Intelligent bed detection method and system

Abstract

The invention provides an intelligent bed detection method and system, relates to the technical field of medical care, and is suitable for an intelligent bed, the intelligent bed comprises a bed and a bedside guardrail, the bed is provided with a corresponding first sensor, the bedside guardrail is provided with an intelligent detector, and the intelligent detector comprises a sensor assembly. A first sensor signal collected by the first sensor and a guardrail state signal, collected by the sensor assembly, of the bedside guardrail are preprocessed to obtain first sensor information and guardrail state information; predicting a patient state of a patient corresponding to the intelligent bed through a pre-trained patient state recognition model, the first sensor information and the guardrail state information; if the patient is in the off-bed state, counting off-bed time when the patient state is in the off-bed state in real time, and detecting whether the off-bed time reaches a target off-bed time threshold value matched with the patient or not; and if yes, outputting early warning information, so that the patient state can be mastered by paying attention to the behavior actions of the patient and the use state of the sickbed, the patient safety is ensured, and the nursing quality is improved.

Description

Technical Field

[0001] The present invention relates to the field of medical care technology, and more particularly to a smart bed detection method and system. Background Art

[0002] In healthcare scenarios, timely and accurate understanding of a patient's condition is crucial for ensuring patient safety and improving the quality of care. Current patient monitoring devices mostly rely on direct monitoring of physiological parameters, such as heart rate and blood pressure, but lack the ability to focus on patient behavior and bed usage to detect potential risks.

[0003] Therefore, how to provide an intelligent bedside detection method to pay attention to the patient's behavior and the usage status of the bed to accurately grasp the patient's condition, ensure the patient's safety and improve the corresponding nursing quality is a problem that this application urgently needs to solve. Summary of the Invention

[0004] In view of this, the present invention provides a smart bed detection method and system, which aims to pay attention to the patient's behavior and the usage status of the bed to accurately grasp the patient's status, ensure the patient's safety and improve the corresponding nursing quality.

[0005] A first aspect of the present application provides a smart bed detection method applicable to a smart bed, wherein the smart bed includes a bed and a bedside guardrail, a corresponding first sensor is installed on the bed, and a smart detector is installed on the bedside guardrail, wherein the smart detector includes a sensor assembly. The method includes:

[0006] receiving a first sensor signal acquired by the first sensor and a guardrail status signal of the bedside guardrail acquired by the sensor assembly;

[0007] Preprocessing the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information;

[0008] Predicting the patient state of the patient corresponding to the smart bed by using a pre-trained patient state recognition model, first sensor information, and guardrail state information; wherein the patient state is a state of leaving the bed or a state of not leaving the bed;

[0009] If the patient is in the out-of-bed state, real-time statistics are collected on the patient's out-of-bed time, and a detection is made as to whether the out-of-bed time reaches a target out-of-bed time threshold matched to the patient; wherein the target out-of-bed time threshold is generated based on the patient information of the patient;

[0010] If it is detected that the time to leave the bed reaches the target time to leave the bed threshold, a corresponding warning message is output.

[0011] Optionally, the guardrail status signal includes a second sensor signal, a third sensor signal, and a fourth sensor signal;

[0012] The preprocessing of the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information includes:

[0013] For each sensor signal, performing a noise characteristic analysis on the sensor signal to obtain a corresponding noise source and signal type, and performing a corresponding filtering operation on the sensor signal according to the noise source, signal type, and the sensor type corresponding to the sensor signal;

[0014] Amplifying the sensor signal after filtering to obtain corresponding sensor information;

[0015] Among them, the sensor signal is the first sensor signal, or the second sensor signal, or the third sensor signal, or the fourth sensor signal; the sensor information is the first sensor information, or the second sensor information, or the third sensor information, or the fourth sensor information; the second sensor information, the third sensor information and the fourth sensor information constitute the guardrail status information.

[0016] Optionally, predicting the patient state of the patient corresponding to the smart bed by using a pre-trained patient state recognition model, first sensor information, and guardrail state information includes:

[0017] extracting a first time characteristic value, a first frequency characteristic value, and a patient pressure characteristic value from the first sensor information;

[0018] extracting a second time feature value, a second frequency feature value, and a motion feature value from the second sensor information;

[0019] Extracting a third time characteristic value, a third frequency characteristic value, and a guardrail pressure characteristic value from the third sensor information;

[0020] Extracting a fourth time characteristic value, a fourth frequency characteristic value, and a guardrail position change characteristic value from the fourth sensor information;

[0021] Each characteristic value is input into a pre-trained patient state recognition model, so that the patient state recognition model performs recognition according to each characteristic value to obtain the patient state of the patient corresponding to the smart bed; wherein the patient state recognition model is obtained by training the patient state recognition model to be trained using multiple historical sensor signals.

[0022] Optionally, the step of training a patient state recognition model to be trained using a plurality of historical sensor signals to obtain the patient state recognition model includes:

[0023] Acquiring historical sensor data of the smart bed, wherein the historical sensor data includes a plurality of historical sensor signals and corresponding target patient states; the plurality of historical sensor signals includes a historical first sensor signal, a historical second sensor signal, a historical third sensor signal, and a historical fourth sensor signal; and the target patient state includes an out-of-bed state and a non-out-of-bed state;

[0024] Preprocessing each of the historical sensor signals to obtain corresponding historical sensor information;

[0025] Performing feature extraction on each piece of the historical sensor information to obtain a plurality of historical feature values ​​of each piece of the historical sensor information;

[0026] The historical characteristic values ​​of each of the historical sensor information and the target patient state are input into the patient state recognition model to be trained, so that the patient state recognition model to be trained performs recognition according to the historical characteristic values ​​of each of the historical sensor information, and takes the recognized patient state approaching the target patient state as the training goal, and adjusts the parameters of the patient state recognition model to be trained until the patient state recognition model to be trained reaches convergence, thereby obtaining a patient state recognition model.

[0027] Optionally, generating the target bed-out time threshold for the patient according to the patient information of the patient includes:

[0028] Acquiring patient information of the patient, wherein the patient information includes basic information, condition information, and activity ability information of the patient;

[0029] A pre-built dynamic threshold library is obtained, and a target bed leaving time threshold matching the patient is determined from each bed leaving time in the dynamic threshold library according to the basic information, the condition information, and the activity ability information.

[0030] Optionally, the method further includes:

[0031] When any sensor is detected to be abnormal, continuous sensor signals sent by sensors that are not abnormal are received; wherein the sensor is the first sensor or any sensor in the sensor assembly;

[0032] Determining whether the continuous sensor signal of each sensor that has not experienced an abnormality satisfies a matching patient in bed condition or patient out of bed condition;

[0033] If the continuous sensor signals of all sensors without abnormalities satisfy the matching patient in bed condition, determining that the patient status of the patient is in bed;

[0034] If the continuous sensor signals of all sensors without abnormalities meet the matching patient leaving bed condition, determining that the patient's patient status is the leaving bed state, and performing the step of real-time counting of the patient's leaving bed time when the patient's patient status is the leaving bed state;

[0035] If any sensor that has not experienced an abnormality continuously sends a sensor signal that does not meet the matching patient leaving bed condition, or sends a sensor signal that does not meet the matching patient staying in bed condition, corresponding abnormality information is displayed.

