Multifunctional medical bracelet and vital sign real-time monitoring method thereof

Abstract

The invention discloses a multifunctional medical bracelet and a vital sign real-time monitoring method thereof, relates to the technical field of medical monitoring, and aims to solve the technical problems that a patient suffering from lung diseases cannot breathe smoothly and is difficult to perceive in time when sleeping at night, and an existing monitoring means is poor in accuracy and cannot monitor continuously. Comprising a medical bracelet body and a silica gel watchband, a vital sign detection module, a data processing module, a communication module and an alarm module are arranged in the medical bracelet body, and the vital sign detection module is used for collecting vital sign data of a lung disease patient in the daily and night sleep process; by means of the bracelet and the matched vital sign real-time monitoring method, the breathing condition of the patient in the sleeping process can be monitored in real time, and the problem that the patient cannot breathe smoothly in the sleeping process and is difficult to perceive in time is effectively solved; and the situations that the sleep quality of the patient is reduced, the body recovery is influenced, organs are damaged and even the life is threatened due to the fact that the abnormal breathing is not found are avoided.

Description

Technical Field

[0001] The present invention relates to the field of medical monitoring technology, and more particularly to a multifunctional medical wristband and a method for real-time monitoring of vital signs thereof. Background Art

[0002] Lung diseases severely affect the patient's respiratory system, making them prone to shortness of breath during both daytime and nighttime sleep. This problem harms the patient's health in many ways: On the one hand, mild shortness of breath can disrupt the patient's sleep quality, leading to insufficient rest, which in turn affects the body's recovery and subsequent treatment. On the other hand, severe shortness of breath can cause a decrease in the patient's blood oxygen level, causing damage to various organs and even endangering the patient's life.

[0003] Currently, there are many deficiencies in monitoring the respiratory status of patients with lung diseases during sleep: Although traditional hospital monitoring equipment can monitor patients' respiratory data relatively accurately, it requires patients to use it in a specific hospital environment and cannot achieve continuous monitoring of patients' sleep status in daily life.

[0004] Existing wearable devices, however, perform poorly in terms of respiratory monitoring accuracy and timely identification of abnormalities, making them unable to meet the needs of lung disease patients for real-time, precise monitoring of their respiratory status. Some wearable devices are susceptible to external interference when collecting respiratory waveforms, resulting in inaccurate data and an inability to effectively reflect the patient's true respiratory condition. When determining whether breathing is abnormal, they rely solely on simple thresholds and lack comprehensive analysis, which can easily lead to misjudgments or missed calls.

[0005] Therefore, it is urgent to develop a new monitoring technology and equipment to solve the problem of respiratory monitoring during sleep in patients with lung diseases and ensure the health and safety of patients. Summary of the Invention

[0006] The purpose of the present invention is to provide a multifunctional medical bracelet and a real-time vital sign monitoring method thereof, so as to solve the technical problems that patients with lung diseases have difficulty in timely detecting breathing difficulties during sleep day and night, and that existing monitoring methods are inaccurate and cannot be continuously monitored.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: a multifunctional medical bracelet, comprising a medical bracelet body and a silicone strap, wherein the medical bracelet body is equipped with a vital sign detection module, a data processing module, a communication module, and an alarm module; The vital signs detection module is used to collect vital signs data of patients with lung diseases during their daytime and nighttime sleep; The data processing module is used to process the data collected by the vital signs detection module; The communication module is used to transmit the processed data to an external device; The alarm module is used to issue an alarm when abnormalities are detected in the patient's vital sign data.

[0008] By designing a multifunctional medical bracelet and a corresponding real-time vital sign monitoring method, the present invention can accurately monitor the respiratory status of patients with lung diseases during sleep, both day and night. The bracelet's built-in high-sensitivity vital sign detection module collects respiratory data. The data processing module undergoes a complex and precise processing flow, including denoising, feature extraction, and comparison with a standard curve. This allows for the timely detection of respiratory abnormalities, and prompts an alarm through the alarm module. This effectively addresses the difficulty in detecting respiratory distress during sleep in patients with lung diseases, preventing undetected respiratory abnormalities from leading to decreased sleep quality, impaired physical recovery, organ damage, and even life-threatening conditions.

