Intelligent monitoring system with line management function and monitoring device

Through the multi-domain feature extraction and automatic wire reeling device of the intelligent monitoring system, the shortcomings of traditional monitors in early warning and cable management are solved, efficient abnormal detection and personalized monitoring are achieved, maintenance costs are reduced, and user experience is improved and equipment hygiene standards.

CN119055201AInactive Publication Date: 2024-12-03NANTONG TUMOR HOSPITAL

Patent Information

Application Number
CN202411573530.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-11-06
Publication Date
2024-12-03
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Traditional monitors lack flexibility in early warning mechanisms, preset thresholds are not scientific enough, and cannot be set in personalized according to individual differences. The early warning form is single, and it cannot effectively improve response capabilities and emergency response efficiency. Cable management needs to be manually sorted, which is easy to This leads to cable tangling, wear and damage, increases maintenance costs and replacement frequency, poor user experience, and cannot ensure the cleanliness and order of the cables.

Method used

An intelligent monitoring system was designed, including cable length adjustment module, physiological signal acquisition module, multi-domain feature extraction module, abnormal signal recognition module and abnormal data early warning module. Multi-scale wavelet decomposition and Kalman filtering algorithm are used to extract multiple physiological signals. Domain characteristics and state estimates are used to identify abnormal signals through the state space model and classification identification module, and cables are managed through automatic wire reeling devices.

Benefits of technology

It realizes comprehensive analysis and abnormal detection of physiological signals, improves the accuracy and sensitivity of abnormal detection, provides a personalized early warning mechanism, reduces cable damage and maintenance costs, and improves user experience and equipment hygiene standards.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119055201A_ABST
    Figure CN119055201A_ABST
Patent Text Reader

Abstract

The invention discloses an intelligent monitoring system with a line management function and a monitoring device, which are used in the field of intelligent data processing, and the intelligent monitoring system comprises a cable length adjustment module, a physiological signal acquisition module, a multi-domain feature extraction module, an abnormal signal identification module and an abnormal data early warning module. According to the method, the Kalman filtering algorithm is used for performing state estimation on the state space model, physiological signal data are judged through iterative prediction and state vector updating, abnormal index construction, comprehensive and useful feature data providing, the established abnormal index and a set detection threshold value, normal and abnormal physiological signals are accurately recognized, and the accuracy and accuracy of state estimation are improved. The accuracy and sensitivity of anomaly detection are effectively improved, and timely decision support is provided for medical monitoring.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of intelligent data processing, and in particular to an intelligent monitoring system and a monitoring device with a line management function. Background Art

[0002] A monitor is a medical device used to monitor the patient's physiological parameters in real time; it is usually used in the intensive care unit (ICU), operating room, emergency room and other places in the hospital, as well as in home care; the monitor can continuously monitor and record a variety of physiological indicators, such as heart rate, blood pressure, respiratory rate, blood oxygen saturation, body temperature, etc., connected to the patient through sensors and electrodes, and display the patient's physiological data in real time.

[0003] The monitor can monitor the patient's physiological parameters in real time and issue an alarm in time when abnormal conditions occur. Early warning can help medical staff take quick measures to prevent the condition from worsening. For critically ill patients, the condition may change rapidly in a short period of time. The monitor provides continuous monitoring to ensure that the patient is under close observation throughout the treatment process. During surgery and anesthesia, the patient's physiological parameters may change dramatically. The monitor helps anesthesiologists and surgeons understand the patient's physiological state in real time to ensure the safety of the operation.

[0004] Traditional physiological signal processing technology can often only extract partial features and cannot capture high-frequency and low-frequency components at the same time, resulting in incomplete analysis of the signal's mutation points, transient characteristics, and steady-state spectral characteristics, and easy to miss important information; the existing anomaly detection methods do not adequately process physiological signal noise and have low state estimation accuracy, resulting in low accuracy and real-time performance of anomaly detection, and are unable to provide timely and reliable abnormal signal identification; the traditional early warning mechanism lacks flexibility, the preset thresholds are not scientific enough, and cannot be personalized according to individual differences. The early warning form is single and cannot effectively improve response capabilities and emergency handling efficiency. Traditional cable management requires manual sorting, which can easily lead to cable entanglement, wear and damage, increasing maintenance costs and replacement frequency. At the same time, the ease of use and user experience are poor, and the cables cannot be kept clean and orderly.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] In order to overcome the above problems, the present invention aims to propose an intelligent monitoring system and monitoring device with line management function, with the aim of solving the problems that the traditional early warning mechanism lacks flexibility, the preset threshold is not scientific enough, and it cannot be personalized according to individual differences. The early warning form is single and cannot effectively improve the response capability and emergency handling efficiency. Traditional cable management requires manual sorting, which easily leads to cable entanglement, wear and damage, increases maintenance costs and replacement frequency, and has poor ease of use and user experience, and cannot ensure the neatness and order of the cables.

[0007] To this end, the specific technical solution adopted by the present invention is as follows: According to one aspect of the present invention, there is provided an intelligent monitoring system with a line management function, the intelligent monitoring system comprising: a cable length adjustment module, a physiological signal acquisition module, a multi-domain feature extraction module, an abnormal signal recognition module and an abnormal data early warning module; The cable length adjustment module, the physiological signal acquisition module, the multi-domain feature extraction module, the abnormal signal recognition module and the abnormal data early warning module are connected in sequence; A cable length adjustment module, used to adjust the cable length through an automatic winding device pre-configured on one side of the monitor body; A physiological signal acquisition module, used to collect and obtain physiological signal data using pre-deployed sensors; A multi-domain feature extraction module is used to extract multi-domain feature data from physiological signals using a spectrum analysis method; The abnormal signal recognition module is used to establish a state space model using pre-labeled physiological signal data, and classify the extracted multi-domain feature data to identify normal and abnormal physiological signal data; An abnormal data warning module is used to compare physiological signal data classified as abnormal with a preset physiological signal threshold and issue an early warning for abnormal data exceeding the preset physiological signal threshold; Among them, the abnormal signal recognition module includes: a state space model building module, a state estimation module, a feature extraction module, an abnormal judgment module and a classification identification module; The state space model building module, the state estimation module, the feature extraction module, the abnormality judgment module and the classification identification module are connected in sequence; A state space model building module is used to build a state space model using pre-labeled physiological signal data and determine parameters of the state space model; The state estimation module is used to perform state estimation on the state space model using the Kalman filter algorithm, process noise and obtain the state estimation value of the physiological signal data by iteratively predicting and updating the state vector; A feature extraction module is used to extract features from the state estimation value and construct an abnormality indicator using the extracted features; An abnormality judgment module is used to judge the physiological signal data using abnormal indicators to determine whether the physiological signal data is abnormal; The classification and identification module is used to classify the physiological signal data according to the recognition result, and identify the physiological signal data as normal or abnormal.