[0036] A second aspect of the present application provides a smart bed detection system applicable to a smart bed, wherein the smart bed includes a bed and a bedside guardrail, a corresponding first sensor is installed on the bed, and a smart detector is installed on the bedside guardrail, wherein the smart detector includes a sensor assembly, and the system includes:

[0037] a data processing module for receiving a first sensor signal acquired by the first sensor and a guardrail status signal of the bedside guardrail acquired by the sensor assembly; preprocessing the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information; predicting the patient status of the patient corresponding to the smart bed through a pre-trained patient status recognition model, the first sensor information, and the guardrail status information; wherein the patient status is either an out-of-bed state or a non-out-of-bed state; if the patient status is an out-of-bed state, real-time statistics are collected on the out-of-bed time when the patient status is an out-of-bed state, and detection is performed on whether the out-of-bed time reaches a target out-of-bed time threshold matched to the patient; wherein the target out-of-bed time threshold is generated based on the patient information of the patient;

[0038] The intelligent early warning module is used to output corresponding early warning information if it is detected that the time to leave the bed reaches the target time to leave the bed threshold.

[0039] Optionally, the guardrail status signal includes a second sensor signal, a third sensor signal, and a fourth sensor signal. A data processing module is provided for preprocessing the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information, specifically for:

[0040] For each sensor signal, performing a noise characteristic analysis on the sensor signal to obtain a corresponding noise source and signal type, and performing a corresponding filtering operation on the sensor signal information according to the noise source, signal type, and the sensor type corresponding to the sensor signal;

[0041] Amplifying the sensor signal after filtering to obtain corresponding sensor information;

[0042] Among them, the sensor signal is the first sensor signal, or the second sensor signal, or the third sensor signal, or the fourth sensor signal; the sensor information is the first sensor information, or the second sensor information, or the third sensor information, or the fourth sensor information; the second sensor information, the third sensor information and the sensor information constitute the guardrail status information.

[0043] Optionally, the data processing module for predicting the patient status of the patient corresponding to the smart bed by using the pre-trained patient status recognition model, the first sensor information, and the guardrail status information is specifically configured to:

[0044] extracting a first time characteristic value, a first frequency characteristic value, and a patient pressure characteristic value from the first sensor information;

[0045] extracting a second time feature value, a second frequency feature value, and a motion feature value from the second sensor information;

[0046] Extracting a third time characteristic value, a third frequency characteristic value, and a guardrail pressure characteristic value from the third sensor information;

[0047] Extracting a fourth time characteristic value, a fourth frequency characteristic value, and a guardrail position change characteristic value from the fourth sensor information;

[0048] Each characteristic value is input into a pre-trained patient state recognition model, so that the patient state recognition model performs recognition according to each characteristic value to obtain the patient state of the patient corresponding to the smart bed; wherein the patient state recognition model is obtained by training the patient state recognition model to be trained using multiple historical sensor signals.

[0049] Optionally, the data processing module for training a patient state recognition model to be trained using a plurality of historical sensor signals to obtain a patient state recognition model is specifically configured to:

[0050] Acquiring historical sensor data of the smart bed, wherein the historical sensor data includes a plurality of historical sensor signals and corresponding target patient states; the plurality of historical sensor signals includes a historical first sensor signal, a historical second sensor signal, a historical third sensor signal, and a historical fourth sensor signal; and the target patient state includes an out-of-bed state and a non-out-of-bed state;

[0051] Preprocessing each of the historical sensor signals to obtain corresponding historical sensor information;

[0052] Performing feature extraction on each piece of the historical sensor information to obtain a plurality of historical feature values ​​of each piece of the historical sensor information;

[0053] The historical characteristic values ​​of each of the historical sensor information and the target patient state are input into the patient state recognition model to be trained, so that the patient state recognition model to be trained performs recognition according to the historical characteristic values ​​of each of the historical sensor information, and takes the recognized patient state approaching the target patient state as the training goal, and adjusts the parameters of the patient state recognition model to be trained until the patient state recognition model to be trained reaches convergence, thereby obtaining a patient state recognition model.

[0054] The present application provides a smart bed detection method and system, applicable to a smart bed, wherein the smart bed includes a bed and a bedside guardrail, wherein a corresponding first sensor is installed on the bed, and an intelligent detector is installed on the bedside guardrail, wherein the intelligent detector includes a sensor component, which receives a first sensor signal acquired by the first sensor and a guardrail status signal of the bedside guardrail acquired by the sensor component; pre-processes the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information; and predicts the patient status of a patient corresponding to the smart bed using a pre-trained patient status recognition model, the first sensor information, and the guardrail status information; wherein the patient status is either an out-of-bed state or a non-out-of-bed state; if the patient status is an out-of-bed state, the out-of-bed time of the patient in the out-of-bed state is counted in real time, and whether the out-of-bed time reaches a target out-of-bed time threshold matched to the patient is detected; wherein the target out-of-bed time threshold is generated based on the patient information of the patient; and if it is detected that the out-of-bed time reaches the target out-of-bed time threshold, a corresponding warning message is output. The technical solution provided by the present application collects sensor signals related to the smart bed in real time through a first sensor, and collects guardrail status signals of the bedside guardrail in real time through a sensor component, so as to process these sensor signals accordingly to obtain corresponding sensor information, and analyze and identify the sensor information to obtain the patient status of the corresponding patient, so as to utilize the patient status of the patient and the target bed leaving event threshold risk monitoring determined according to the patient information of the patient, so as to trigger a corresponding early warning when it is determined that there is a risk. It can be seen that the present application focuses on the patient's condition and the usage status of the bedside guardrail of the smart bed, accurately grasps the patient's status, can ensure the patient's safety and improve the corresponding nursing quality, and reduce the probability of patient risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0056] Figure 1 A schematic structural diagram of a smart bed provided in an embodiment of the present application;

[0057] Figure 2 A flow chart of a smart bed detection method provided in an embodiment of the present application;

[0058] Figure 3 A schematic diagram of the structure of a nurse station monitoring terminal provided in an embodiment of the present application;

[0059] Figure 4 This is a structural diagram of a smart bed detection system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0060] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0061] In this application, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus comprising the element.

[0062] See also Figure 1 , shows a structural schematic diagram of a smart bed provided in an embodiment of the present application, the smart bed includes a bed and a bedside guardrail, a corresponding first sensor is installed on the bed, and a smart detector is installed on the bedside guardrail, and the smart detector includes a sensor component.

[0063] It should be noted that the first sensor can be a pressure sensor, which is used to detect pressure changes of the patient on the bed and generate a corresponding pressure sensor signal based on the detected pressure changes (for ease of distinction, the pressure sensor signal collected by the first sensor is referred to as the first sensor signal).

[0064] In the embodiment of the present application, the sensor assembly includes a second sensor, a third sensor, and a fourth sensor.

[0065] It should be noted that the second sensor can be a micro switch or a photoelectric sensor, which is used to detect the opening and closing movements of the bedside guardrail in real time; when the bedside guardrail is detected to be opened or closed, a corresponding motion sensor signal is generated (for ease of distinction, the motion sensor signal generated by the second sensor is referred to as the second sensor signal).

[0066] The third sensor can be a pressure sensor, which can sense the pressure changes generated when the bedside guardrail is lowered or raised (opened or closed). Specifically, it can detect the operating handle on the bedside guardrail in real time; when it detects that the operating handle is pressed or released, it generates a corresponding pressure sensor signal (for ease of distinction, the pressure sensor signal generated by the third sensor according to the pressure change on the operating handle is called the third sensor signal).