[0009] A method for real-time monitoring of vital signs, characterized by comprising the following steps: S1: First, perform denoising on the respiratory waveform, and then obtain the patient's respiratory waveform during sleep; S2: Collect, calibrate and judge the respiratory waveform data, perform waveform fitting, calculate the mean square error, determine the standard value and establish a standard curve to generate a respiratory signal processing diagram; S3: Set up a respiratory waveform library, screen high-quality data, and generate a respiratory waveform sub-library, i.e., a respiratory waveform curve library; S4: Obtain respiratory waveform curves, extract curves for specific periods and compare and analyze them with normal respiratory curves, mark abnormal curves, and traverse and mark all curves; S5: Analyze and process the marking curve, obtain the marking curve and related information, determine whether to alarm, store the related information if no alarm is needed, and send the alarm information if an alarm is needed.

[0010] Preferably, the step S3 specifically includes the following steps: S301: Constructing a database indexed by time to store respiratory waveforms and respiratory signal processing diagrams, i.e., a respiratory waveform library; S302: Count the number of respiratory waveforms and respiratory signal processing graphs within the monitoring range every day and sort them, and select the graphs with the highest number. The data are used as high-quality monitoring data; S303: Perform cluster analysis on the high-quality monitoring data using the K-Means algorithm to generate a respiratory waveform sub-library, namely a respiratory waveform curve library.

[0011] Preferably, the step S4 specifically includes the following steps: S401: Acquire stored respiratory waveform curves from a respiratory waveform curve library in chronological order; S402: Extract the patient's sleep process to The respiratory waveform curve within the time period is recorded as , and simultaneously obtain the corresponding normal breathing curve; S403: Using the feature vector-based comparison method, calculate Euclidean distance from the normal breathing curve eigenvector If the distance is greater than the set value, the respiratory waveform curve of this section is marked as abnormal; S404: performing the operations of steps S402 to S403 in sequence on all the respiratory waveform curves in the respiratory waveform curve library.

[0012] Preferably, the S403 is specifically performed by the following method for comparison analysis and labeling: First, extract and the characteristic vector of the normal breathing curve, let the breathing waveform curve The Fourier transform of , take the front Fourier coefficients as descriptors ,calculate Euclidean distance from the normal breathing curve eigenvector : ; Where, and They are and the normal breathing curve feature vector elements; Then according to the preset distance threshold ,like , then mark this respiratory waveform curve as abnormal.

[0013] Preferably, the step S5 specifically includes the following steps: S501: Acquire a marked respiratory waveform curve and its corresponding respiratory waveform graph, respiratory signal processing graph, and patient position information; S502: Analyze the marked respiratory waveform curve to determine whether an alarm is required. If an alarm is not required, store the corresponding information; if an alarm is required, send an alarm message.

[0014] Preferably, the step S502 specifically includes the following steps: S502a: Obtain the marked respiratory waveform curve and the type of lung disease the corresponding patient suffers from, and establish a targeted analysis model through a deep learning model; S502b: matching and comparing the marked respiratory waveform curve with respiratory curves in a respiratory waveform curve library; S502c: Finding mismatching points between the marked respiratory waveform curve and the respiratory curves in the respiratory waveform curve library, and generating respiratory waveform curve marking points; S502d: Compare and analyze the respiratory signal processing graph corresponding to the respiratory waveform curve with the respiratory signal processing curves in the respiratory signal processing curve library using the structural similarity index (SSIM), and set a threshold , if the SSIM value is less than , then calculate the mismatch points The respiratory signal processing curve points in the respiratory signal processing curve library Error : ; S502e: Setting Thresholds ,like , mark the point with an alarm; if , no marking is required; S502f: sequentially calculating the errors between the respiratory waveform curve and all respiratory signal processing curves in the respiratory signal processing curve library, and setting an alarm or non-alarm mark based on the results; S502g: Determine whether there are three consecutive If it exists, an alarm message is sent to the respiratory waveform curve; otherwise, normal breathing information is sent.

