Nursing monitoring system and method based on anti-bedsore intelligent mattress

Through deep learning technology, multi-scale analysis of the body movement status data of smart mattresses with anti-bedroom sores intelligently judges the turnover frequency, solving the problem that the existing technology cannot be analyzed in depth and integrated, realizing timely monitoring and early warning of turnover frequency, reducing the need for reduction risk of complications of bedsores.

CN118806270BActive Publication Date: 2024-11-22JILIN JIMING BIOTECHNOLOGY CO LTD
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411297705.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-18
Publication Date
2024-11-22
Estimated Expiration
2044-09-18

AI Technical Summary

Technical Problem

The existing anti-bedsore smart mattresses cannot deeply analyze the patient's physical movement status data, cannot promptly detect insufficient turnover frequency, and cannot integrate with other medical systems, which limits its application in clinical environments.

Method used

Using artificial intelligence technology based on deep learning, the features of body dynamic state are extracted through multi-scale semantic coding modules, and the feature guide fusion module is used to generate multi-scale turnover semantic coding feature vectors. Combined with the nursing monitoring results to generate the module to intelligently determine whether the turnover frequency is insufficient. Automatically remind nursing staff to intervene.

Benefits of technology

It realizes in-depth analysis of the patient's physical movement status, intelligently judges the frequency of turnover, and automatically reminds nursing staff to conduct timely intervention to prevent the occurrence of bedsores and reduces the risk of complications.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118806270B_ABST
    Figure CN118806270B_ABST
Patent Text Reader

Abstract

The present application discloses a nursing monitoring system and method based on an anti-bedsore smart mattress, which relates to the field of medical information processing technology. The anti-bedsore smart mattress is used to monitor the patient's body movement state, and the deep learning-based artificial intelligence technology is used to analyze and process the patient's body movement state data, thereby mining the patient's body movement state information such as being in bed, leaving bed, and turning over, so as to intelligently judge whether the patient's turning over frequency is insufficient. In this way, the nursing staff can be automatically reminded to carry out timely nursing intervention on the patient, prevent the occurrence of bedsores, and reduce the risk of complications caused by bedsores.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of medical information processing technology, and more specifically, to a nursing monitoring system and method based on an anti-bedsore intelligent mattress. Background Art

[0002] With the aging trend of the population, the care problem of bedridden patients has become increasingly prominent. Among them, bedsores (also known as pressure sores or pressure ulcers) are one of the most common complications of long-term bedridden patients, seriously affecting the quality of life and health of patients. Specifically, bedsores are mainly caused by the patient being in the same position for a long time, resulting in long-term pressure on local tissues, which hinders blood circulation and leads to tissue ischemia and necrosis. For long-term bedridden patients, regular turning is a key measure to prevent bedsores.

[0003] Traditional measures to prevent bedsores rely on nursing staff to observe and record the patient's body position and adjust the patient's position regularly. However, nursing staff cannot always observe the patient, so they may not be able to detect the patient's insufficient turning frequency in time. In addition, for critically ill patients or patients who stay in bed frequently, nursing staff need to spend a lot of time monitoring and recording, which is a heavy workload and easy to get tired.

[0004] In recent years, as an advanced medical device, anti-bedsore smart mattresses have been widely used in the field of medical care. Anti-bedsore smart mattresses can monitor the patient's body movement status in real time through sensors, and provide medical staff with the patient's body activity data, so that medical staff can take corresponding nursing measures in time according to the patient's body activity. However, some existing anti-bedsore smart mattresses can only provide simple body movement status data, lack in-depth analysis and understanding of the data, and cannot be integrated with other medical systems, which limits their application in clinical environments. Therefore, a nursing monitoring system and method based on an anti-bedsore smart mattress is expected. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed.

[0006] Accordingly, according to one aspect of the present application, a nursing monitoring system based on an anti-bedsore smart mattress is provided, comprising:

[0007] A patient body movement state monitoring module, used to obtain a time series of the body movement state of the monitored patient collected by the anti-bedsore intelligent mattress;

[0008] A multi-scale semantic coding module, used for extracting multi-scale semantic coding features of the time series of the body movement state to obtain a first-scale turning semantic coding feature vector and a second-scale turning semantic coding feature vector;

[0009] A feature-guided fusion module, used for performing feature-guided fusion on the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector to obtain a multi-scale turning-over semantic coding feature vector;

[0010] A nursing monitoring result generating module, used for determining the nursing monitoring result based on the multi-scale turning semantic coding feature vector;

[0011] Wherein, the feature-guided fusion module includes:

[0012] A Gaussian distribution random number generating unit, configured to generate Gaussian distribution function coefficients of the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector using a Gaussian distribution random number function to obtain a first Gaussian distribution function coefficient and a second Gaussian distribution function coefficient;

[0013] A feature fusion unit is used to perform guided fusion processing on the first-scale turn-over semantic coding feature vector and the second-scale turn-over semantic coding feature vector based on the first Gaussian distribution function coefficient and the second Gaussian distribution function coefficient to obtain the multi-scale turn-over semantic coding feature vector.