[0067] The fourth sensor can be a Hall sensor, which can use changes in the magnetic field to detect the position of the bedside guardrail. Specifically, the position of the bedside guardrail can be detected in real time. When a change in the position of the bedside guardrail is detected, it can be determined that the surrounding magnetic field has changed, thereby generating a corresponding Hall sensor signal (for ease of distinction, the Hall sensor signal generated by the fourth sensor is referred to as the fourth sensor signal).

[0068] In some embodiments, when the second sensor is a micro switch, the bedside guardrail can be detected in real time through the micro switch; when the bedside guardrail is detected to be open or closed, the corresponding mechanical structure action can be triggered to generate a corresponding motion sensor signal (second sensor signal); when the second sensor is a photoelectric sensor, the photoelectric sensor can be used to detect the light obstruction of the bedside guardrail, so as to determine the state of the bedside guardrail (open or closed) according to the detected light obstruction, and then generate a corresponding motion sensor signal (second sensor signal).

[0069] It should be noted that photoelectric sensors are more suitable for medical environments that are sensitive to electromagnetic interference and have the advantage of fast response speed.

[0070] based on Figure 1 The smart bed shown, accordingly, the embodiment of the present application provides a smart bed detection method, such as Figure 2 As shown, the smart bed detection method is applicable to Figure 1 The smart bed shown in FIG. 1 includes the following steps:

[0071] S201: Receive a first sensor signal collected by a first sensor and a guardrail status signal of a bedside guardrail collected by a sensor assembly.

[0072] In an embodiment of the present application, the first sensor can detect the bed in the smart bed in real time to detect the pressure changes of the user on the bed in real time and send a corresponding pressure sensor signal (first sensor signal); the second sensor in the sensor assembly can detect the opening and closing of the bedside guardrail of the smart bed in real time and send a corresponding motion sensor signal (second sensor signal); the third sensor assembly can detect the release of the operating handle on the bedside guardrail in real time and send a corresponding pressure sensor signal (third sensor signal); the fourth sensor can detect the movement of the bedside guardrail in real time and send a corresponding Hall sensor signal, that is, send a corresponding position change sensor signal (fourth sensor signal).

[0073] In summary, the present application forms a four-dimensional detection matrix using the first sensor and the sensor assembly to respectively collect the operating pressure of the bed, the mechanical movement of the bedside guardrail opening and closing, the operating pressure of the bedside guardrail operating handle, and the magnetic field position change of the bedside guardrail in the smart bed. This allows the detected operating pressure, mechanical movement, and magnetic field position change to be used to accurately predict the patient status of the patient using the smart bed. The patient status is either out of bed or in bed.

[0074] It should be noted that the sampling frequency of each sensor can be set to 200Hz to ensure zero-delay capture of state changes of the bed, bedside guardrails, and operating handles. The sampling frequency of each sensor can be set according to actual application and is not limited in this embodiment of the application.

[0075] It's also worth noting that a dual-protocol network combining Zigbee (short-range) and 5G (long-range) can be used. Once sensors collect sensor signals, they can transmit them via the Zigbee 3.0 protocol. This transmission method offers low power consumption and short transmission delays.

[0076] During the specific execution of step S201, the first sensor signal collected and sent by the first sensor and the guardrail status signal of the bedside guardrail collected and sent by the sensor component can be detected and received in real time; wherein the guardrail status signal includes the second sensor signal, the third sensor signal and the fourth sensor signal.

[0077] S202: Preprocess the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information.

[0078] In the specific process of executing step S202, after receiving each sensor signal, for each sensor signal, a corresponding filtering operation can be performed on each sensor signal, and the sensor signal after the filtering operation can be subjected to signal enhancement processing to convert the sensor signal transmitted by the sensor into recognizable sensor information; wherein, the sensor signal is the first sensor signal, or the second sensor signal, or the third sensor signal, or the fourth sensor signal; the sensor information is the first sensor information, or the second sensor information, or the third sensor information, or the fourth sensor information; the second sensor information, the third sensor information and the fourth sensor information constitute the guardrail status information.

[0079] Optionally, the first sensor signal and the guardrail status signal are preprocessed to obtain the first sensor information and the guardrail status information. The process can be: for each sensor signal, the noise characteristics of the sensor signal are analyzed to obtain the corresponding noise source and signal type, and the sensor signal information is filtered according to the noise source, signal type and sensor type corresponding to the sensor signal; the sensor signal after the filtering operation is amplified to obtain the corresponding sensor information.

[0080] It should be noted that the sensor type can be a mechanical action type, an operating pressure type, and a magnetic field position change type. Among them, the sensor corresponding to the mechanical action type can be a micro switch, the sensor corresponding to the operating pressure type can be a pressure sensor, and the sensor corresponding to the magnetic field position change type can be a Hall sensor.

[0081] It should also be noted that the noise source may include low-frequency noise and high-frequency noise.

[0082] In the embodiment of the present application, filtering the sensor signal can effectively remove the noise signal generated by external interference; amplifying the sensor signal after the filtering operation can enhance the strength of the signal.

[0083] As an implementation method of an embodiment of the present application, when the first sensor is a pressure sensor, that is, the sensor type is an operating pressure type, the first sensor signal is a pressure sensor signal, and then the noise characteristic analysis can be performed on the first sensor signal to identify the low-frequency noise, high-frequency noise and corresponding signal type in the first sensor signal; the high-frequency noise in the first sensor is filtered using a double-order RC low-pass filter to remove the corresponding noise; the low-frequency noise in the first sensor signal is filtered using a sliding average filter and a wavelet denoising algorithm to remove the corresponding noise; and the signal after the filtering operation is amplified using an instrument amplifier to eliminate the corresponding common-mode noise (such as ground loop interference), and the baseline offset in the signal is corrected using a temperature compensation circuit to obtain the corresponding first sensor information.

[0084] It should be noted that the first sensor information may include key parameters such as the corresponding state of the smart bed (whether the patient is in bed or not in bed), the corresponding state change time (in bed or not in bed), and the corresponding state change frequency.

[0085] As an implementation method of an embodiment of the present application, when the second sensor is a micro-switch, that is, the sensor type is a mechanical action type, the second sensor signal is an action sensor signal, and then the noise feature analysis can be performed on the second sensor signal to identify the physical jitter, electric spark interference, external electromagnetic coupling noise and corresponding signal type caused by the closing / opening of the mechanical contact in the second sensor signal; the noise is filtered using an RC elimination circuit to eliminate the high-frequency burrs caused by mechanical jitter; the noise is filtered using electromagnetic shielding to reduce the coupling of external electromagnetic fields to the signal line and remove external electromagnetic coupling noise; the validity of the signal is detected using a delayed debouncing algorithm and a state threshold; when it is determined that the signal is valid, the corresponding signal of the signal after the filtering operation is amplified to obtain the corresponding second sensor information.

[0086] It should be noted that the second sensor information may include key parameters such as whether the bedside guardrail is open or closed (lowered or raised), the corresponding state change (open or closed) time, and the corresponding state change frequency.

[0087] It's also worth noting that the delay debounce algorithm uses a timer to detect the level of the signal. If the level within the timer doesn't change again, the signal is considered valid (to avoid false triggering). The status threshold can be set to 5 times. In other words, if the second sensor signals the same level five times, the signal is considered valid.

[0088] As an implementation method of an embodiment of the present application, when the third sensor is a pressure sensor, that is, the sensor type is an operating pressure type, the third sensor signal is a pressure sensor signal, and then the noise characteristic analysis of the third sensor signal can be performed to identify the low-frequency noise, high-frequency noise and corresponding signal type in the third sensor signal; the high-frequency noise in the third sensor signal is filtered using a double-order RC low-pass filter to remove the corresponding noise; the low-frequency noise in the third sensor signal is filtered using a sliding average filter and a wavelet denoising algorithm to remove the corresponding noise; and the signal after the filtering operation is amplified using an instrument amplifier to eliminate the corresponding common-mode noise (such as ground loop interference), and the baseline offset in the signal is corrected using a temperature compensation circuit to obtain the corresponding third sensor information.

