An intelligent weight monitoring method and system based on a through-hole pressure sensor

Abstract

The present invention provides a method and system for intelligent weight monitoring based on a through-hole pressure sensor. The system includes a detachable through-hole pressure sensor, an A / D conversion module, a communication module, an alarm module, and a data processing module that are matched with each through-hole pressure sensor. The method performs state determination based on the K-clustering method and monitors the patient's weight by calculating the T<supgt;2< / supgt; parameter. The system provided by the present invention is easy to retrofit on a traditional hospital bed. The method provided by the present invention has high processing efficiency, overcomes the drawback of the high cost of traditional weight monitoring systems, and is easy to deploy and retrofit.

Description

Technical Field

[0001] The present invention relates to an intelligent weight detection method and system, and in particular to an intelligent weight detection method and system for patients used on a hospital medical bed. Background Art

[0002] During the treatment of inpatients in hospitals, real-time monitoring of patient weight changes is an important reference factor for evaluating the effectiveness of treatment plans and adjusting treatment plans. Especially in the treatment of new coronavirus pneumonia, the patient's weight parameter is an important indicator for measuring the effectiveness of glucocorticoid drugs. Existing weight detection instruments require patients to leave their beds for weighing, which is not feasible and comfortable for critically ill patients, and cannot meet the needs of intensively weighing patients in time.

[0003] The present invention aims at the problem of difficulty in obtaining real-time weight information of hospitalized patients, especially patients with severe illnesses, and proposes a weight intelligent monitoring method and system based on a through-the-heart pressure sensor. The system collects information on patient weight changes in real time by installing a detachable through-the-heart pressure sensor on the bed support, and has a built-in data analysis module with a self-learning function. The patient's weight data is analyzed and monitored by the data analysis module, and alarm conditions are set in the data analysis module. When the patient's weight data changes to meet the alarm conditions, the system automatically alarms to remind the doctor to intervene in the treatment.

[0004] The intelligent weight detection method and system proposed in the present invention are detachable and can upgrade existing hospital beds to beds with intelligent weight detection functions for patients without any rigid modification to existing beds. They can also be connected and communicated with existing hospital systems, such as the Dicom system, to form an AI diagnosis and treatment reference system. At the same time, the intelligent weight detection method and system proposed in the present invention have a large degree of freedom and can set a variety of bed resting states for patients. They can automatically distinguish and extract the patient's state during the cluster analysis self-learning process, and only extract data that contributes greatly to monitoring the patient's weight for analysis. Therefore, the work efficiency is high, the energy consumption is low, and it has the characteristics of good durability. Summary of the invention

[0005] The present invention proposes a method for intelligent weight monitoring based on a through-the-heart pressure sensor, which is characterized by comprising the following steps:

[0006] S1: setting a cluster analysis model training cycle and a weight data monitoring cycle, and training the model within the training cycle;

[0007] S2: Setting the statistical analysis threshold according to the training results of the cluster analysis model;

[0008] S3: In each body weight data monitoring period, collect new body weight data twice and take the average value, and output the average value as the new body weight data;

[0009] S4: Input the output new body weight data into the clustering analysis model to obtain component data related to body weight detection;

[0010] S5: Analyze the component data related to body weight detection after clustering processing;

[0011] S6: If the statistical analysis index of the component data related to body weight detection exceeds the threshold value, issue an alarm and recommend that a doctor intervene for treatment; otherwise, collect new body weight data again and perform subsequent processing.

[0012] Further, step S1 further includes:

[0013] S11: According to the treatment requirements, set the number of clusters in the clustering analysis model to 2, which respectively represent lying in bed and an empty bed;

[0014] S12: Collect body weight data once every 10 minutes and perform model training according to the K-means method;

[0015] S13: Set the training period to 1 day, and stop training when the training time reaches the training period.

[0016] Further, the body weight data collected in step S12 is the output value of all through-hole pressure sensors, and the body weight data is saved as an n-dimensional vector, where n is the number of all through-hole pressure sensors.