[0014] In the above-mentioned nursing monitoring system based on the anti-bedsore intelligent mattress, the multi-scale semantic encoding module includes: a one-hot encoding unit, which is used to perform one-hot encoding on each body movement state in the time series of the body movement state to obtain a time series of one-hot encoding representation vectors of the body movement state; and a multi-scale feature extraction unit, which is used to perform short-term dependency association feature extraction and long-term dependency association feature extraction on the time series of the one-hot encoding representation vectors of the body movement state to obtain the first-scale turning over semantic encoding feature vector and the second-scale turning over semantic encoding feature vector.

[0015] In the above-mentioned nursing monitoring system based on the anti-bedsore intelligent mattress, the multi-scale feature extraction unit is used to: pass the time series of the one-hot encoding representation vector of the body movement state through a turning over semantic dual-stream feature extractor including an RNN model and a Bi-LSTM model to obtain the first-scale turning over semantic encoding feature vector and the second-scale turning over semantic encoding feature vector.

[0016] In the above-mentioned nursing monitoring system based on the anti-bedsore intelligent mattress, the feature-guided fusion module is used to: use the feature-guided fusion module based on prior distribution to process the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector to obtain the multi-scale turning semantic coding feature vector.

[0017] In the above-mentioned nursing monitoring system based on the anti-bedsore intelligent mattress, the feature fusion unit is used to: determine the weight coefficients of the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector to obtain the first weight coefficient and the second weight coefficient; use the first weight coefficient to weight the first-scale turning semantic coding feature vector to obtain the weighted first-scale turning semantic coding feature vector, and use the second weight coefficient to weight the second-scale turning semantic coding feature vector to obtain the weighted second-scale turning semantic coding feature vector; respectively calculate the exponential function values ​​of each eigenvalue of the weighted first-scale turning semantic coding feature vector and the weighted second-scale turning semantic coding feature vector with the natural constant as the base to obtain the exponential first-scale turning semantic coding feature vector and the exponential second-scale turning semantic coding feature vector; add and sum the corresponding elements of the product of the first Gaussian distribution function coefficient multiplied by the exponential first-scale turning semantic coding feature vector and the product of the second Gaussian distribution function coefficient multiplied by the exponential second-scale turning semantic coding feature vector to obtain the multi-scale turning semantic coding feature vector.

[0018] In the above-mentioned nursing monitoring system based on the anti-bedsore intelligent mattress, the nursing monitoring result generation module includes: a classification unit, which is used to pass the multi-scale turning over semantic encoding feature vector through a classifier-based nursing monitoring result generator to obtain the nursing monitoring result, and the nursing monitoring result is used to indicate whether the turning over frequency is insufficient.

[0019] In the above-mentioned nursing monitoring system based on the anti-bedsore intelligent mattress, the classification unit is used to: use the fully connected layer of the nursing monitoring result generator to fully connect the multi-scale turning semantic coding feature vector to obtain a fully connected coding feature vector; input the fully connected coding feature vector into the Softmax classification function of the nursing monitoring result generator to obtain the probability value of the multi-scale turning semantic coding feature vector belonging to each classification label, wherein the classification labels include insufficient turning frequency and normal turning frequency; and determine the classification label corresponding to the largest one among the probability values ​​as the nursing monitoring result.