[0089] It should be noted that the third sensor information may include key parameters such as whether the operating handle is pressed or released, the corresponding state change (pressed or released) time, and the corresponding state change frequency.

[0090] As an implementation method of an embodiment of the present application, when the fourth sensor is a Hall sensor, that is, the sensor type is a magnetic field position change type, the fourth sensor signal is a Hall sensor signal, and then the fourth sensor signal can be subjected to noise characteristic analysis to identify high-frequency noise, hysteresis effect, temperature drift and corresponding signal type in the fourth sensor signal; a bandpass filter is used to filter the high-frequency noise in the fourth sensor to remove the corresponding noise; the hysteresis characteristic is used to eliminate edge oscillations caused by the hysteresis effect; the series common-mode inductor is used to suppress the common-mode noise in the signal; the temperature compensation algorithm is used to remove the temperature drift in the signal; and the corresponding signal of the signal after the filtering operation is amplified to obtain the corresponding fourth sensor information.

[0091] It should be noted that the fourth sensor information may include key parameters such as the moving position of the bedside guardrail, the corresponding state change (moving position of the bedside guardrail) time, and the corresponding state change frequency.

[0092] S203: Predicting the patient state of the patient corresponding to the smart bed by using the pre-trained patient state recognition model, the first sensor information, and the guardrail state information.

[0093] In an embodiment of the present application, a deep learning model (a patient status recognition model to be trained) can be pre-trained using multiple historical sensor signals, allowing the model to automatically learn the features and patterns in the data, thereby more accurately judging the patient's status and analyzing potential risks, thereby obtaining a patient status recognition model that can accurately identify the patient's status.

[0094] It should be noted that using a deep learning model as the patient status recognition model to be trained can have advantages in processing complex data and status recognition methods, and may further improve the accuracy of detection and early warning.

[0095] It should also be noted that the deep learning model can be a hidden Markov model (HMM), and the corresponding deep learning model can be selected according to actual usage, which is not limited in this embodiment of the present application.

[0096] In the specific process of executing step S203, after converting each sensor signal into identifiable sensor information, the wavelet transform method can be further used to extract features of each sensor information to obtain multiple eigenvalues ​​of each sensor information, and each eigenvalue is input into the patient state recognition model, so that the patient state recognition model can identify the patient state corresponding to the smart bed according to the eigenvalues ​​of each sensor information.

[0097] Optionally, the process of predicting the patient state of the patient corresponding to the smart bed through a pre-trained patient state recognition model, first sensor information and guardrail state information can be: extracting a first time characteristic value, a first frequency characteristic value and a patient pressure characteristic value from the first sensor information; extracting a second time characteristic value, a second frequency characteristic value and a motion characteristic value from the second sensor information; extracting a third time characteristic value, a third frequency characteristic value and a guardrail pressure characteristic value from the third sensor information; extracting a fourth time characteristic value, a fourth frequency characteristic value and a guardrail position change characteristic value from the fourth sensor information; inputting each characteristic value into the pre-trained patient state recognition model, so that the patient state recognition model performs recognition according to each characteristic value to obtain the patient state of the patient corresponding to the smart bed; wherein the patient state recognition model is obtained by training the patient state recognition model to be trained using multiple historical sensor signals.

[0098] It should be noted that the various characteristic values ​​may include: first time characteristic value, first frequency characteristic value, patient pressure characteristic value, second time characteristic value, second frequency characteristic value, action characteristic value, third time characteristic value, third frequency characteristic value, guardrail pressure characteristic value, fourth time characteristic value, fourth frequency characteristic value and guardrail position change characteristic value.

[0099] In an embodiment of the present application, the process of using multiple historical sensor signals to train a patient state recognition model to obtain a patient state recognition model can be: obtaining historical sensor data of the smart bed, wherein the historical sensor data includes multiple historical sensor signals and corresponding target patient states; the multiple historical sensor signals include a historical first sensor signal, a historical second sensor signal, a historical third sensor signal and a historical fourth sensor signal; the target patient state is an out-of-bed state and a not-out-of-bed state; preprocessing each historical sensor signal to obtain corresponding historical sensor information; performing feature extraction on each historical sensor information to obtain multiple historical feature values ​​of each historical sensor information; inputting each historical feature value of each historical sensor information and the target patient state into the patient state recognition model to be trained, so that the patient state recognition model to be trained performs recognition according to each historical feature value of each historical sensor information, and taking the recognized patient state approaching the target patient state as the training goal, adjusting the parameters of the patient state recognition model to be trained until the patient state recognition model to be trained converges to obtain a patient state recognition model.

[0100] It should be noted that the historical first time characteristic value, historical first frequency characteristic value and historical patient pressure characteristic value can be extracted from the historical first sensor information; the historical second time characteristic value, historical second frequency characteristic value and motion characteristic value can be extracted from the historical second sensor information; the historical third time characteristic value, historical third frequency characteristic value and historical guardrail pressure characteristic value can be extracted from the historical third sensor information; and the historical fourth time characteristic value, historical fourth frequency characteristic value and historical guardrail position change characteristic value can be extracted from the historical fourth sensor signal.

[0101] S204: Determine whether the patient's status is out of bed; if the patient's status is out of bed, execute step S205.

[0102] In the specific process of executing step S204, after obtaining the patient's patient status, it can be further determined whether the patient's patient status is the bed leaving state; if it is the bed leaving state, execute step S205 to count the patient's bed leaving time; if it is not the bed leaving state, that is, it is determined that the bed leaving state is the bed leaving state, it can return to execute step S201 to continue monitoring the patient's patient status until it is detected that the patient's patient status is the bed leaving state, and then execute step S205.

[0103] S205: Real-time statistics are collected on the patient's out-of-bed time, and the out-of-bed time is detected to see if it reaches the target out-of-bed time threshold matched to the patient. If it is detected that the out-of-bed time reaches the target out-of-bed time threshold, step S206 is executed.

[0104] In an embodiment of the present application, historical patient information of each historical patient can be obtained, wherein the historical patient information includes basic information, medical condition information, and activity ability information of the historical patient; the basic information, medical condition information, and activity ability information of the historical patient are analyzed to determine the corresponding bed leaving time threshold, and the basic information, medical condition information, and activity ability information of the historical patient are bound to the corresponding bed leaving time threshold, so as to construct a corresponding dynamic threshold library with the bound basic information, medical condition information, and activity ability information of the historical patient and the corresponding bed leaving time threshold.

[0105] For example, if the historical patient information of a historical patient is elderly, has a stroke, and has poor mobility, the corresponding bed leaving time threshold may be 15 seconds; if the historical patient information of a historical patient is middle-aged, has common diseases, and has medium mobility, the corresponding bed leaving time threshold may be 30 seconds. This embodiment of the present application is not limited to this.

[0106] It should be noted that, since there is a certain delay time in collecting corresponding sensor signals through sensors, each bed leaving time threshold in the dynamic threshold library can be further adjusted according to the transmission delay time of the sensor signal.

[0107] It should also be noted that combining the patient's own basic information, medical condition information and activity ability to determine the corresponding target bed leaving time threshold can effectively solve the problem of triggering the corresponding warning according to a fixed time threshold in the existing technology, which easily leads to a large number of false alarms, low warning reliability, and interference with the normal work of nursing staff.