[0017] Further, the body weight data monitoring period in step S1 is 1 hour.

[0018] Further, in step S5, the new body weight data is tested in the manner of T2 test.

[0019] Further, in the T2 test method, use the Hotelling model to calculate the statistic T2, and use the following formula to calculate the control limit of T2:

[0020]

[0021] Among them:

[0022] j is the number of eigenvalues and eigenvectors of the covariance matrix of the component data related to body weight detection selected during data processing;

[0023] α is the number of the through-hole pressure sensor;

[0024] Line represents calculating the control limit.

[0025] Furthermore, the criteria for selecting the number of eigenvalues and eigenvectors are as follows:

[0026]

[0027] Wherein:

[0028] λ i is the eigenvalue of the covariance matrix;

[0029] β is the contribution rate, β ∈ (0, 1], and common values are 0.80, 0.85, 0.90, 0.95, 0.98 or 0.99.

[0030] The present invention also provides an intelligent body weight monitoring system based on a through-hole pressure sensor, including a detachable through-hole pressure sensor disposed on a weight bearing portion, an A / D conversion module, a communication module and an alarm module that match each through-hole pressure sensor, wherein:

[0031] The communication module is used to control the data communication of the intelligent body weight detection system, receive the body weight data from the A / D conversion module, send an alarm signal to the alarm module and perform data interaction with the outside world;

[0032] The alarm module is used to send an alarm message to notify the doctor to intervene;

[0033] It is characterized in that it further includes a data analysis module, and the data analysis module is connected to the communication module and is used to process and analyze the body weight data.

[0034] Furthermore, the data analysis module processes the body weight data in any one of the manners described in claims 1-7.

[0035] Furthermore, the A / D conversion module communicates with the communication module using the ZigBee protocol, the alarm module uses a buzzer for alarm, and the communication module can be connected to a host computer to form a network.

[0036] Using the above intelligent body weight monitoring method and system based on a through-hole pressure sensor, it is possible to provide real-time body weight detection for inpatients, especially critically ill inpatients. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a schematic diagram of the transformation of an embodiment of the present invention;

[0038] Figure 2 is a schematic diagram of the through-hole pressure sensor of an embodiment of the present invention;

[0039] Figure 3 is a schematic diagram of data processing of an embodiment of the present invention;

[0040] Figure 4 Flow chart of an embodiment of the present invention;

[0041] Figure 5 Screenshot of the detection result record of an embodiment of the present invention. Detailed implementation manners

[0042] The intelligent body weight monitoring method and system based on a through-hole pressure sensor proposed by the present invention can be used for the transformation of traditional hospital beds. The implementation methods of the method and system proposed by the present invention are introduced as follows:

[0043] Embodiment 1:

[0044] In this embodiment, a solution for transforming a traditional four-leg bed frame is adopted.

[0045] As Figure 1 shown, in this embodiment, one through-hole pressure sensor is respectively installed on each bed leg of the bed frame, and they are respectively marked as {α1, α2, α3, α4}. Each through-hole pressure sensor is matched with one A / D conversion module and a wireless signal transmitter, which are used for converting the data type and sending the data to the data analysis module.

[0046] Let the pressure of each through-hole pressure sensor be {F 1,t , F 2,t , F 3,t , F 4,t}. Among them, t is the body weight data acquisition moment. At different moments, the pressure change amounts are recorded as ΔF1, ΔF2, ΔF3, ΔF4. Therefore, the total pressure change is The body weight change of the patient is recorded as △M. Then, the body weight change amount of the patient between two adjacent acquisition moments is G is the acceleration of gravity.

[0047] As Figure 2 shown, in this embodiment, the output of the through-hole pressure sensor is an analog current signal of 4 - 20 mA. In the A / D conversion module, the output analog current signal is converted into a digital signal through 8-bit sampling and then sent out by the wireless signal transmitter.