[0008] Optionally, the multi-domain feature extraction module includes: Acquire the collected physiological signal data, perform multi-scale wavelet decomposition on the physiological signal data, and record high-frequency wavelet coefficients and low-frequency wavelet coefficients; By analyzing high-frequency wavelet coefficients, the mutation points and transient features in physiological signal data are captured, and the start and end moments of transient changes are identified through the modulus maximum method to extract transient feature data; The low-frequency wavelet coefficients are reconstructed to obtain low-frequency signals representing steady-state components and baseline drift, and the steady-state spectrum feature data are extracted using Fourier transform; The transient feature data and the steady-state spectrum feature data are integrated to form complete time-frequency feature data, and the multi-domain feature data most relevant to the monitoring classification is selected. At the same time, the linear discriminant method is used for dimensionality reduction to produce the final multi-domain feature data.

[0009] Optionally, acquiring the collected physiological signal data, performing multi-scale wavelet decomposition on the physiological signal data, and recording high-frequency wavelet coefficients and low-frequency wavelet coefficients include: Acquire the collected physiological signal data, perform filtering on the collected physiological signal data, and remove noise; Select wavelet basis functions according to the signal length and frequency characteristics of physiological signal data, and determine the number of wavelet decomposition layers according to analysis requirements and signal characteristics; Based on the selected wavelet basis function and the number of wavelet decomposition layers, the preprocessed physiological signal data is subjected to multi-scale wavelet decomposition, and the high-frequency wavelet coefficients and low-frequency wavelet coefficients obtained by the decomposition are recorded.

[0010] Optionally, by analyzing high-frequency wavelet coefficients, the mutation points and transient features in the physiological signal data are captured, and the start and end moments of the transient changes are identified by the modulus maximum method, and the transient feature data are extracted, including: Calculate the modulus maximum of high-frequency wavelet coefficients; By detecting the change of the modulus maximum value, the mutation point and transient characteristics in the physiological signal data are captured, and the start and end time of the transient change are determined; Extracting transient characteristic data based on the start and end time of transient changes; Verify transient feature data and record transient feature data that meets physiological signal expectations; The formula for calculating the modulus maximum is: ; In the formula, represents the modulus maximum; C ( n ) indicates the n wavelet coefficients; Represents wavelet coefficients C ( n )’s absolute value.

[0011] Optionally, the low-frequency wavelet coefficients are reconstructed to obtain a low-frequency signal representing a steady-state component and a baseline drift, and the steady-state spectrum characteristic data are extracted using Fourier transform, including: Reconstruct the low-frequency wavelet coefficients to obtain the low-frequency signal; Perform de-averaging on low-frequency signals to remove DC components; Apply Fourier transform to the low-frequency signal after de-averaging; Obtain the frequency domain signal after Fourier transformation and extract the steady-state spectrum features.

[0012] Optionally, the transient feature data and the steady-state spectrum feature data are integrated to form complete time-frequency feature data, and the multi-domain feature data most relevant to the monitoring classification is selected, and the linear discriminant method is used for dimensionality reduction processing to produce the final multi-domain feature data including: Integrate the transient characteristic data extracted from the high-frequency wavelet coefficients into a transient matrix, where each row in the transient matrix represents transient characteristic data; The steady-state spectrum feature data extracted from the Fourier transform of the low-frequency wavelet coefficients are integrated into a spectrum matrix, where each row in the spectrum matrix represents the spectrum feature data; Integrate the transient matrix and the spectrum matrix in time sequence to form a complete time-frequency feature matrix; Variance analysis is used to select the multi-domain feature data most relevant to the monitoring task from the integrated time-frequency feature data; The linear discriminant method is used to reduce the dimension of the screened multi-domain feature data, and the data after feature screening and dimensionality reduction are integrated into the final multi-domain feature data.

[0013] Optionally, the state estimation module performs state estimation on the state space model by using a Kalman filter algorithm, processes noise and obtains a state estimation value of the physiological signal data by iteratively predicting and updating the state vector, including: Set the initial state vector and initial observation noise covariance matrix; Predict the state vector and covariance matrix; Calculate the Kalman gain, update the state vector and update the covariance matrix; Repeat the prediction and calculation steps to iteratively update the state vector and covariance matrix.

[0014] Optionally, the formula for calculating the Kalman gain is: ; In the formula, represents the Kalman gain; represents the prior error covariance matrix; represents the observation matrix; represents the transpose of the observation matrix; represents the observation noise covariance matrix.

[0015] Optionally, the feature extraction module extracts features from the state estimation value and constructs an abnormality indicator using the extracted features, including: Compute statistical properties of state estimates; Normalize the statistical characteristics; Using the statistical characteristics after normalization, the anomaly index is constructed by calculating the angle of the eigenvector; An abnormality detection threshold is set and the new physiological signal data is judged using the abnormality detection threshold. If the physiological signal data exceeds the abnormality detection threshold, it is judged as an abnormal signal.

[0016] According to another aspect of the present invention, there is also provided a monitoring device with a line management function, the monitoring device comprising a monitor body, an automatic winding device is arranged on one side of the monitor body, a cable is arranged on the outer side of the automatic winding device, and a cuff is arranged on one end of the cable away from the automatic winding device; The automatic winding device includes a mounting slot provided on one side of the monitor body, a motor is provided inside the mounting slot, a plurality of first magnetic blocks are provided at one end of the output shaft of the motor, and the magnetic poles of adjacent first magnetic blocks are opposite magnetic poles; A plurality of first magnetic blocks are provided with hollow grooves on their outer sides, and a plurality of second magnetic blocks are provided at one end of the hollow grooves away from the first magnetic blocks. A winding roller is provided at one end of the second magnetic blocks, and a winding baffle matching the cable is provided on the outer surface of the winding roller.

[0017] Compared with the prior art, this application has the following beneficial effects: 1. The present invention utilizes multi-scale wavelet decomposition technology to extract high-frequency and low-frequency components from physiological signals, realizes a comprehensive analysis of signal mutation points, transient characteristics and steady-state spectrum characteristics, and analyzes high-frequency wavelet coefficients by modulus maximum method, which can accurately capture mutation points and transient changes in physiological signals, identify the start and end time of signal changes, and extract key transient feature data. By reconstructing low-frequency wavelet coefficients and fast Fourier transform, the spectrum characteristics of steady-state components can be accurately extracted, the main frequency components and amplitudes can be identified, and detailed spectrum analysis results can be provided. The transient feature data is integrated with the steady-state spectrum feature data to form a complete time-frequency feature matrix, providing a full range of signal feature descriptions, and enhancing the depth and breadth of data analysis. Through variance analysis and linear discriminant method, the multi-domain features most relevant to the monitoring classification task are screened out, and dimensionality reduction processing is performed to reduce the data dimension while retaining important classification information, thereby improving computational efficiency and classification performance. The module adopts a variety of signal processing technologies to ensure high accuracy and signal-to-noise ratio of data processing, and provides a reliable data basis for subsequent analysis and monitoring.