[0020] According to another aspect of the present application, a nursing monitoring method based on an anti-bedsore smart mattress is provided, comprising:

[0021] Acquire a time series of body motion status of a monitored patient object collected by the anti-bedsore smart mattress;

[0022] Extracting multi-scale semantic coding features of the time series of the body movement state to obtain a first-scale turning-over semantic coding feature vector and a second-scale turning-over semantic coding feature vector;

[0023] Performing feature-guided fusion on the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector to obtain a multi-scale turning-over semantic coding feature vector;

[0024] Determining nursing monitoring results based on the multi-scale turning semantic coding feature vector;

[0025] The step of performing feature-guided fusion on the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector to obtain a multi-scale turning-over semantic coding feature vector includes:

[0026] Generate Gaussian distribution function coefficients of the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector using a Gaussian distribution random number function to obtain a first Gaussian distribution function coefficient and a second Gaussian distribution function coefficient;

[0027] Based on the first Gaussian distribution function coefficient and the second Gaussian distribution function coefficient, the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector are subjected to guided fusion processing to obtain the multi-scale turning semantic coding feature vector.

[0028] Compared with the prior art, the nursing monitoring system and method based on the anti-bedsore smart mattress provided by the present application uses the anti-bedsore smart mattress to monitor the patient's body movement state, and uses artificial intelligence technology based on deep learning to analyze and process the patient's body movement state data, thereby mining the patient's body movement state information such as being in bed, leaving bed, and turning over, so as to intelligently judge whether the patient's turning over frequency is insufficient. In this way, the nursing staff can be automatically reminded to carry out timely nursing intervention on the patient, prevent the occurrence of bedsores, and reduce the risk of complications caused by bedsores. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0030] Figure 1 It is a block diagram of a nursing monitoring system based on an anti-bedsore smart mattress according to an embodiment of the present application.

[0031] Figure 2 Schematic diagram of the architecture of a nursing monitoring system based on an anti-bedsore smart mattress according to an embodiment of the present application.

[0032] Figure 3It is a block diagram of a multi-scale semantic coding module in a nursing monitoring system based on an anti-bedsore smart mattress according to an embodiment of the present application.

[0033] Figure 4 It is a block diagram of a feature-guided fusion module in a nursing monitoring system based on an anti-bedsore smart mattress according to an embodiment of the present application.

[0034] Figure 5 The present invention is a flowchart of a nursing monitoring method based on an anti-bedsore smart mattress according to an embodiment of the present application. DETAILED DESCRIPTION

[0035] Below, the embodiments of the present application will be described in more detail in conjunction with the accompanying drawings, and the above and other purposes, features and advantages of the present application will become more apparent. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein.

[0036] As mentioned in the above background technology, the anti-bedsore smart mattress, as an advanced medical device, can monitor the patient's body movement state in real time and provide medical staff with the patient's physical activity data, so that medical staff can take corresponding nursing measures in time according to the patient's behavior. However, some existing anti-bedsore smart mattresses can only provide basic body movement data, lack in-depth analysis and understanding of the data, and cannot be integrated with other medical systems, which to a certain extent restricts its application potential in clinical environments. In response to the above technical problems, the technical concept of this application is to use an anti-bedsore smart mattress to monitor the patient's body movement state, and use artificial intelligence technology based on deep learning to analyze and process the patient's body movement state data, from which the patient's body movement state information such as being in bed, leaving bed and turning over can be mined, so as to intelligently judge whether the patient's turning over frequency is insufficient. In this way, nursing staff can be automatically reminded to carry out timely nursing intervention on patients, prevent the occurrence of bedsores, and reduce the risk of complications caused by bedsores.

[0037] Figure 1 It is a block diagram of a nursing monitoring system based on an anti-bedsore smart mattress according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the architecture of a nursing monitoring system based on an anti-bedsore smart mattress according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to an embodiment of the present application, a nursing monitoring system 100 based on an anti-bedsore smart mattress includes: a patient object body movement state monitoring module 110, which is used to obtain a time series of body movement states of a monitored patient object collected by the anti-bedsore smart mattress; a multi-scale semantic coding module 120, which is used to extract multi-scale semantic coding features of the time series of the body movement state to obtain a first-scale turning semantic coding feature vector and a second-scale turning semantic coding feature vector; a feature-guided fusion module 130, which is used to perform feature-guided fusion on the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector to obtain a multi-scale turning semantic coding feature vector; and a nursing monitoring result generation module 140, which is used to determine a nursing monitoring result based on the multi-scale turning semantic coding feature vector.