[0108] During the execution of step S205, when it is determined that the patient status of the patient of the smart bed is the out-of-bed state, the patient information of the patient can be further obtained, and a target out-of-bed time threshold value matching the patient can be determined from the dynamic threshold library based on the patient information; steps S201 to S204 are executed in a loop to detect the patient status of the patient in real time, to count the out-of-bed time when the patient status is the out-of-bed state, and to detect in real time whether the counted out-of-bed time reaches the target out-of-bed threshold time; if it is detected that the out-of-bed time reaches the target out-of-bed time threshold, step S206 is executed.

[0109] It should be noted that if the counted bed leaving time does not reach the target bed leaving time threshold, it has been detected that the patient's patient status is not in bed. At this time, the counted bed leaving time can be counted and the process returns to step S201 to re-monitor the patient's patient status.

[0110] Optionally, the process of generating a target bed exit time threshold that matches the patient based on the patient information of the patient may be: obtaining the patient information of the patient, wherein the patient information includes the patient's basic information, medical condition information, and mobility information; obtaining a pre-built dynamic threshold library, and determining a target bed exit time threshold that matches the patient from each bed exit time in the dynamic threshold library based on the basic information, medical condition information, and mobility information.

[0111] In actual application, the similarity between the patient information of the patient and the patient information of historical patients in the dynamic threshold library can be calculated, and the bed leaving time threshold corresponding to the patient information of the historical patient with high similarity can be determined as the target bed leaving time threshold for patient matching.

[0112] S206: Output corresponding warning information.

[0113] In the specific process of executing step S206 , when it is determined that the time for the patient to leave the bed when the patient's status is out of bed reaches the target time threshold, a preset warning method is used to output corresponding warning information.

[0114] In the embodiment of the present application, the warning method can be: corresponding warning information can be sent to the nurse station monitoring terminal, so that after receiving the warning information, the nurse station monitoring terminal broadcasts the corresponding warning information through the corresponding voice announcer to prompt the nursing staff to check it in time. Figure 3 As shown, the nurse station monitoring terminal includes at least a micro-electric switch and a voice announcer.

[0115] It should be noted that the corresponding warning information can be sent through the 5G SA network (bandwidth ≥ 100Mbps).

[0116] In some embodiments, corresponding handheld devices can also be configured for corresponding nursing staff so that corresponding warning information can be sent to the corresponding nursing staff's handheld devices so that the nursing staff can detect and take corresponding measures in time to prevent unexpected events such as patients falling.

[0117] It should be noted that the handheld device may be a smart watch, which is not limited in the embodiments of the present application.

[0118] In some embodiments, a corresponding edge server (4-core CPU + 8GB memory) can be deployed at the nurse station monitoring terminal to complete the corresponding data preprocessing (filtering and signal amplification of the sensor signals of each sensor, and feature extraction of each sensor information) through the edge service. At the same time, the generated early warning information can be uploaded to the cloud to reduce network bandwidth occupancy.

[0119] In summary, the research found that the bedside guardrail is an important part of the hospital bed. Its state changes when it is lowered or raised, as well as its changes in movement position are closely related to the patient's movements. Therefore, various sensors are used to monitor the relevant states of the bedside guardrail and the patient in real time, and transmit corresponding sensor signals; the received sensor signals are filtered, amplified, denoised and other algorithms are used to perform corresponding processing and feature extraction to obtain characteristic values ​​such as the time and frequency of the guardrail state and its changes, and characteristic values ​​such as the time and frequency of the patient's state and its changes; the patient's state is identified according to each characteristic value, so that corresponding warnings can be issued according to the patient's state and the target bed leaving time threshold matched with the patient, which can effectively prevent accidents such as patient falls.

[0120] Furthermore, in an embodiment of the present application, after sending the corresponding warning information, the camera in the room of the smart bed can also be turned on to record the corresponding video through the camera, and the recorded video can be pushed to the nursing robot, so that the nursing robot can perform corresponding warning operations based on the received video.

[0121] Furthermore, in an embodiment of the present application, each sensor can be monitored in real time. When any sensor is detected to be abnormal, the "dual sensor cross-validation" mode is automatically started (for example, when a micro switch fails, the status is judged by fusing the pressure sensor and Hall sensor data) to ensure monitoring continuity.

[0122] Optionally, when any sensor is detected to have an abnormality, continuous sensor signals sent by all sensors that have not experienced an abnormality are received; it is determined whether the continuous sensor signals of all sensors that have not experienced an abnormality meet the matching patient in-bed condition or patient out-of-bed condition; if the continuous sensor signals of all sensors that have not experienced an abnormality meet the matching patient in-bed condition, the patient's patient status is determined to be in-bed; if the continuous sensor signals of all sensors that have not experienced an abnormality meet the matching patient out-of-bed condition, the patient's patient status is determined to be out-of-bed, and the step of real-time statistics of the patient's out-of-bed time is performed; if any of the sensors that have not experienced an abnormality have a continuous sensor signal that does not meet the matching patient out-of-bed condition or a continuous sensor signal that does not meet the matching patient in-bed condition, corresponding abnormality information is displayed. The sensor is the first sensor or any sensor in the sensor assembly.

[0123] It should be noted that, for each sensor, when it is detected that the signal of the sensor is continuously abnormal, it can be determined that the sensor is abnormal.

[0124] For example, for the second sensor (micro switch), if it is detected that the micro switch signal is interrupted, the signal has no fluctuation, or there is a long-term contradiction between the signal transmitted by the micro switch and the signal transmitted by the pressure (first sensor / third sensor) / Hall sensor (fourth sensor), it can be considered that the micro switch is abnormal.

[0125] It should be noted that when the microswitch is in a normal state, the signal collected by the microswitch (whether the bedside guardrail is lowered or raised) and the signals transmitted by the pressure sensor (first sensor and second sensor) and the Hall sensor (fourth sensor) form a "quadruple confirmation" (for example, when the patient is in bed, the microswitch detects the microswitch signal, the signal transmitted by the pressure sensor contains a pressure value, and the signal transmitted by the Hall sensor indicates an active magnetic field change); therefore, if the signal transmitted by the pressure sensor indicates "pressure (patient in bed)" but the microswitch always has no signal (the transmitted signal is empty), the microswitch can be determined to be abnormal through the corresponding preset threshold (such as abnormality duration > 5 minutes) or algorithm (such as signal consistency check), and the "pressure + Hall" dual-sensor cross-validation mode is automatically triggered, that is, the continuous sensor signals sent by each sensor that has not experienced an abnormality (pressure sensor and Hall sensor that have not experienced an abnormality) are received, so that the patient's status can be determined based on the continuous sensor signals sent by each sensor that has not experienced an abnormality.

[0126] In some embodiments, the matching in-bed state conditions and out-of-bed state conditions for each sensor can be pre-set. For example, when the sensor is a pressure sensor, the matching in-bed state condition can be: pressure value ≥ preset threshold value (such as pressure ≥ 50N for adult patients), and the pressure distribution is stable (no sudden drop or return to zero); the matching out-of-bed state condition can be: pressure value ≤ preset threshold value (such as <10N), and the pressure value shows a sudden drop trend (consistent with the pressure change pattern when getting out of bed). When the sensor is a Hall sensor, the matching in-bed state condition can be: detection of continuous dynamic changes in the magnetic field (magnetic field fluctuations caused by the movement of the bedside guardrail), or the magnetic field strength is within the range corresponding to "human presence"; the matching out-of-bed state condition can be: the magnetic field strength is stable at the "no interference" baseline value, or there is no magnetic field change for a long time.