[0018] 2. The present invention uses the Kalman filtering algorithm to perform state estimation on the state space model, and through iterative prediction and updating of the state vector, accurately processes the noise and obtains the state estimation value of the physiological signal data, thereby ensuring the accuracy and real-time performance of the signal data. A variety of statistical characteristics are extracted from the state estimation value, and normalized, and anomaly indicators are constructed to provide comprehensive and useful feature data, thereby providing a basis for subsequent anomaly detection. The physiological signal data is judged through the established anomaly indicators and the set detection thresholds, and normal and abnormal physiological signals are accurately identified, thereby effectively improving the accuracy and sensitivity of anomaly detection. The classification and identification module can perform real-time classification according to the identification results of the physiological signal data, and identify the signal data as normal or abnormal, thereby providing timely decision support for medical monitoring.

[0019] 3. The present invention further analyzes the physiological signal data that has been classified as abnormal to ensure that each abnormal signal receives timely attention, and further confirms its severity by comparing with preset thresholds, thereby improving the accuracy and effectiveness of abnormality detection. The preset physiological signal thresholds can be flexibly set according to medical standards or historical data to ensure the scientific nature and applicability of the thresholds, adapt to the physiological characteristics of different individuals, and provide personalized monitoring and early warning. The early warning mechanism can take various forms, such as sending instant alarms, recording logs, or triggering automated response measures to ensure that abnormal data receives sufficient attention and processing, and improve response capabilities and emergency handling efficiency.

[0020] 4. With the action of the automatic cable winding device, the present invention allows the cable to be automatically wound up after use, keeping it tidy and orderly, reducing the trouble of manually arranging the cables, improving ease of use and user experience. The design of the automatic cable winding device and the winding baffle can effectively reduce cable damage and wear, reduce the frequency of cable maintenance and replacement, and thus reduce maintenance costs. The cuff is a detachable structure and can be taken off separately for disinfection, ensuring that it can be kept hygienic before each use, thereby improving the safety and hygiene standards of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The above characteristics, features and advantages of the present invention and their implementation methods and methods will become more clearly understood in conjunction with the following description of the embodiments, which are described in detail in conjunction with the accompanying drawings. Herein, a schematic diagram is shown: Figure 1 is a principle block diagram of an intelligent monitoring system with line management function according to an embodiment of the present invention; Figure 2 It is a structural schematic diagram of a monitoring device with line management function according to an embodiment of the present invention; Figure 3 It is a partial structural diagram of a monitoring device with line management function according to an embodiment of the present invention; Figure 4 yes Figure 3 A partial enlarged view of point A in the middle.

[0022] In the figure: 1. Monitor body; 100. Cable length adjustment module; 110. Automatic winding device; 111. Mounting slot; 112. Motor; 113. First magnetic block; 114. Hollow slot; 115. Second magnetic block; 116. Winding roller; 117. Winding bar; 120. Cable; 130. Cuff; 200. Physiological signal acquisition module; 300. Multi-domain feature extraction module; 400. Abnormal signal recognition module; 500. Abnormal data warning module. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0024] According to an embodiment of the present invention, an intelligent monitoring system and a monitoring device with a line management function are provided.

[0025] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to one embodiment of the present invention, there is provided an intelligent monitoring system with a line management function, the intelligent monitoring system comprising: a cable length adjustment module 100, a physiological signal acquisition module 200, a multi-domain feature extraction module 300, an abnormal signal recognition module 400 and an abnormal data early warning module 500; The cable length adjustment module 100, the physiological signal acquisition module 200, the multi-domain feature extraction module 300, the abnormal signal recognition module 400 and the abnormal data warning module 500 are connected in sequence; The cable length adjustment module 100 is used to adjust the cable length through an automatic winding device 110 pre-configured on one side of the monitor body 1 .

[0026] The physiological signal acquisition module 200 is used to collect and acquire physiological signal data using pre-deployed sensors.

[0027] It should be explained that the collected physiological signal data include electrocardiogram, electroencephalogram, blood oxygen saturation, blood pressure, etc.

[0028] The physiological signal acquisition module 200 module includes a sensor, a signal processing unit, a data transmission unit and a storage unit. The sensor is used to collect different types of physiological signals. Common sensors include ECG electrodes, EEG electrodes, pulse oximeters, blood pressure sensors, etc. The signal processing unit is responsible for amplifying, filtering, and analog-to-digital conversion (ADC) of the collected original signal to ensure the accuracy and availability of the signal.

[0029] The data transmission unit is responsible for transmitting the processed physiological signal data to the central processing unit or storage device, and can adopt wired or wireless transmission methods such as Bluetooth, Wi-Fi, etc.

[0030] The data storage unit is used to store the processed physiological signal data for subsequent analysis and historical data tracing.

[0031] The operation process is as follows: Start the physiological signal acquisition module, the power management unit supplies power, and each sensor starts working; Each sensor collects physiological signals and transmits the original signals to the signal processing unit; The signal processing unit amplifies, filters and performs analog-to-digital conversion on the original signal to obtain a usable digital signal; The processed digital signal is sent to the central processing unit or the storage unit through the data transmission unit; Physiological signal data is stored in a local storage unit or transmitted to a cloud server via a network, and users can view real-time and historical data through a display interface or a remote terminal.

[0032] The multi-domain feature extraction module 300 is used to extract multi-domain feature data from physiological signals using a spectrum analysis method.

[0033] Preferably, when the multi-domain feature extraction module 300 extracts multi-domain feature data from the physiological signal using the spectrum analysis method, it includes: Acquire the collected physiological signal data, perform multi-scale wavelet decomposition on the physiological signal data, and record high-frequency wavelet coefficients and low-frequency wavelet coefficients; By analyzing high-frequency wavelet coefficients, the mutation points and transient features in physiological signal data are captured, and the start and end moments of transient changes are identified through the modulus maximum method to extract transient feature data; The low-frequency wavelet coefficients are reconstructed to obtain low-frequency signals representing steady-state components and baseline drift, and the steady-state spectrum feature data are extracted using fast Fourier transform; The transient feature data and the steady-state spectrum feature data are integrated to form complete time-frequency feature data, and the multi-domain feature data most relevant to the monitoring classification is selected. At the same time, the linear discriminant method is used for dimensionality reduction to produce the final multi-domain feature data.