[0038] In the above-mentioned nursing monitoring system 100 based on the anti-bedsore smart mattress, the patient object body movement state monitoring module 110 is used to obtain the time series of the body movement state of the monitored patient object collected by the anti-bedsore smart mattress. It should be understood that the anti-bedsore smart mattress can collect the patient's body movement state data in real time through the built-in sensor, including turning over, sitting up, slight activity, stillness, leaving the bed, etc., reflecting the patient's activity in bed. In the technical solution of the present application, in order to effectively warn the patient of insufficient turning frequency, the patient's body movement state is analyzed and processed in the time dimension through artificial intelligence technology to reveal the patient's body movement state time series pattern. Then, abnormal patient body movement patterns are identified, such as low turning frequency, so as to issue a warning in time, so that the nursing staff can take intervention measures in time to adjust the patient's body position, which helps to prevent complications such as bedsores caused by long-term bed rest.

[0039] In the nursing monitoring system 100 based on the anti-bedsore smart mattress, the multi-scale semantic coding module 120 is used to extract the multi-scale semantic coding features of the time series of the body movement state to obtain the first-scale turning over semantic coding feature vector and the second-scale turning over semantic coding feature vector. Figure 3 FIG. 1 is a block diagram of a multi-scale semantic encoding module in a nursing monitoring system based on an anti-bedsore smart mattress according to an embodiment of the present application. Figure 3 As shown, the multi-scale semantic coding module 120 includes: a one-hot coding unit 121, which is used to perform one-hot coding on each body movement state in the time series of the body movement state to obtain a time series of one-hot coding representation vectors of the body movement state; a multi-scale feature extraction unit 122, which is used to perform short-term dependency association feature extraction and long-term dependency association feature extraction on the time series of the one-hot coding representation vectors of the body movement state to obtain the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector.

[0040] Specifically, the one-hot encoding unit 121 is used to perform one-hot encoding on each body movement state in the time series of the body movement state to obtain a time series of one-hot encoding representation vectors of the body movement state. It should be understood that since the patient's body movement state data is a discrete, non-numerical data form (such as turning over, sitting up, getting out of bed, etc.), in order to convert it into a numerical form that can be processed by a machine learning algorithm, in the technical solution of the present application, one-hot encoding is used to encode each body movement state in the time series of the body movement state. Specifically, one-hot encoding is a method of converting a discrete categorical variable into a binary numerical form. Through one-hot encoding, each body movement state can be converted into a binary numerical vector of a fixed length, and only one element in the binary numerical vector is 1, and the rest of the elements are 0, wherein the position of the element that is 1 corresponds to the position of the body movement state in the coding table. In this way, each body movement state is mapped to a unique numerical vector representation, so as to facilitate subsequent mathematical operations and processing of machine learning models.

[0041] Specifically, the multi-scale feature extraction unit 122 is used to perform short-term dependency association feature extraction and long-term dependency association feature extraction on the time series of the one-hot encoding representation vector of the body movement state to obtain the first-scale turning semantic encoding feature vector and the second-scale turning semantic encoding feature vector. It should be understood that since the patient's body movement state data has continuity and time series in the time dimension, in order to capture the time series information of the patient's body movement state, it is necessary to further perform time series association feature encoding on the time series of the one-hot encoding representation vector of the body movement state. At the same time, considering that the patient's body movement state may present different characteristic patterns on different time scales, for example, it may appear as frequent micro-activities of the limbs on a shorter time scale, and may present periodic turning behavior or leaving the bed on a longer time scale. Therefore, in the technical solution of the present application, the time series of the one-hot encoding representation vector of the body movement state is further encoded with long-term and short-term dependency association.

[0042] In a specific example of the present application, the encoding method for extracting short-term dependency association features and long-term dependency association features from the time series of the one-hot encoding representation vector of the body movement state is to pass the time series of the one-hot encoding representation vector of the body movement state through a turning semantic dual-stream feature extractor including an RNN model and a Bi-LSTM model to obtain the first-scale turning semantic encoding feature vector and the second-scale turning semantic encoding feature vector. Specifically, the RNN model enables the network to memorize and process the input information of the previous moment through its internal cyclic connection structure, so that the temporal dependency relationship in the time series of the one-hot encoding representation vector of the body movement state can be effectively captured. However, the RNN model is limited by the problems of gradient disappearance and gradient explosion, and it is difficult to learn and maintain the long-term dependency association in the patient's body movement state time series data. The Bi-LSTM model introduces a bidirectional information flow mechanism based on the RNN, combines the forward and reverse LSTM (long short-term memory network) layers, and can simultaneously consider the information in the two directions of the time series of the one-hot encoding representation vector of the body movement state, thereby capturing the long-term dependency association relationship in the patient's body movement state time series data. In other words, by comprehensively utilizing the RNN model and the Bi-LSTM model, the complex characteristic patterns of the patient's body movement state can be captured at different time scales to fully express the patient's turning behavior characteristics, providing a more comprehensive and accurate basis for the subsequent patient turning frequency analysis.

