursing device with early warning function according to claim 1, characterized in that: The device also includes a wireless communication module, which sends the first warning signal, the second warning signal, and the risk index R value distribution map to a remote monitoring terminal or nurse station.

8. The active nursing device with early warning function according to claim 1, characterized in that: During the decompression operation, the micro-airbag is deflated to a first predetermined pressure of 0.1-0.3 atmospheres, and the adjacent micro-airbag is inflated to a second predetermined pressure of 0.6-0.8 atmospheres. The decompression operation lasts for 2-5 minutes, and then automatically returns to the initial pressure.

9. The active nursing device with early warning function according to claim 1, characterized in that: It also includes a patient identification module, which is an RFID reader or a QR code scanner, used to read patient information and associate and store the patient's sensor data and warning history with the patient's file.

10. A nursing method using an active nursing device with an early warning function, comprising the active nursing device with an early warning function as described in any one of claims 1-9, characterized in that, Includes the following steps: S1, initialize the device and collect pressure, temperature and humidity data of the contact surface between the patient and the nursing pad; S2, calculates the risk index R in real time for each sensor unit location; S3, when any R value is greater than the first risk threshold R1, perform active decompression operation: control the corresponding micro airbag to deflate while controlling the adjacent micro airbag to inflate. S4, when any R value is greater than the second risk threshold R2, the first warning signal is triggered; S5. Input the historical R value sequence into the prediction model to predict the future R prediction value. When the R prediction value is greater than the third risk threshold R3, trigger the second early warning signal. S6. Repeat steps S1-S5 to achieve proactive care and early warning of pressure ulcer risk.

Images