[0034] It should be explained that the multi-domain feature extraction module 300 first obtains the original physiological signal data from the sensor or data acquisition device and performs filtering processing to remove noise and interference to ensure the clarity and accuracy of the data. Next, the multi-scale wavelet decomposition technology is used to select the appropriate wavelet basis function and decomposition layer number according to the length and frequency characteristics of the signal, and the filtered physiological signal data is decomposed to record the obtained high-frequency and low-frequency wavelet coefficients. The high-frequency wavelet coefficients are used to capture the mutation points and transient features in the physiological signal. The start and end times of the signal mutation are identified by the modulus maximum method, and the transient feature data including the transient amplitude and duration are extracted. The low-frequency wavelet coefficients are used to capture the mutation points and transient features in the physiological signal. The data are used to reconstruct the low-frequency signals representing the steady-state components and baseline drift, and the steady-state spectrum features such as the main frequency components and amplitudes are extracted by fast Fourier transform (FFT). Then, the extracted transient feature data and steady-state spectrum feature data are integrated to form a complete time-frequency feature data matrix. Finally, the multi-domain feature data most relevant to the monitoring and classification tasks are screened out by variance analysis or other feature selection methods, and the linear discriminant method (LDA) is used for dimensionality reduction processing to retain the classification information and generate the final multi-domain feature data. These processing steps ensure the efficiency and accuracy from the original physiological signals to the final feature data, providing strong support for subsequent monitoring and classification.

[0035] In wavelet decomposition, "high frequency" and "low frequency" are relative and depend on the current decomposition level (scale). Generally speaking: High frequencies contain higher frequency components relative to the current scale. The specific frequency range depends on the sampling rate and the decomposition level; for example, at a certain decomposition level, high frequencies may refer to frequency components between half and one quarter of the sampling frequency.

[0036] Low frequency includes components with lower frequencies relative to the current scale. The specific frequency range also depends on the sampling rate and decomposition level; for example, at the same decomposition level, low frequency may refer to frequency components below one quarter of the sampling frequency.

[0037] Assume there is a signal with a sampling frequency of 1000Hz and perform three-level wavelet decomposition: 1) First level decomposition: The high-frequency wavelet coefficients represent the components between 500 Hz and 250 Hz.

[0038] The low-frequency wavelet coefficients represent the components between 0 Hz and 250 Hz.

[0039] 2) Second level decomposition: Continue to decompose the low-frequency wavelet coefficients of the first level.

[0040] At this time, the high-frequency wavelet coefficients represent the components between 250Hz and 125Hz.

[0041] The low-frequency wavelet coefficients represent the components between 0 Hz and 125 Hz.

[0042] 3) Third level decomposition: Continue to decompose the low-frequency wavelet coefficients of the second level.

[0043] At this time, the high-frequency wavelet coefficients represent the components between 125Hz and 62.5Hz.

[0044] The low-frequency wavelet coefficients represent the components between 0 Hz and 62.5 Hz.

[0045] Preferably, acquiring the collected physiological signal data, performing multi-scale wavelet decomposition on the physiological signal data, and recording high-frequency wavelet coefficients and low-frequency wavelet coefficients include: Acquire the collected physiological signal data, perform filtering on the collected physiological signal data, and remove noise; Select wavelet basis functions according to the signal length and frequency characteristics of physiological signal data, and determine the number of wavelet decomposition layers according to analysis requirements and signal characteristics; Based on the selected wavelet basis function and the number of wavelet decomposition layers, the preprocessed physiological signal data is subjected to multi-scale wavelet decomposition, and the high-frequency wavelet coefficients and low-frequency wavelet coefficients obtained by the decomposition are recorded.

[0046] It should be explained that the original physiological signal data is obtained from the pre-arranged sensors or data acquisition equipment. The original physiological signal data includes different types of physiological signals such as electrocardiogram (ECG), electroencephalogram (EEG), blood oxygen saturation, etc. The collected physiological signal data is filtered to remove noise and interference. The purpose of filtering is to improve the signal-to-noise ratio of the signal and ensure the accuracy of subsequent analysis. The appropriate wavelet basis function is selected according to the length and frequency characteristics of the physiological signal data. For example, Daubechies wavelet (db) and Haar wavelet are commonly used wavelet basis functions. When selecting the wavelet basis function, it is necessary to consider that different wavelet basis functions have different performances on the local and global characteristics of the signal; analysis requirements: select a wavelet basis function suitable for signal feature extraction; when determining the number of layers of wavelet decomposition, it is necessary to consider the detail resolution and computational complexity of the signal. The more layers, the more details of the decomposition, but the computational complexity will also increase. In general, the number of decomposition layers is determined according to the frequency characteristics of the signal and the analysis requirements.

[0047] Based on the selected wavelet basis function and decomposition layer number, the preprocessed physiological signal data is subjected to multi-scale wavelet decomposition. The specific steps are as follows: Apply wavelet transform: Apply discrete wavelet transform (DWT) to the preprocessed signal data to decompose the signal into high-frequency and low-frequency components of different scales; Record wavelet coefficients: Record the high-frequency wavelet coefficients (detail components) and low-frequency wavelet coefficients (approximate components) obtained by each layer of decomposition. The coefficients reflect the characteristics of the signal at different scales; Record detail components to capture the transient characteristics and mutation points of the signal, and record approximate components to describe the steady-state components and baseline changes of the signal.

[0048] Preferably, by analyzing high-frequency wavelet coefficients, capturing mutation points and transient features in physiological signal data, and identifying the start and end times of transient changes by modulus maximum method, the transient feature data is extracted, including: Calculate the modulus maximum of high-frequency wavelet coefficients; By detecting the change of the modulus maximum value, the mutation point and transient characteristics in the physiological signal data are captured, and the start and end time of the transient change are determined; Extracting transient characteristic data based on the start and end time of transient changes; Verify transient feature data and record transient feature data that meets physiological signal expectations; The formula for calculating the modulus maximum is: ; In the formula, represents the modulus maximum; C ( n ) indicates the nwavelet coefficients; Represents wavelet coefficients C ( n )’s absolute value.

[0049] It should be explained that, first, physiological signal data are collected through pre-deployed sensors, and these data are transmitted to the processing unit for further analysis and processing. Next, the collected physiological signal data are subjected to multi-scale wavelet decomposition. The purpose of this step is to decompose the signal into multiple components of different frequencies, including high-frequency and low-frequency components, select appropriate wavelet basis functions (such as Daubechies wavelet or Haar wavelet), and use discrete wavelet transform to decompose the signal to obtain high-frequency wavelet coefficients and low-frequency wavelet coefficients. By analyzing the high-frequency wavelet coefficients, the mutation points and transient features in the physiological signal data are captured. This step usually adopts the modulus maximum method. By analyzing the changes in high-frequency components, the mutation points in the signal are identified, thereby determining the starting and ending times of the transient changes. According to the captured mutation points and transient features, relevant data features are extracted. The data features may include the amplitude and duration of the mutation point, which describe the changes in the signal at the mutation point. The low-frequency wavelet coefficients are reconstructed to obtain a low-frequency signal representing the steady-state component and baseline drift. Through the inverse wavelet transform, the low-frequency wavelet coefficients are recombined into a new low-frequency signal, which can reflect the baseline and steady-state components of the physiological signal. The reconstructed low-frequency signal is subjected to a fast Fourier transform to extract its spectral features. Through spectral analysis, the main frequency components and corresponding amplitudes of the low-frequency signal can be obtained. The features describe the steady-state spectral characteristics of the signal. The extracted transient feature data are integrated with the steady-state spectral feature data to form complete time-frequency feature data. The purpose of this step is to unify the features of all different time and frequency domains to provide comprehensive feature information for subsequent analysis, select the multi-domain feature data most relevant to the monitoring classification, and perform dimensionality reduction processing. Through correlation analysis, select the features that are most useful for the classification task. Then, use dimensionality reduction techniques such as linear discriminant analysis to reduce the high-dimensional feature data to a lower dimension to reduce computational complexity and improve classification performance.