[0043] In the above-mentioned nursing monitoring system 100 based on the anti-bedsore intelligent mattress, the feature-guided fusion module 130 is used to perform feature-guided fusion on the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector to obtain a multi-scale turning-over semantic coding feature vector. It should be understood that the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector respectively reflect the patient's turning-over behavior pattern characteristics from different time scales. However, simply concatenating or adding the two together may not fully utilize the complementarity between the two. Based on this, in the technical solution of the present application, the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector are further feature-guided fused.

[0044] In a specific example of the present application, the encoding method for feature-guided fusion of the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector is to use a feature-guided fusion module based on prior distribution to process the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector to obtain the multi-scale turning semantic coding feature vector. Specifically, the feature-guided fusion module based on prior distribution introduces Gaussian prior distribution parameters in the fusion process to simulate the potential statistical relationship between turning semantic coding features of different scales, so that the fusion process can take into account the relative importance and complementarity between features of different scales, so as to achieve effective fusion of turning semantic coding features of different scales, thereby enhancing the expressive power of the semantic features of the patient's turning behavior. At the same time, due to the introduction of random Gaussian prior distribution parameters, the overfitting problem can be alleviated to a certain extent, and the generalization ability of the model can be improved.

[0045] Figure 4 FIG. 1 is a block diagram of a feature-guided fusion module in a nursing monitoring system based on an anti-bedsore smart mattress according to an embodiment of the present application. Figure 4 As shown, the feature guided fusion module 130 includes: a Gaussian distribution random number generation unit 131, which is used to use a Gaussian distribution random number function to generate the Gaussian distribution function coefficients of the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector to obtain the first Gaussian distribution function coefficient and the second Gaussian distribution function coefficient; a feature fusion unit 132, which is used to guide and fuse the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector based on the first Gaussian distribution function coefficient and the second Gaussian distribution function coefficient to obtain the multi-scale turning semantic coding feature vector.

[0046] More specifically, the feature fusion unit 132 is used to: determine the weight coefficients of the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector to obtain a first weight coefficient and a second weight coefficient; use the first weight coefficient to weight the first-scale turning semantic coding feature vector to obtain a weighted first-scale turning semantic coding feature vector, and use the second weight coefficient to weight the second-scale turning semantic coding feature vector to obtain a weighted second-scale turning semantic coding feature vector; respectively calculate the exponential function values ​​of each eigenvalue of the weighted first-scale turning semantic coding feature vector and the weighted second-scale turning semantic coding feature vector with a natural constant as the base to obtain an exponential first-scale turning semantic coding feature vector and an exponential second-scale turning semantic coding feature vector; add and sum the corresponding elements of the product of the first Gaussian distribution function coefficient multiplied by the exponential first-scale turning semantic coding feature vector and the product of the second Gaussian distribution function coefficient multiplied by the exponential second-scale turning semantic coding feature vector to obtain the multi-scale turning semantic coding feature vector.

[0047] That is, the feature-guided fusion module 130 is used to: fuse the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector using the following feature-guided fusion formula to obtain the multi-scale turning semantic coding feature vector, wherein the feature-guided fusion formula is:

[0048] ;

[0049] in, is the first scale turn-over semantic encoding feature vector, is the second scale turn-over semantic encoding feature vector, is the multi-scale turning semantic encoding feature vector, and denote the first weight coefficient and the second weight coefficient respectively, and They respectively represent the first Gaussian distribution function coefficient and the second Gaussian distribution function coefficient generated by the Gaussian distribution random number function.

[0050] In the above-mentioned nursing monitoring system 100 based on the anti-bedsore intelligent mattress, the nursing monitoring result generation module 140 is used to determine the nursing monitoring result based on the multi-scale turning semantic coding feature vector. The nursing monitoring result generation module 140 includes: a classification unit, which is used to pass the multi-scale turning semantic coding feature vector through a nursing monitoring result generator based on a classifier to obtain the nursing monitoring result, and the nursing monitoring result is used to indicate whether the turning frequency is insufficient.