[0048] As Figure 3 shown, in this embodiment, in the intelligent body weight monitoring system, the ZigBee protocol is adopted for signal reception and transmission, and a database is deployed in the system to store the body weight data of the patient. After the system is started, the communication module first receives the pressure data collected by the 4 through-hole pressure sensors and stores the original data in the database. At the same time, the dynamic body weight data of the patient is also sent to the data analysis module for processing and analysis. When external intervention is required, a prompt alarm is sent to the outside.

[0049] In this embodiment, the system operation is divided into two working states: training and real-time monitoring. When monitoring the weight of a new patient or when it is necessary to initiate a new round of weight monitoring for the same patient, the system first enters the training state. When the training time reaches the set duration, the system automatically enters the real-time monitoring state. In this embodiment, the learning cycle is set to 1 day, the data collection interval is 0.5 hours, and the number of sample clusters k is set to 2, representing the patient in bed and the empty bed respectively. During the training cycle, at time point t, the collected sample data is {x (1) ,x (2) ,L x (t)}, and the pressure value x (t) = {F 1,t ,F 2,t ,F 3,t ,F 4,t}. During the training cycle, the following method is used to perform a clustering analysis on each collected data sample:

[0050] a) For each pressure value data collection sample x (t) = {F 1,t ,F 2,t ,F 3,t ,F 4,t}, calculate the cluster it belongs to

[0051] where:

[0052] c (t) is the class closest to the sample among the k clusters;

[0053] ||x (t) - μ k || is the Euclidean distance between x (t) and μ k in space;

[0054] represents the operation of taking the minimum value when comparing the distances to each centroid.

[0055] b) Recalculate the centroid center point μ k .

[0056]

[0057] where m is the total number of sample collections.

[0058] c) Perform iteration.

[0059] Continuously iterate to calculate the μ k value, and calculate the difference between two adjacent centroids until the loop termination condition is met. In this embodiment, the loop termination condition is set to Δμ k < 0.001. Where △μk is the difference in centroid between two adjacent times.

[0060] When the training duration reaches the set learning cycle, the system stops training the model and enters the real-time monitoring state.

[0061] In the real-time monitoring state, the system needs to judge the system state according to the centroid model. When the collected sample is close to the empty bed state, the system does not process the data; when the collected sample is close to the lying state, the system processes it according to the following steps:

[0062] Record the lying data sample set S as: S = {s (1) , s (2) , L, s (q)}, where the number of samples is q and the feature dimension is 4. Calculate the covariance matrix between the 4 feature values:

[0063]

[0064] By solving the matrix equation, obtain the eigenvalue λ i and eigenvector p i of the covariance matrix, and arrange the eigenvalues in ascending order, denoted as λ = {λ1, λ2, λ3, λ4}. Select the elements that satisfy the condition , where j is an integer set as needed in the interval [1, 4], set to 3 in this embodiment; β ∈ (0, 1] is the contribution rate, common values are 0.80, 0.85, 0.90, 0.95, 0.98 or 0.99, set to 0.90 in this embodiment.

[0065] Adopt the Hotelling model, use the F distribution with degrees of freedom (3, 1) and confidence level α ∈ (0, 1], and calculate the control limit of T 2 using the following formula:

[0066]

[0067] When a new lying data x (t+1) is collected, substitute x (t+1) into the following formula for calculation:

[0068]

[0069] where Λ4 is a 4×4 identity diagonal matrix. (x (t+1) ) T is the transposed rank matrix of x (t+1) .

[0070] When , it is judged that an abnormality has occurred in the patient and an alarm is given.

[0071] As Figure 5 shown, when the system is in the real-time monitoring state, the weight monitoring of the first 50 acquisition points of the patient is basically normal. The samples after 50 have a large gap from the standard samples. The method issues a warning and recommends accessing the diagnosis and treatment of the patient's health condition.

[0072] Example 2:

[0073] In this embodiment, the traditional hospital bed is transformed in the manner of Embodiment 1, and the patient's weight is monitored in real time according to the working mode described in Embodiment 1. At the same time, a host computer capable of receiving data of multiple hospital beds is set in the system, and the database system is deployed in the host computer.