[0050] Preferably, reconstructing the low-frequency wavelet coefficients to obtain a low-frequency signal representing a steady-state component and a baseline drift, and extracting steady-state spectrum feature data using a fast Fourier transform includes: Reconstruct the low-frequency wavelet coefficients to obtain the low-frequency signal; Perform de-averaging on low-frequency signals to remove DC components; Apply fast Fourier transform to the low-frequency signal after de-averaging; Obtain the frequency domain signal after fast Fourier transformation and extract the steady-state spectrum features.

[0051] It should be explained that, first, the low-frequency wavelet coefficients are reconstructed to obtain a low-frequency signal representing the steady-state component and the baseline drift. This step is achieved by inverse wavelet transform, which recombines the decomposed low-frequency wavelet coefficients into a low-frequency signal.

[0052] The reconstructed low-frequency signal usually contains a DC component, that is, the mean value of the signal. In order to obtain more accurate spectral characteristics, the low-frequency signal needs to be de-meaned to remove the DC component. The specific steps are as follows: Calculate Signal Mean: Calculates the average value of a low-frequency signal.

[0053] Remove the mean: subtract the calculated mean from the low-frequency signal to obtain the low-frequency signal after the mean removal processing.

[0054] Fast Fourier transform (FFT) is applied to the low-frequency signal after de-averaging to convert the signal from time domain to frequency domain. The frequency domain representation of the signal can be obtained through fast Fourier transform (FFT), revealing the amplitude and phase information of the signal at different frequencies.

[0055] The steady-state spectrum features are extracted from the frequency domain signal after the fast Fourier transform. The steady-state spectrum features include but are not limited to the following: Main frequency components: Identify the main frequency components in the signal; Frequency amplitude: extract the amplitude of each main frequency component; Spectral energy distribution: Analyze the distribution of signal energy in the frequency domain.

[0056] Preferably, the transient feature data and the steady-state spectrum feature data are integrated to form complete time-frequency feature data, and the multi-domain feature data most relevant to the monitoring classification is selected, and the linear discriminant method is used for dimensionality reduction processing to produce the final multi-domain feature data including: Integrate the transient characteristic data extracted from the high-frequency wavelet coefficients into a transient matrix, where each row in the transient matrix represents transient characteristic data; The steady-state spectrum feature data extracted from the fast Fourier transform of the low-frequency wavelet coefficients are integrated into a spectrum matrix, where each row in the spectrum matrix represents the spectrum feature data; Integrate the transient matrix and the spectrum matrix in time sequence to form a complete time-frequency feature matrix; Variance analysis is used to select the multi-domain feature data most relevant to the monitoring task from the integrated time-frequency feature data; The linear discriminant method is used to reduce the dimension of the screened multi-domain feature data, and the data after feature screening and dimensionality reduction are integrated into the final multi-domain feature data.

[0057] It should be explained that the transient feature data extracted from the high-frequency wavelet coefficients need to be integrated into a matrix, called a transient matrix. Each row of the transient matrix represents a transient feature data. The transient feature data includes information such as the amplitude and duration of the mutation point.

[0058] Create a transient matrix: Arrange all transient feature data in chronological order, with each row representing a transient event.

[0059] The steady-state spectrum feature data extracted from the fast Fourier transform of the low-frequency wavelet coefficients needs to be integrated into a matrix, called a spectrum matrix. Each row of the spectrum matrix represents a steady-state spectrum feature data. The features may include the main frequency components, amplitudes, etc.

[0060] Create a spectrum matrix: Arrange all steady-state spectrum feature data in chronological order, with each row representing the spectrum feature of a time period.

[0061] Integrating the transient matrix and the spectrum matrix in chronological order to form a complete time-frequency feature matrix can combine transient and steady-state features to provide a comprehensive feature description for each time point. Integrate the time-frequency feature matrix: Align the transient features and spectral features in time to form a comprehensive matrix, where each row represents the complete features of a time point.

[0062] Variance analysis is used to select the multi-domain feature data most relevant to the monitoring task from the integrated time-frequency feature data. Variance analysis can identify features with significant differences between different categories.

[0063] The linear discriminant algorithm (LDA) is used to reduce the dimension of the screened multi-domain feature data. The linear discriminant algorithm reduces the feature dimension while maintaining the classification information by finding the projection direction that can maximize the ratio of between-class variance to intra-class variance. The data after feature screening and dimensionality reduction are integrated into the final multi-domain feature data set. The multi-domain feature data set contains the key features that best characterize the changes in physiological signals and is suitable for subsequent monitoring classification and analysis. The feature data processed by the linear discriminant method are combined into a new feature matrix for subsequent analysis and classification.

[0064] The abnormal signal identification module 400 is used to establish a state space model using pre-labeled physiological signal data, and classify the extracted multi-domain feature data to identify normal and abnormal physiological signal data.

[0065] The abnormal signal identification module 400 includes: a state space model building module, a state estimation module, a feature extraction module, an abnormality judgment module and a classification identification module; The state space model building module, the state estimation module, the feature extraction module, the abnormality judgment module and the classification identification module are connected in sequence; A state space model building module is used to build a state space model using pre-labeled physiological signal data and determine parameters of the state space model; The state estimation module is used to perform state estimation on the state space model using the Kalman filter algorithm, process noise and obtain the state estimation value of the physiological signal data by iteratively predicting and updating the state vector; A feature extraction module is used to extract features from the state estimation value and construct an abnormality indicator using the extracted features; An abnormality judgment module is used to judge the physiological signal data using abnormal indicators to determine whether the physiological signal data is abnormal; The classification and identification module is used to classify the physiological signal data according to the recognition result, and identify the physiological signal data as normal or abnormal.