[0051] Specifically, the classification unit is used to pass the multi-scale turning semantic coding feature vector through a classifier-based nursing monitoring result generator to obtain the nursing monitoring result, and the nursing monitoring result is used to indicate whether the turning frequency is insufficient. It should be understood that the nursing monitoring result generator adopts a classification algorithm to perform feature learning on the multi-scale turning semantic coding feature vector to understand the patient's body movement pattern in the multi-scale turning semantic coding feature vector, and performs feature classification mapping on it, thereby outputting a two-class nursing monitoring result, namely "insufficient turning frequency" or "normal turning frequency". Furthermore, an automatic alarm can be issued according to the nursing monitoring result to achieve real-time monitoring and low-frequency early warning of the patient's turning frequency, so that nursing staff can pay attention to the patient's improper body position in time and take appropriate intervention measures to prevent bedsores caused by long-term bed rest, and further improve the patient's nursing effect and safety.

[0052] In a specific example of the present application, the classification unit is used to: use the fully connected layer of the nursing monitoring result generator to fully connect the multi-scale turning semantic encoding feature vector to obtain a fully connected encoding feature vector; input the fully connected encoding feature vector into the Softmax classification function of the nursing monitoring result generator to obtain the probability value of the multi-scale turning semantic encoding feature vector belonging to each classification label, wherein the classification labels include insufficient turning frequency and normal turning frequency; determine the classification label corresponding to the largest one of the probability values ​​as the nursing monitoring result.

[0053] Here, considering that the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector, which are representations of different-scale time series, are subjected to feature-guided fusion based on prior distribution, the difference in prior distributions of different-scale time series will guide the fusion to obtain a multi-scale turning semantic coding feature vector with a diversified set expression distribution of time series multi-scale aggregation features. Therefore, it is expected to improve the balance between the accuracy and completeness of the regression mapping when the multi-scale turning semantic coding feature vector is input into a classifier-based nursing monitoring result generator for class regression, thereby improving the accuracy of the obtained nursing monitoring results.

[0054] In a preferred example, the multi-scale turning semantic encoding feature vector is input into a classifier-based nursing monitoring result generator to obtain a nursing monitoring result, including:

[0055] Determine the maximum eigenvalue of the multi-scale turning semantic coding and the minimum eigenvalue of the multi-scale turning semantic coding feature vector, and calculate the multi-scale turning semantic coding mean and the multi-scale turning semantic coding standard deviation of the feature set of the multi-scale turning semantic coding feature vector;

[0056] Calculating the quotient of the multi-scale turning semantic coding mean and the multi-scale turning semantic coding standard deviation to obtain a multi-scale turning semantic coding statistical normalization value;

[0057] Calculate the inverse of each eigenvalue of the multi-scale turning semantics coding feature vector, multiply it by the difference between the maximum eigenvalue of the multi-scale turning semantics coding and the minimum eigenvalue of the multi-scale turning semantics coding, and then perform point subtraction from the statistical normalization value of the multi-scale turning semantics coding to obtain a multi-scale turning semantics coding distribution approximation vector;

[0058] Calculate an exponential function with a natural constant as the base and each eigenvalue of the multi-scale turning over semantic coding distribution approximation vector as the exponent to obtain a multi-scale turning over semantic coding distribution class approximation vector;

[0059] Performing dot-wise addition of the multi-scale turning semantic coding distribution class approximation vector and the multi-scale turning semantic coding statistical normalization value, and calculating the base-2 logarithm of the absolute value of each eigenvalue of the dot-wise addition vector to obtain an optimized multi-scale turning semantic coding feature vector;

[0060] The optimized multi-scale turning semantic encoding feature vector is input into a classifier-based nursing monitoring result generator to obtain a nursing monitoring result.

[0061] The optimized representation of the multi-scale turning semantic coding feature vector is:

[0062] ;

[0063] ;

[0064] ;

[0065] ;

[0066] in, represents the maximum eigenvalue of the multi-scale turning semantic encoding, Indicates the minimum eigenvalue of multi-scale turning semantic encoding, represents the mean value of multi-scale turning semantic encoding, represents the standard deviation of multi-scale turning semantic encoding, is the statistical normalization value of the multi-scale turning semantic encoding, and is the difference between the maximum eigenvalue of the multi-scale turning semantics encoding and the minimum eigenvalue of the multi-scale turning semantics encoding, represents the multi-scale turning semantic coding feature vector, It means to calculate the inverse of each eigenvalue of the vector. It means point multiplication by position. Add by position, Subtract by position, represents the approximate vector of the multi-scale semantic coding distribution class, Represents the optimized multi-scale turn-over semantic encoding feature vector.