[0127] It should be noted that if the continuous sensor signals of each sensor include both a continuous sensor signal that satisfies the matching patient leaving the bed condition and a continuous sensor signal that satisfies the patient staying in bed condition, it can be considered that an abnormality has occurred, and corresponding abnormal information can be output at this time.

[0128] For example, if the continuous sensor signal of the pressure sensor indicates that the patient is "in bed" but there is no activity signal from the Hall sensor for a long time (such as more than 30 minutes), it can be combined with historical data (such as the patient's previous activity frequency) to determine it as a "potential abnormality (such as drowsiness, risk of fixed posture)", triggering an early warning and outputting the corresponding abnormal information, that is, triggering an audible and visual alarm or pushing a reminder to the medical staff's mobile terminal to ensure timely intervention.

[0129] It should also be noted that the output abnormal information and the abnormal time of the sensor where the abnormality occurred can also be recorded at the same time to facilitate subsequent device dimension and data tracing.

[0130] It is worth noting that in complex medical environments, individual sensors are susceptible to interference. When a sensor fails or is subject to external interference and produces erroneous data, the existing system lacks an effective fault-tolerance mechanism, which may cause deviations in the entire monitoring and early warning process, making it impossible to accurately reflect the true status of the bedside guardrail, thereby affecting the judgment of patient safety risks. Therefore, when this application detects an abnormality in any sensor, it automatically activates the "dual sensor cross-validation" mode (for example, when a micro switch fails, the status is determined by fusing the pressure sensor and Hall sensor data) to ensure monitoring continuity.

[0131] Furthermore, in an embodiment of the present application, the present application also has corresponding nursing record and management functions, which records, stores and manages the sensor signals of each sensor received each time, sensor information, identified patient status, and related information such as the nursing staff and their operation information who process the warning information after sending it, so that the corresponding nursing staff can query it, and provide corresponding support for nursing work review, quality assessment, and the formulation and adjustment of subsequent nursing plans.

[0132] It should be noted that the existing nursing record and management functions only stay at the simple data recording level, lacking in-depth analysis of these data. For example, it is impossible to analyze the patient's activity patterns and nursing demand trends based on the long-term accumulated guardrail status change data, and cannot provide strong support for the formulation of personalized nursing plans, which limits the further improvement of nursing quality.

[0133] Therefore, this application can use the management rule mining algorithm to mine the sensor signals, sensor information, identified patient status, and other related information such as the nursing staff and their operation information that process the early warning information after sending it, to identify the correlation between nursing operations and patient risks, so that the corresponding nursing staff can provide corresponding support for the formulation and adjustment of subsequent nursing plans based on the correlation.

[0134] Furthermore, in an embodiment of the present application, the ECharts chart engine can also be used to generate visual reports such as guardrail operation heat maps and trend curves, supporting multi-dimensional retrieval by department, patient type, and time period to further improve the corresponding nursing quality.

[0135] In some embodiments, structured records can be automatically generated based on each guardrail operation without manual entry, providing data support for subsequent nursing plans and further improving the corresponding nursing quality.

[0136] It should be noted that corresponding nursing intervention suggestions can also be automatically generated based on the generated structured records of guardrail operations to reduce the probability of patient accidents.

[0137] It should also be noted that if the structured record shows that the patient frequently gets out of bed, the generated suggestion can be to add bed guards.

[0138] In some embodiments, the nursing operations can also be recorded and stored through blockchain, enabling full traceability of the nursing operations.

[0139] Furthermore, in an embodiment of the present application, after issuing a corresponding warning message, the weight sensor and human infrared sensor of the corresponding smart field can also be automatically triggered so that the weight sensor and human infrared sensor can collect corresponding data to form a multi-dimensional risk assessment to determine whether the patient has returned to the smart bed. If the patient has not returned to the smart bed, the corresponding warning message can continue to be issued.

[0140] The present application provides a smart bed detection method, which is applicable to a smart bed. The smart bed includes a bed and a bedside guardrail. A corresponding first sensor is installed on the bed, and an intelligent detector is installed on the bedside guardrail. The intelligent detector includes a sensor component. The method receives a first sensor signal collected by the first sensor and a guardrail status signal of the bedside guardrail collected by the sensor component; pre-processes the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information; predicts the patient status of a patient corresponding to the smart bed through a pre-trained patient status recognition model, the first sensor information and the guardrail status information; wherein the patient status is either an out-of-bed state or a non-out-of-bed state; if the patient status is an out-of-bed state, the out-of-bed time of the patient when the patient status is the out-of-bed state is counted in real time, and detects whether the out-of-bed time reaches a target out-of-bed time threshold matched to the patient; wherein the target out-of-bed time threshold is generated based on the patient information of the patient; if it is detected that the out-of-bed time reaches the target out-of-bed time threshold, outputs a corresponding warning message. The technical solution provided by the present application collects sensor signals related to the smart bed in real time through a first sensor, and collects guardrail status signals of the bedside guardrail in real time through a sensor component, so as to process these sensor signals accordingly to obtain corresponding sensor information, and analyze and identify the sensor information to obtain the patient status of the corresponding patient, so as to utilize the patient status of the patient and the target bed leaving event threshold risk monitoring determined according to the patient information of the patient, so as to trigger a corresponding early warning when it is determined that there is a risk. It can be seen that the present application focuses on the patient's condition and the usage status of the bedside guardrail of the smart bed, accurately grasps the patient's status, can ensure the patient's safety and improve the corresponding nursing quality, and reduce the probability of patient risks.

[0141] Based on the smart bed detection method provided in the above embodiment of the present application, the embodiment of the present application also provides a smart bed detection system, such as Figure 4 As shown, it is applicable to a smart bed, which includes a bed and a bedside guardrail. A corresponding first sensor is installed on the bed, and a smart detector is installed on the bedside guardrail. The smart detector includes a sensor component. The smart bed detection system includes:

[0142] a data processing module for receiving a first sensor signal acquired by a first sensor and a guardrail status signal of a bedside guardrail acquired by a sensor assembly; preprocessing the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information; predicting the patient status of a patient corresponding to the smart bed using a pre-trained patient status recognition model, the first sensor information, and the guardrail status information; wherein the patient status is either an out-of-bed state or a non-out-of-bed state; if the patient status is an out-of-bed state, real-time statistics of the patient's out-of-bed time when the patient status is an out-of-bed state are collected, and detection is performed on whether the out-of-bed time reaches a target out-of-bed time threshold matched to the patient; wherein the target out-of-bed time threshold is generated based on the patient information;

[0143] The intelligent early warning module is used to output corresponding early warning information if it detects that the time to leave the bed reaches the target time threshold.

[0144] The specific principles and execution processes of each unit in the smart bed detection system disclosed in the above embodiment of the present application are the same as those of the smart bed detection method disclosed in the above embodiment of the present application. Please refer to the corresponding parts of the smart bed detection method disclosed in the above embodiment of the present application, and no further details will be given here.