[0066] It should be explained that the main function of the state space model building module is to use the pre-labeled physiological signal data to establish the state space model and determine the parameters of the model. The state space model is used to describe the dynamics, including the state equation and the observation equation. The state equation and the observation equation can be used to predict and analyze the state changes. The state estimation module uses the Kalman filter algorithm to estimate the state of the state space model. The Kalman filter algorithm iteratively predicts and updates the state vector, processes the noise and obtains the state estimation value of the physiological signal data. This process includes a prediction step and an update step to achieve accurate estimation of the state. The feature extraction module is responsible for extracting features from the state estimation value and using these features to construct anomaly indicators. The extracted features may include statistical characteristics and spectral characteristics of the signal, etc. These features provide basic data for subsequent anomaly detection. The abnormality judgment module uses the extracted abnormality indicators to judge the physiological signal data and identify whether there is an abnormal situation. By comparing the characteristic data with the standard under the normal state, it can determine whether the physiological signal deviates from the normal range, thereby identifying the abnormality; The classification and identification module classifies the physiological signal data according to the recognition results, and identifies the data as normal or abnormal. Based on the output results of the abnormality judgment module, this step determines the health status of each segment of physiological signal data.

[0067] The state is estimated through the Kalman filtering algorithm, the extracted feature data is used to construct anomaly indicators, and anomaly detection is performed through the anomaly judgment module. Finally, the classification identification module identifies the signal data as normal or abnormal.

[0068] Preferably, the state estimation module uses the Kalman filter algorithm to perform state estimation on the state space model, iteratively predicts and updates the state vector, processes noise and obtains the state estimation value of the physiological signal data, including: Set the initial state vector and initial observation noise covariance matrix; Predict the state vector and covariance matrix; Calculate the Kalman gain, update the state vector and update the covariance matrix; Repeat the prediction and calculation steps to iteratively update the state vector and covariance matrix.

[0069] Preferably, the formula for calculating the Kalman gain is: ; In the formula, represents the Kalman gain; represents the prior error covariance matrix; represents the observation matrix; represents the transpose of the observation matrix; represents the observation noise covariance matrix.

[0070] It should be explained that in the state estimation module, the Kalman filter algorithm is used to estimate the state of the state space model, and the noise is processed and the state estimation value of the physiological signal data is obtained by iteratively predicting and updating the state vector. The specific steps are as follows: Set the initial state vector and the initial observation noise covariance matrix. Set the initial state vector, which is usually based on prior knowledge or initial observation data. Set the initial state covariance matrix to describe the uncertainty of the initial state vector. Set the covariance matrix of the observation noise to describe the noise characteristics in the observation data.

[0071] Predict the state vector and covariance matrix, and predict the state vector at the next moment based on the state transfer equation. This step uses the current state vector and state transfer matrix to predict the covariance matrix at the next moment based on the state transfer equation, taking into account the impact of process noise.

[0072] Calculate the Kalman gain, which is a dynamic adjustment factor used to weigh the difference between the predicted value and the observed value, update the state vector and update the covariance matrix. Use the current observed value and the Kalman gain to update the state vector to make it closer to the true value. Use the Kalman gain to update the covariance matrix to reduce the uncertainty of the state. Repeat the prediction and update steps to gradually adjust the state vector and covariance matrix so that it can dynamically track the state, handle noise and improve the accuracy of state estimation.

[0073] Preferably, the feature extraction module extracts features from the state estimation value and constructs an abnormality indicator using the extracted features, including: Compute statistical properties of state estimates; Normalize the statistical characteristics; Using the statistical characteristics after normalization, the anomaly index is constructed by calculating the angle of the eigenvector; An abnormality detection threshold is set and the new physiological signal data is judged using the abnormality detection threshold. If the physiological signal data exceeds the abnormality detection threshold, it is judged as an abnormal signal.

[0074] It is necessary to explain that the statistical properties of the state estimates are calculated, including the calculation of mean, variance, skewness and kurtosis; Among them, mean: calculates the average value of the state estimate, reflecting the central trend of the data.

[0075] Variance: Calculates the variance of the state estimate, which measures the dispersion of the data.

[0076] Skewness: Calculates the skewness of the state estimate, which describes the symmetry of the data distribution.

[0077] Kurtosis: Calculates the kurtosis of the state estimate, which describes the peakedness of the data distribution.

[0078] Normalize the statistical characteristics so that different features have the same dimension for easy comparison. Common methods include minimum-maximum normalization and Z-score normalization.

[0079] The data are scaled to between 0 and 1 and transformed into a standard normal distribution.

[0080] By using the statistical characteristics after normalization, the anomaly index is constructed by calculating the angle of the feature vector. The angle calculation can use cosine similarity or other vector angle measurement methods. According to historical data and experimental results, an anomaly detection threshold is set. This threshold is used to determine whether the physiological signal data is abnormal. The set anomaly detection threshold is used to judge the newly collected physiological signal data. If the feature vector angle of the physiological signal data exceeds the anomaly detection threshold, it is determined to be an abnormal signal. The feature vector angle of the new data is calculated in real time and compared with the threshold.

[0081] The abnormal data warning module 500 is used to compare the physiological signal data classified as abnormal with the preset physiological signal threshold, and to issue a warning for the abnormal data exceeding the preset physiological signal threshold.

[0082] It should be explained that the main function of the abnormal data warning module 500 is to further analyze and warn the physiological signal data classified as abnormal. Specifically, the module compares the abnormal physiological signal data with the preset physiological signal threshold and issues a warning for abnormal data that exceeds the preset threshold.

[0083] Data input classified as abnormal: This module receives the output from the classification and identification module, that is, the physiological signal data that has been identified as abnormal; the abnormal physiological signal data includes various characteristic indicators, such as heart rate, blood pressure, respiratory rate, etc., which have been determined to be abnormal through the previous processing steps.

[0084] Compare with the preset physiological signal threshold: The preset physiological signal threshold is set according to medical standards or historical data, representing the safe range of normal physiological signals. The abnormal data warning module compares each abnormal physiological signal data with the corresponding preset threshold. For example, if a person's heart rate abnormal data is 120 beats / minute, and the preset heart rate threshold is 100 beats / minute, then this data will be marked as exceeding the threshold.

[0085] Early warning mechanism: When abnormal physiological signal data exceeds the preset threshold, the early warning mechanism is triggered, which means that the signal data is not only classified as abnormal, but also its severity exceeds the safety limit and requires special attention; the early warning mechanism can take many forms, such as sending alarm notifications, recording logs, or triggering further automated response measures; sending instant alarms to medical personnel or monitoring terminals to prompt the need for immediate action, recording abnormal data that exceeds the threshold in the log for subsequent analysis and auditing, and triggering preset automated response measures, such as adjusting treatment plans or initiating emergency procedures.