[0067] Therefore, by performing random statistical normalization on the diversified feature set of the multi-scale turning semantic coding feature vector, the confidence space constructed based on the overall eigenvalue of the multi-scale turning semantic coding feature vector is approximated by the standardized continuous probability density distribution relative to the response hypothesis test of each eigenvalue of the multi-scale turning semantic coding feature vector, thereby establishing the target reachability of the diversified feature distribution of the multi-scale turning semantic coding feature vector to a unified regression target, so as to achieve a balanced executableness between mapping accuracy and mapping completeness in the class regression process based on the diversified feature distribution of the multi-scale turning semantic coding feature vector, and improve the accuracy of the nursing monitoring results obtained by inputting the multi-scale turning semantic coding feature vector into the classifier-based nursing monitoring result generator.

[0068] In summary, the nursing monitoring system based on the anti-bedsore smart mattress according to the embodiment of the present application is explained, which uses the anti-bedsore smart mattress to monitor the patient's body movement state, and uses artificial intelligence technology based on deep learning to analyze and process the patient's body movement state data, thereby mining the patient's body movement state information such as being in bed, leaving bed, and turning over, so as to intelligently judge whether the patient's turning over frequency is insufficient. In this way, the nursing staff can be automatically reminded to carry out timely nursing intervention on the patient, prevent the occurrence of bedsores, and reduce the risk of complications caused by bedsores.

[0069] Figure 5 FIG. 1 is a flow chart of a nursing monitoring method based on an anti-bedsore smart mattress according to an embodiment of the present application. Figure 5 As shown, according to the embodiment of the present application, the nursing monitoring method based on the anti-bedsore smart mattress includes the following steps: S1, acquiring the time series of the body movement state of the monitored patient object collected by the anti-bedsore smart mattress; S2, extracting the multi-scale semantic coding features of the time series of the body movement state to obtain a first-scale turning semantic coding feature vector and a second-scale turning semantic coding feature vector; S3, performing feature-guided fusion on the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector to obtain a multi-scale turning semantic coding feature vector; S4, determining the nursing monitoring result based on the multi-scale turning semantic coding feature vector.

[0070] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned nursing monitoring method based on the anti-bedsore intelligent mattress have been referred to above. Figures 1 to 4The description of the nursing monitoring system based on the anti-bedsore intelligent mattress has been introduced in detail, and therefore, its repeated description will be omitted.

[0071] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0072] In the above embodiments, the description of each embodiment has its own emphasis. For the parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiment described above is only schematic. For example, the unit division is only a logical function division, and there may be other division methods in actual implementation. The unit described as a separate component may or may not be physically separated, and the component displayed as a unit may or may not be a physical sub-unit, that is, it may be located in one place, or it may be distributed on multiple network sub-units. Some or all of the units can be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0073] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0074] In addition, it is obvious that the word "comprising" does not exclude other sub-units or steps, and the singular does not exclude the plural. Multiple sub-units stated in the system claim can also be implemented by one sub-unit through software or hardware.

[0075] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A nursing monitoring system based on an anti-bedsore intelligent mattress, characterized in that: include: A patient body movement state monitoring module, used to obtain a time series of the body movement state of the monitored patient collected by the anti-bedsore intelligent mattress; A multi-scale semantic coding module, used for extracting multi-scale semantic coding features of the time series of the body movement state to obtain a first-scale turning semantic coding feature vector and a second-scale turning semantic coding feature vector; A feature-guided fusion module, used for performing feature-guided fusion on the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector to obtain a multi-scale turning-over semantic coding feature vector; A nursing monitoring result generating module, used for determining the nursing monitoring result based on the multi-scale turning semantic coding feature vector; Wherein, the feature-guided fusion module includes: A Gaussian distribution random number generating unit, configured to generate Gaussian distribution function coefficients of the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector using a Gaussian distribution random number function to obtain a first Gaussian distribution function coefficient and a second Gaussian distribution function coefficient; A feature fusion unit is used to perform guided fusion processing on the first-scale turn-over semantic coding feature vector and the second-scale turn-over semantic coding feature vector based on the first Gaussian distribution function coefficient and the second Gaussian distribution function coefficient to obtain the multi-scale turn-over semantic coding feature vector.