[0074] In this embodiment, the server of the hospital Dicom system is selected as the host computer. In the A / D conversion module of each hospital bed, a 1-byte address code and a 4-bit sensor number code are added. When the system is started for the first time, all hospital beds are first reset, and address codes are assigned to all beds. When a new hospital bed is connected to the system, after the initialization of the new hospital bed system, the port data is actively sent to the communication module, which is pushed to the host computer by the communication module. The host computer updates the network data, assigns an address code to the newly added hospital bed, and sends it back to the original data receiving port.

[0075] In addition, the host computer can also regularly and actively traverse and query the data information of all hospital beds in the system and perform backups. In this embodiment, the time for regular traversal and query is set to 00:00 every day.

[0076] When the weight T 2 value of a patient on a certain hospital bed exceeds the preset threshold, the data analysis module of the hospital bed where the patient is located sends an alarm to the host computer through the communication module. The host computer can send an alarm to the medical system terminal of the department where the hospital bed is located, and at the same time retrieve the basic case and recent medical images in the Dicom system for medical staff to refer to.

[0077] The weight intelligent monitoring method and system based on the through-hole pressure sensor provided by the present invention have been introduced in detail above. Specific examples are used in this article to elaborate on the principle and implementation manner of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manner and application scope. The content of this specification should not be construed as a limitation to the technical solution of the present invention.

Claims

1. An intelligent body weight monitoring method based on a through-hole pressure sensor, characterized in that It includes the following steps: S1: Set the training period of the clustering analysis model and the monitoring period of the weight data, and train the model within the training period; the monitoring period of the weight data in step S1 is 1 hour; S2: Set the statistical analysis threshold according to the training result of the clustering analysis model; S3: Within each weight data monitoring period, collect new weight data twice and take the average value, and output the average value as the new weight data; S4: Input the output new weight data into the clustering analysis model to obtain component data related to weight detection; S5: Analyze the component data related to weight detection after clustering processing; S6: If the statistical analysis index of the component data related to weight detection exceeds the threshold, send an alarm and recommend that a doctor intervene for treatment, otherwise collect new weight data again and perform subsequent processing; Step S1 further includes: S11: According to the treatment requirements, set the number of clusters in the clustering analysis model to 2, which respectively represent lying in bed and an empty bed; S12: Collect weight data every 10 minutes and perform model training according to the K-means method; S13: Set the training period to 1 day, and stop training when the training time reaches the training period; The weight data collected in step S12 is the output value of all through-hole pressure sensors, and the weight data is saved as an n-dimensional vector, where n is the number of all through-hole pressure sensors; In the step S5, the new weight data is tested in accordance with T 2 for inspection; In the said T 2 inspection method, the Hotelling model is used to calculate the statistic T 2 , and the following formula is used to calculate the control limit of T 2 : Where: j is the number of eigenvalues and eigenvectors of the covariance matrix of the component data related to weight detection selected during data processing; α is the number of the through-hole pressure sensor; F α is the pressure value of the α-th through-hole pressure sensor; Line represents calculating the control limit; The criteria for selecting the number of eigenvalues and eigenvectors are as follows: Where: λ i is the eigenvalue of the covariance matrix; β is the contribution rate, β ∈ (0, 1].

2. A detection system adopting the weight intelligent monitoring method based on a through-hole pressure sensor described in claim 1, including a detachable through-hole pressure sensor arranged on a weight bearing part, an A / D conversion module, a communication module and an alarm module that match each through-hole pressure sensor, where: The communication module is used to control the data communication of the weight intelligent detection system, receive weight data from the A / D conversion module, send an alarm signal to the alarm module, and perform data interaction with the outside world; The alarm module is used to send an alarm message to notify a doctor to intervene; It further includes a data analysis module, and the data analysis module is connected to the communication module and is used to process and analyze weight data; The A / D conversion module communicates with the communication module using the ZigBee protocol, the alarm module uses a buzzer for alarm, and the communication module can be connected to a host computer to form a network.

Images