[0086] According to another embodiment of the present invention, Figure 2-Figure 4 As shown, a monitoring device with a line management function is also provided, the monitoring device includes a monitor body 1, an automatic winding device 110 is arranged on one side of the monitor body 1, a cable 120 is arranged on the outer side of the automatic winding device 110, and a cuff 130 is arranged on one end of the cable 120 away from the automatic winding device 110; The automatic winding device 110 includes a mounting slot 111 provided on one side of the monitor body 1, a motor 112 is provided inside the mounting slot 111, a plurality of first magnetic blocks 113 are provided at one end of the output shaft of the motor 112, and the magnetic poles of adjacent first magnetic blocks 113 are opposite magnetic poles; A hollow groove 114 is arranged on the outer side of the plurality of first magnetic blocks 113, and a plurality of second magnetic blocks 115 are arranged at one end of the hollow groove 114 away from the first magnetic block 113, a winding roller 116 is arranged at one end of the second magnetic block 115, and a winding baffle 117 cooperating with the cable 120 is arranged on the outer surface of the winding roller 116.

[0087] It needs to be explained that the magnetic poles of the adjacent first magnetic blocks 113 are set as opposite magnetic poles, so that a stable magnetic field can be formed. When the output shaft of the motor 112 in the automatic winding device 110 rotates, the first magnetic block 113 will be driven to rotate. When the first magnetic block 113 rotates, the second magnetic block 115 will be driven to rotate, and the winding roller 116 will be driven to rotate at the same time. The outer surface of the winding roller 116 is provided with a winding baffle 117 that cooperates with the cable 120. The winding baffle can be used to keep the same spacing between the cables during the winding process, avoid cable entanglement, and realize cable collection.

[0088] It should be noted that the cuff is a detachable structure and can be taken off separately for final disinfection. The cuff is connected to the cable during use.

[0089] In summary, with the help of the above technical scheme of the present invention, the present invention utilizes multi-scale wavelet decomposition technology to extract high-frequency and low-frequency components from physiological signals, realizes a comprehensive analysis of signal mutation points, transient characteristics and steady-state spectrum characteristics, and analyzes high-frequency wavelet coefficients by modulus maximum method, which can accurately capture mutation points and transient changes in physiological signals, identify the start and end time of signal changes, and extract key transient feature data. By reconstructing low-frequency wavelet coefficients and fast Fourier transform, the spectrum characteristics of steady-state components can be accurately extracted, the main frequency components and amplitudes can be identified, and detailed spectrum analysis results can be provided. Transient feature data The data is integrated with the steady-state spectrum feature data to form a complete time-frequency feature matrix, providing a full range of signal feature descriptions, enhancing the depth and breadth of data analysis, and screening out the multi-domain features most relevant to the monitoring classification task through variance analysis and linear discriminant method, and performing dimensionality reduction processing to reduce the data dimension while retaining important classification information, improving computing efficiency and classification performance. The module uses a variety of signal processing technologies to ensure high accuracy and signal-to-noise ratio of data processing, providing a reliable data basis for subsequent analysis and monitoring; the present invention uses the Kalman filtering algorithm to estimate the state of the state space model, and through iterative prediction and updating of the state vector, Accurately process noise and obtain state estimation values ​​of physiological signal data to ensure the accuracy and real-time performance of signal data. Extract multiple statistical characteristics from the state estimation values, perform normalization processing, construct abnormal indicators, provide comprehensive and useful feature data, and provide a basis for subsequent abnormal detection. Through the established abnormal indicators and the set detection thresholds, the physiological signal data is judged, and normal and abnormal physiological signals are accurately identified, effectively improving the accuracy and sensitivity of abnormal detection. The classification and identification module can perform real-time classification according to the identification results of the physiological signal data, identify the signal data as normal or abnormal, and provide timely decision support for medical monitoring. The present invention further analyzes the physiological signal data classified as abnormal to ensure that each abnormal signal is paid attention to in time, and further confirms its severity by comparing with the preset threshold, thereby improving the accuracy and effectiveness of abnormal detection. The preset physiological signal threshold can be flexibly set according to medical standards or historical data to ensure the scientificity and applicability of the threshold, adapt to the physiological characteristics of different individuals, and provide personalized monitoring and early warning. The early warning mechanism can take various forms, such as sending instant alarms, recording logs, or triggering automated response measures, to ensure that abnormal data is fully paid attention to and processed, and improve response capabilities and emergency response efficiency.The present invention, under the action of the automatic winding device, allows the cable to be automatically wound after use, keeping it neat and orderly, reducing the trouble of manually arranging the cables, improving the convenience of use and user experience, and through the design of the automatic winding device and the winding baffle, it can effectively reduce the damage and wear of the cable, reduce the frequency of cable maintenance and replacement, and thus reduce the maintenance cost. The cuff is a detachable structure and can be taken off separately for disinfection, ensuring that it can be kept hygienic before each use, thereby improving the safety and sanitation standards of the equipment. ;

[0090] Although the present invention has been disclosed as above with preferred embodiments, the embodiments are merely examples for the convenience of description and are not intended to limit the present invention. Those skilled in the art may make several changes and modifications without departing from the spirit and scope of the present invention. The scope of protection claimed by the present invention shall be based on the claims.

Claims

1. An intelligent monitoring system with line management function, characterized in that: The intelligent monitoring system comprises: a cable length adjustment module (100), a physiological signal acquisition module (200), a multi-domain feature extraction module (300), an abnormal signal recognition module (400) and an abnormal data early warning module (500); The cable length adjustment module (100), the physiological signal acquisition module (200), the multi-domain feature extraction module (300), the abnormal signal recognition module (400) and the abnormal data warning module (500) are connected in sequence; The cable length adjustment module (100) is used to adjust the cable length through an automatic cable winding device (110) pre-configured on one side of the monitor body (1); The physiological signal acquisition module (200) is used to collect and obtain physiological signal data using pre-deployed sensors; The multi-domain feature extraction module (300) is used to extract multi-domain feature data from physiological signals using a spectrum analysis method; The abnormal signal identification module (400) is used to establish a state space model using pre-labeled physiological signal data, and classify the extracted multi-domain feature data to identify normal and abnormal physiological signal data; The abnormal data warning module (500) is used to compare physiological signal data classified as abnormal with a preset physiological signal threshold, and to issue a warning for abnormal data exceeding the preset physiological signal threshold; The abnormal signal identification module (400) comprises: a state space model construction module, a state estimation module, a feature extraction module, an abnormality judgment module and a classification identification module; The state space model building module, the state estimation module, the feature extraction module, the abnormality judgment module and the classification identification module are connected in sequence; A state space model building module is used to build a state space model using pre-labeled physiological signal data and determine parameters of the state space model; The state estimation module is used to perform state estimation on the state space model using the Kalman filter algorithm, process noise and obtain the state estimation value of the physiological signal data by iteratively predicting and updating the state vector; A feature extraction module is used to extract features from the state estimation value and construct an abnormality indicator using the extracted features; An abnormality judgment module is used to judge the physiological signal data using abnormal indicators to determine whether the physiological signal data is abnormal; The classification and identification module is used to classify the physiological signal data according to the recognition result, and identify the physiological signal data as normal or abnormal.