[0145] The present application provides an intelligent bed detection system, which is applicable to an intelligent bed. The intelligent bed includes a bed and a bedside guardrail. A corresponding first sensor is installed on the bed, and an intelligent detector is installed on the bedside guardrail. The intelligent detector includes a sensor component. The system receives a first sensor signal collected by the first sensor and a guardrail status signal of the bedside guardrail collected by the sensor component; pre-processes the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information; predicts the patient status of a patient corresponding to the intelligent bed through a pre-trained patient status recognition model, the first sensor information and the guardrail status information; wherein the patient status is either an out-of-bed state or a non-out-of-bed state; if the patient status is an out-of-bed state, the out-of-bed time of the patient when the patient status is the out-of-bed state is counted in real time, and detects whether the out-of-bed time reaches a target out-of-bed time threshold matched to the patient; wherein the target out-of-bed time threshold is generated based on the patient information of the patient; if it is detected that the out-of-bed time reaches the target out-of-bed time threshold, outputs a corresponding warning message. The technical solution provided by the present application collects sensor signals related to the smart bed in real time through a first sensor, and collects guardrail status signals of the bedside guardrail in real time through a sensor component, so as to process these sensor signals accordingly to obtain corresponding sensor information, and analyze and identify the sensor information to obtain the patient status of the corresponding patient, so as to utilize the patient status of the patient and the target bed leaving event threshold risk monitoring determined according to the patient information of the patient, so as to trigger a corresponding early warning when it is determined that there is a risk. It can be seen that the present application focuses on the patient's condition and the usage status of the bedside guardrail of the smart bed, accurately grasps the patient's status, can ensure the patient's safety and improve the corresponding nursing quality, and reduce the probability of patient risks.

[0146] Optionally, the guardrail status signal includes a second sensor signal, a third sensor signal, and a fourth sensor signal. A data processing module is provided for preprocessing the first sensor signal and the guardrail status signal to obtain the first sensor information and the guardrail status information. The module is specifically configured to:

[0147] For each sensor signal, the noise characteristics of the sensor signal are analyzed to obtain the corresponding noise source and signal type, and the sensor signal information is filtered accordingly according to the noise source, signal type and sensor type corresponding to the sensor signal;

[0148] Amplifying the sensor signal after filtering to obtain corresponding sensor information;

[0149] Among them, the sensor signal is the first sensor signal, or the second sensor signal, or the third sensor signal, or the fourth sensor signal; the sensor information is the first sensor information, or the second sensor information, or the third sensor information, or the fourth sensor information; the second sensor information, the third sensor information and the fourth sensor information constitute the guardrail status information.

[0150] Optionally, a data processing module for predicting the patient status of a patient corresponding to the smart bed using a pre-trained patient status recognition model, first sensor information, and guardrail status information is specifically used to:

[0151] extracting a first time characteristic value, a first frequency characteristic value, and a patient pressure characteristic value from the first sensor information;

[0152] extracting a second time characteristic value, a second frequency characteristic value, and a motion characteristic value from the second sensor information;

[0153] extracting a third time characteristic value, a third frequency characteristic value, and a guardrail pressure characteristic value from the third sensor information;

[0154] extracting a fourth time characteristic value, a fourth frequency characteristic value, and a guardrail position change characteristic value from the fourth sensor information;

[0155] Each eigenvalue is input into a pre-trained patient state recognition model, so that the patient state recognition model performs recognition according to each eigenvalue to obtain the patient state of the patient corresponding to the smart bed; wherein the patient state recognition model is obtained by training the patient state recognition model to be trained using multiple historical sensor signals.

[0156] Optionally, a data processing module for obtaining a patient state recognition model by training the patient state recognition model to be trained using multiple historical sensor signals is specifically used for:

[0157] Acquiring historical sensor data of the smart bed, wherein the historical sensor data includes a plurality of historical sensor signals and corresponding target patient states; the plurality of historical sensor signals includes a historical first sensor signal, a historical second sensor signal, a historical third sensor signal, and a historical fourth sensor signal; and the target patient state includes an out-of-bed state and a non-out-of-bed state;

[0158] Preprocess each historical sensor signal to obtain corresponding historical sensor information;

[0159] Perform feature extraction on each historical sensor information to obtain multiple historical feature values ​​of each historical sensor information;

[0160] The various historical characteristic values ​​of the various historical sensor information and the target patient state are input into the patient state recognition model to be trained, so that the patient state recognition model to be trained performs recognition according to the various historical characteristic values ​​of the various historical sensor information, and the training goal is to make the recognized patient state approach the target patient state, and the parameters of the patient state recognition model to be trained are adjusted until the patient state recognition model to be trained reaches convergence, thereby obtaining a patient state recognition model.

[0161] Optionally, a data processing module is provided for generating a target bed-out time threshold value for the patient according to the patient information, specifically for:

[0162] Obtaining patient information of the patient, wherein the patient information includes the patient's basic information, condition information, and activity ability information;

[0163] Obtain a pre-built dynamic threshold library, and determine a target bed leaving time threshold that matches the patient from each bed leaving time in the dynamic threshold library based on basic information, condition information, and activity ability information.

[0164] Optionally, the smart bed detection system provided by the present application further includes a dual-sensor cross-validation module, which is specifically used to:

[0165] When any sensor is detected to be abnormal, continuous sensor signals sent by sensors that are not abnormal are received; wherein the sensor is the first sensor or any sensor in the sensor assembly;

[0166] Determining whether the continuous sensor signal of each sensor that has not experienced an abnormality satisfies a matching patient in bed condition or patient out of bed condition;

[0167] If the continuous sensor signals of all sensors without abnormality satisfy the matching patient in bed condition, determining that the patient status of the patient is in bed;

[0168] If the continuous sensor signals of all sensors without abnormalities meet the matching patient leaving bed condition, determining that the patient's patient status is the leaving bed state, and performing the step of real-time counting of the patient's leaving bed time when the patient status is the leaving bed state;

[0169] If any sensor that has not experienced an abnormality continuously sends a sensor signal that does not meet the matching patient leaving bed condition, or sends a sensor signal that does not meet the matching patient staying in bed condition, corresponding abnormality information is displayed.

[0170] Each embodiment in this specification is described in a progressive manner. The same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple. For relevant parts, refer to the partial description of the method embodiment. The system and system embodiments described above are merely schematic. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without making any creative efforts.

[0171] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0172] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

[0173] The above are only preferred embodiments of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A smart bed detection method, characterized in that: Applicable to a smart bed, the smart bed includes a bed and a bedside guardrail, the bed is installed with a corresponding first sensor, the bedside guardrail is installed with a smart detector, the smart detector includes a sensor component, the method includes: receiving a first sensor signal acquired by the first sensor and a guardrail status signal of the bedside guardrail acquired by the sensor assembly; Preprocessing the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information; Predicting the patient state of the patient corresponding to the smart bed by using a pre-trained patient state recognition model, first sensor information, and guardrail state information; wherein the patient state is a state of leaving the bed or a state of not leaving the bed; If the patient is in the out-of-bed state, real-time statistics are collected on the patient's out-of-bed time, and a detection is made as to whether the out-of-bed time reaches a target out-of-bed time threshold matched to the patient; wherein the target out-of-bed time threshold is generated based on the patient information of the patient; If it is detected that the time to leave the bed reaches the target time to leave the bed threshold, a corresponding warning message is output.

2. The method according to claim 1, characterized in that The guardrail status signal includes a second sensor signal, a third sensor signal and a fourth sensor signal; The preprocessing of the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information includes: For each sensor signal, performing a noise characteristic analysis on the sensor signal to obtain a corresponding noise source and signal type, and performing a corresponding filtering operation on the sensor signal according to the noise source, signal type, and the sensor type corresponding to the sensor signal; Amplifying the sensor signal after filtering to obtain corresponding sensor information; Among them, the sensor signal is the first sensor signal, or the second sensor signal, or the third sensor signal, or the fourth sensor signal; the sensor information is the first sensor information, or the second sensor information, or the third sensor information, or the fourth sensor information; the second sensor information, the third sensor information and the fourth sensor information constitute the guardrail status information.