2. The nursing monitoring system based on the anti-bedsore intelligent mattress according to claim 1 is characterized in that: The multi-scale semantic encoding module comprises: A one-hot encoding unit, used for performing one-hot encoding on each body motion state in the time series of the body motion state to obtain a time series of one-hot encoding representation vectors of the body motion state; The multi-scale feature extraction unit is used to perform short-term dependency association feature extraction and long-term dependency association feature extraction on the time series of the one-hot encoding representation vector of the body movement state to obtain the first-scale turning semantic encoding feature vector and the second-scale turning semantic encoding feature vector.

3. The nursing monitoring system based on the anti-bedsore intelligent mattress according to claim 2 is characterized in that: The multi-scale feature extraction unit is used to: The time series of the one-hot encoding representation vector of the body movement state is passed through a turning semantic dual-stream feature extractor including an RNN model and a Bi-LSTM model to obtain the first-scale turning semantic encoding feature vector and the second-scale turning semantic encoding feature vector.

4. The nursing monitoring system based on the anti-bedsore intelligent mattress according to claim 3 is characterized in that: The feature-guided fusion module is used to: The first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector are processed using a feature-guided fusion module based on prior distribution to obtain the multi-scale turning semantic coding feature vector.

5. The nursing monitoring system based on the anti-bedsore intelligent mattress according to claim 4 is characterized in that: The feature fusion unit is used to: Determining weight coefficients of the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector to obtain a first weight coefficient and a second weight coefficient; Using the first weight coefficient to weight the first-scale turning semantic coding feature vector to obtain a weighted first-scale turning semantic coding feature vector, and using the second weight coefficient to weight the second-scale turning semantic coding feature vector to obtain a weighted second-scale turning semantic coding feature vector; Respectively calculating the exponential function values ​​of each eigenvalue of the weighted first-scale turning semantic coding feature vector and the weighted second-scale turning semantic coding feature vector with the natural constant as the base to obtain the exponential first-scale turning semantic coding feature vector and the exponential second-scale turning semantic coding feature vector; The product of the first Gaussian distribution function coefficient multiplied by the exponential first-scale turning semantic coding feature vector and the product of the second Gaussian distribution function coefficient multiplied by the exponential second-scale turning semantic coding feature vector are added and summed up to obtain the multi-scale turning semantic coding feature vector.

6. The nursing monitoring system based on the anti-bedsore intelligent mattress according to claim 5 is characterized in that: The nursing monitoring result generating module comprises: The classification unit is used to pass the multi-scale turning semantic encoding feature vector through a classifier-based nursing monitoring result generator to obtain the nursing monitoring result, and the nursing monitoring result is used to indicate whether the turning frequency is insufficient.

7. The nursing monitoring system based on the anti-bedsore intelligent mattress according to claim 6 is characterized in that: The classification unit is used for: Using the fully connected layer of the nursing monitoring result generator to perform fully connected encoding on the multi-scale turning semantic encoding feature vector to obtain a fully connected encoding feature vector; Inputting the fully connected encoded feature vector into the Softmax classification function of the nursing monitoring result generator to obtain the probability value of the multi-scale turning semantic encoding feature vector belonging to each classification label, wherein the classification labels include insufficient turning frequency and normal turning frequency; The classification label corresponding to the largest probability value is determined as the nursing monitoring result.

8. A nursing monitoring method based on an anti-bedsore intelligent mattress, characterized in that: include: Acquire a time series of body motion status of a monitored patient object collected by the anti-bedsore smart mattress; Extracting multi-scale semantic coding features of the time series of the body movement state to obtain a first-scale turning-over semantic coding feature vector and a second-scale turning-over semantic coding feature vector; Performing feature-guided fusion on the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector to obtain a multi-scale turning-over semantic coding feature vector; Determining nursing monitoring results based on the multi-scale turning semantic coding feature vector; The step of performing feature-guided fusion on the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector to obtain a multi-scale turning-over semantic coding feature vector includes: Generate Gaussian distribution function coefficients of the first-scale turning-over semantic coding feature vector and the second-scale turning-over semantic coding feature vector using a Gaussian distribution random number function to obtain a first Gaussian distribution function coefficient and a second Gaussian distribution function coefficient; Based on the first Gaussian distribution function coefficient and the second Gaussian distribution function coefficient, the first-scale turning semantic coding feature vector and the second-scale turning semantic coding feature vector are subjected to guided fusion processing to obtain the multi-scale turning semantic coding feature vector.

Citation Information

Cited By

  • Pressure sore prevention mattress regulation and control system based on multi-modal perception and dynamic risk zoning

    CN122031202A