2. The intelligent monitoring system with line management function according to claim 1, characterized in that: The multi-domain feature extraction module (300) comprises, when extracting multi-domain feature data from a physiological signal using a spectrum analysis method: Acquire the collected physiological signal data, perform multi-scale wavelet decomposition on the physiological signal data, and record high-frequency wavelet coefficients and low-frequency wavelet coefficients; By analyzing high-frequency wavelet coefficients, the mutation points and transient features in physiological signal data are captured, and the start and end moments of transient changes are identified through the modulus maximum method to extract transient feature data; The low-frequency wavelet coefficients are reconstructed to obtain low-frequency signals representing steady-state components and baseline drift, and the steady-state spectrum feature data are extracted using Fourier transform; The transient feature data and the steady-state spectrum feature data are integrated to form complete time-frequency feature data, and the multi-domain feature data most relevant to the monitoring classification is selected. At the same time, the linear discriminant method is used for dimensionality reduction to produce the final multi-domain feature data.

3. The intelligent monitoring system with line management function according to claim 2 is characterized in that: The step of acquiring the collected physiological signal data, performing multi-scale wavelet decomposition on the physiological signal data, and recording high-frequency wavelet coefficients and low-frequency wavelet coefficients includes: Acquire the collected physiological signal data, perform filtering on the collected physiological signal data, and remove noise; Select wavelet basis functions according to the signal length and frequency characteristics of physiological signal data, and determine the number of wavelet decomposition layers according to analysis requirements and signal characteristics; Based on the selected wavelet basis function and the number of wavelet decomposition layers, the preprocessed physiological signal data is subjected to multi-scale wavelet decomposition, and the high-frequency wavelet coefficients and low-frequency wavelet coefficients obtained by the decomposition are recorded.

4. The intelligent monitoring system with line management function according to claim 3 is characterized in that: The method captures the mutation points and transient features in the physiological signal data by analyzing the high-frequency wavelet coefficients, and identifies the start and end time of the transient change by the modulus maximum method, and extracts the transient feature data, including: Calculate the modulus maximum of high-frequency wavelet coefficients; By detecting the change of the modulus maximum value, the mutation point and transient characteristics in the physiological signal data are captured, and the start and end time of the transient change are determined; Extracting transient characteristic data based on the start and end time of transient changes; Verify transient feature data and record transient feature data that meets physiological signal expectations; The formula for calculating the modulus maximum is: ; In the formula, represents the modulus maximum; C ( n ) indicates the n wavelet coefficients; Represents the wavelet coefficients C ( n )’s absolute value.

5. The intelligent monitoring system with line management function according to claim 4, characterized in that: The reconstructing of the low-frequency wavelet coefficients to obtain a low-frequency signal representing a steady-state component and a baseline drift, and extracting steady-state spectrum feature data using Fourier transform includes: Reconstruct the low-frequency wavelet coefficients to obtain the low-frequency signal; Perform de-averaging on low-frequency signals to remove DC components; Apply Fourier transform to the low-frequency signal after de-averaging; Obtain the frequency domain signal after Fourier transformation and extract the steady-state spectrum features.

6. The intelligent monitoring system with line management function according to claim 5, characterized in that: The transient feature data and the steady-state spectrum feature data are integrated to form complete time-frequency feature data, and the multi-domain feature data most relevant to the monitoring classification is selected, and the linear discriminant method is used for dimensionality reduction processing to produce the final multi-domain feature data including: Integrating transient feature data extracted from high-frequency wavelet coefficients into a transient matrix, wherein each row in the transient matrix represents transient feature data; Integrate the steady-state spectrum feature data extracted from the Fourier transform of the low-frequency wavelet coefficients into a spectrum matrix, wherein each row in the spectrum matrix represents the spectrum feature data; Integrate the transient matrix and the spectrum matrix in time sequence to form a complete time-frequency feature matrix; Variance analysis is used to select the multi-domain feature data most relevant to the monitoring task from the integrated time-frequency feature data; The linear discriminant method is used to reduce the dimension of the screened multi-domain feature data, and the data after feature screening and dimensionality reduction are integrated into the final multi-domain feature data.

7. The intelligent monitoring system with line management function according to claim 1, characterized in that: The state estimation module uses the Kalman filter algorithm to perform state estimation on the state space model, iteratively predicts and updates the state vector, processes noise and obtains the state estimation value of the physiological signal data, including: Set the initial state vector and initial observation noise covariance matrix; Predict the state vector and covariance matrix; Calculate the Kalman gain, update the state vector and update the covariance matrix; Repeat the prediction and calculation steps to iteratively update the state vector and covariance matrix.

8. The intelligent monitoring system with line management function according to claim 7, characterized in that: The formula for calculating the Kalman gain is: ; In the formula, represents the Kalman gain; represents the prior error covariance matrix; represents the observation matrix; represents the transpose of the observation matrix; represents the observation noise covariance matrix.

9. The intelligent monitoring system with line management function according to claim 8, characterized in that: The feature extraction module extracts features from the state estimation value and constructs an abnormality indicator using the extracted features, including: Compute statistical properties of state estimates; Normalize the statistical characteristics; Using the statistical characteristics after normalization, the anomaly index is constructed by calculating the angle of the eigenvector; An abnormality detection threshold is set and the new physiological signal data is judged using the abnormality detection threshold. If the physiological signal data exceeds the abnormality detection threshold, it is judged as an abnormal signal.

10. A monitoring device with line management function, used to implement the monitoring analysis of any intelligent monitoring system in claims 1-9, characterized in that: The monitoring device comprises a monitor body (1), an automatic wire winding device (110) being arranged on one side of the monitor body (1), a cable (120) being arranged on the outside of the automatic wire winding device (110), and a cuff (130) being arranged on one end of the cable (120) away from the automatic wire winding device (110); The automatic wire winding device (110) comprises a mounting groove (111) provided on one side of the monitor body (1), a motor (112) being arranged inside the mounting groove (111), a plurality of first magnetic blocks (113) being arranged at one end of an output shaft of the motor (112), and the magnetic poles of adjacent first magnetic blocks (113) being opposite magnetic poles; A plurality of hollow grooves (114) are arranged on the outer sides of the first magnetic blocks (113), and a plurality of second magnetic blocks (115) are arranged at one end of the hollow grooves (114) away from the first magnetic blocks (113), a winding roller (116) is arranged at one end of the second magnetic blocks (115), and a winding stop bar (117) matching the cable (120) is arranged on the outer surface of the winding roller (116).

Citation Information

Patent Citations

  • Abnormality monitoring and alarming method, device and equipment for wearable intelligent underwear

    CN118542652A

  • Automatic winder of medical monitor

    CN213231055U

Cited By

  • Battery health state detection method and system

    CN120577703A