3. The method according to claim 2, characterized in that The predicting the patient state of the patient corresponding to the smart bed by using the pre-trained patient state recognition model, the first sensor information, and the guardrail state information includes: extracting a first time characteristic value, a first frequency characteristic value, and a patient pressure characteristic value from the first sensor information; extracting a second time characteristic value, a second frequency characteristic value, and a motion characteristic value from the second sensor information; Extracting a third time characteristic value, a third frequency characteristic value, and a guardrail pressure characteristic value from the third sensor information; Extracting a fourth time characteristic value, a fourth frequency characteristic value, and a guardrail position change characteristic value from the fourth sensor information; Each characteristic value is input into a pre-trained patient state recognition model, so that the patient state recognition model performs recognition according to each characteristic value to obtain the patient state of the patient corresponding to the smart bed; wherein the patient state recognition model is obtained by training the patient state recognition model to be trained using multiple historical sensor signals.

4. The method according to claim 3, characterized in that The method of training a patient state recognition model to be trained using a plurality of historical sensor signals to obtain a patient state recognition model includes: Acquiring historical sensor data of the smart bed, wherein the historical sensor data includes a plurality of historical sensor signals and corresponding target patient states; the plurality of historical sensor signals includes a historical first sensor signal, a historical second sensor signal, a historical third sensor signal, and a historical fourth sensor signal; and the target patient state includes an out-of-bed state and a non-out-of-bed state; Preprocessing each of the historical sensor signals to obtain corresponding historical sensor information; Performing feature extraction on each piece of the historical sensor information to obtain a plurality of historical feature values ​​of each piece of the historical sensor information; The historical characteristic values ​​of each of the historical sensor information and the target patient state are input into the patient state recognition model to be trained, so that the patient state recognition model to be trained performs recognition according to the historical characteristic values ​​of each of the historical sensor information, and takes the recognized patient state approaching the target patient state as the training goal, and adjusts the parameters of the patient state recognition model to be trained until the patient state recognition model to be trained reaches convergence, thereby obtaining a patient state recognition model.

5. The method according to claim 1, wherein Generating the target bed leave time threshold for the patient according to the patient information of the patient includes: Acquiring patient information of the patient, wherein the patient information includes basic information, condition information, and activity ability information of the patient; A pre-built dynamic threshold library is obtained, and a target bed leaving time threshold matching the patient is determined from each bed leaving time in the dynamic threshold library according to the basic information, the condition information, and the activity ability information.

6. The method according to claim 1, wherein The method further comprises: When any sensor is detected to be abnormal, continuous sensor signals sent by sensors that are not abnormal are received; wherein the sensor is the first sensor or any sensor in the sensor assembly; Determining whether the continuous sensor signal of each sensor that has not experienced an abnormality satisfies a matching patient in bed condition or patient out of bed condition; If the continuous sensor signals of all sensors without abnormalities satisfy the matching patient in bed condition, determining that the patient status of the patient is in bed; If the continuous sensor signals of all sensors without abnormalities meet the matching patient leaving bed condition, determining that the patient's patient status is the leaving bed state, and performing the step of real-time counting of the patient's leaving bed time when the patient's patient status is the leaving bed state; If any sensor that has not experienced an abnormality continuously sends a sensor signal that does not meet the matching patient leaving bed condition, or sends a sensor signal that does not meet the matching patient staying in bed condition, corresponding abnormality information is displayed.

7. A smart bed detection system, characterized in that: Applicable to a smart bed, the smart bed includes a bed and a bedside guardrail, the bed is installed with a corresponding first sensor, the bedside guardrail is installed with a smart detector, the smart detector includes a sensor component, and the system includes: a data processing module for receiving a first sensor signal acquired by the first sensor and a guardrail status signal of the bedside guardrail acquired by the sensor assembly; preprocessing the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information; predicting the patient status of the patient corresponding to the smart bed through a pre-trained patient status recognition model, the first sensor information, and the guardrail status information; wherein the patient status is either an out-of-bed state or a non-out-of-bed state; if the patient status is an out-of-bed state, real-time statistics are collected on the out-of-bed time when the patient status is an out-of-bed state, and detection is performed on whether the out-of-bed time reaches a target out-of-bed time threshold matched to the patient; wherein the target out-of-bed time threshold is generated based on the patient information of the patient; The intelligent early warning module is used to output corresponding early warning information if it is detected that the time to leave the bed reaches the target time to leave the bed threshold.

8. The system according to claim 7, characterized in that The guardrail status signal includes a second sensor signal, a third sensor signal, and a fourth sensor signal. A data processing module is used to pre-process the first sensor signal and the guardrail status signal to obtain first sensor information and guardrail status information. For each sensor signal, performing a noise characteristic analysis on the sensor signal to obtain a corresponding noise source and signal type, and performing a corresponding filtering operation on the sensor signal information according to the noise source, signal type, and the sensor type corresponding to the sensor signal; Amplifying the sensor signal after filtering to obtain corresponding sensor information; Among them, the sensor signal is the first sensor signal, or the second sensor signal, or the third sensor signal, or the fourth sensor signal; the sensor information is the first sensor information, or the second sensor information, or the third sensor information, or the fourth sensor information; the second sensor information, the third sensor information and the sensor information constitute the guardrail status information.

9. The system according to claim 8, characterized in that The data processing module for predicting the patient status of the patient corresponding to the smart bed by using the pre-trained patient status recognition model, the first sensor information, and the guardrail status information is specifically used to: extracting a first time characteristic value, a first frequency characteristic value, and a patient pressure characteristic value from the first sensor information; extracting a second time characteristic value, a second frequency characteristic value, and a motion characteristic value from the second sensor information; Extracting a third time characteristic value, a third frequency characteristic value, and a guardrail pressure characteristic value from the third sensor information; Extracting a fourth time characteristic value, a fourth frequency characteristic value, and a guardrail position change characteristic value from the fourth sensor information; Each characteristic value is input into a pre-trained patient state recognition model, so that the patient state recognition model performs recognition according to each characteristic value to obtain the patient state of the patient corresponding to the smart bed; wherein the patient state recognition model is obtained by training the patient state recognition model to be trained using multiple historical sensor signals.

10. The system according to claim 9, characterized in that The data processing module for training a patient state recognition model to be trained using a plurality of historical sensor signals to obtain a patient state recognition model is specifically used to: Acquiring historical sensor data of the smart bed, wherein the historical sensor data includes a plurality of historical sensor signals and corresponding target patient states; the plurality of historical sensor signals includes a historical first sensor signal, a historical second sensor signal, a historical third sensor signal, and a historical fourth sensor signal; and the target patient state includes an out-of-bed state and a non-out-of-bed state; Preprocessing each of the historical sensor signals to obtain corresponding historical sensor information; Performing feature extraction on each piece of the historical sensor information to obtain a plurality of historical feature values ​​of each piece of the historical sensor information; The historical characteristic values ​​of each of the historical sensor information and the target patient state are input into the patient state recognition model to be trained, so that the patient state recognition model to be trained performs recognition according to the historical characteristic values ​​of each of the historical sensor information, and takes the recognized patient state approaching the target patient state as the training goal, and adjusts the parameters of the patient state recognition model to be trained until the patient state recognition model to be trained reaches convergence, thereby obtaining a patient state recognition model.

Images