[0015] Preferably, the deep learning large model adopts a convolutional neural network (CNN), and the network structure includes multiple convolution layers, pooling layers and fully connected layers. The input data is a labeled respiratory waveform curve and its corresponding respiratory signal processing graph. The convolution layer extracts features, the pooling layer performs downsampling, and the fully connected layer performs classification and judgment. The model is trained using a backpropagation algorithm, and the loss function is a cross entropy loss function: ; Where, is the number of categories, is the true label, is the probability value predicted by the model.

[0016] Preferably, the step S502b of matching and comparing the marked respiratory waveform curve with the respiratory curves in the respiratory waveform curve library specifically includes: using a dynamic time warping (DTW) algorithm to calculate the similarity between the two; If the similarity is greater than the set threshold , it is considered a match and proceed to the next step; Otherwise, a preliminary alarm will be processed for the respiratory waveform curve, and the unmatched respiratory waveform curve will be sent to medical staff for further determination whether to issue a formal alarm.

[0017] Preferably, the structural similarity index (SSIM) is used for comparative analysis in step S502d, specifically comprising: assuming that the two respiratory signal processing graphs are and , the calculation formula of SSIM is: ; Where, and is an image and The mean of and is an image and The standard deviation of is an image and The covariance of and is a constant; The closer the calculated SSIM value is to 1, the more similar the two respiratory signal processing graphs are.

[0018] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention, through the design of a multifunctional medical bracelet and a supporting real-time vital signs monitoring method, can accurately monitor the respiratory status of patients with lung diseases during their sleep during the day and at night. The bracelet's built-in high-sensitivity vital signs detection module is used to collect respiratory data. After a complex and precise processing flow of the data processing module, including denoising, feature extraction, and comparison with a standard curve, abnormal breathing can be detected in a timely manner, and an alarm can be issued in a timely manner through the alarm module. This effectively solves the problem of difficulty in detecting shortness of breath during sleep in patients with lung diseases, and avoids the occurrence of situations where the patient's sleep quality is reduced, physical recovery is affected, organs are damaged, and even life-threatening situations occur due to undetected respiratory abnormalities.

[0019] 2. The present invention also provides a more accurate reference for abnormal breathing judgment by designing and constructing a respiratory waveform library and screening high-quality data to generate a respiratory waveform sub-library. Based on a large amount of monitoring data, high-quality data is selected for cluster analysis, covering different types of normal and abnormal respiratory waveform patterns. This allows for more accurate judgment of respiratory abnormalities when comparing and analyzing a patient's respiratory waveform curve, reducing misjudgments and further improving the accuracy of monitoring abnormal breathing in patients, thereby more reliably protecting the patient's health.

[0020] 3. The present invention also achieves more accurate identification of respiratory abnormalities in patients with different types of lung diseases by designing a deep learning large model to establish a targeted analysis model structure. The model takes the labeled respiratory waveform curve and its corresponding respiratory signal processing diagram as input, and after processing by the convolution layer, pooling layer and fully connected layer, it can deeply learn the differences in respiratory characteristics under different disease types. This helps to comprehensively consider the patient's disease type when judging respiratory abnormalities, improve the specificity of abnormal judgment, provide more targeted and accurate monitoring and early warning for different patients, further improve the monitoring effect of the respiratory condition of patients with lung diseases, and better serve the treatment and rehabilitation of patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a schematic structural diagram of the multifunctional medical wristband of the present invention; Figure 2 Schematic diagram of the process of the real-time monitoring method of vital signs in the present invention. DETAILED DESCRIPTION

[0022] Example 1: Figure 1 As shown, the present invention relates to a multifunctional medical bracelet, comprising a medical bracelet body and a silicone strap, wherein the medical bracelet body is equipped with a vital sign detection module, a data processing module, a communication module and an alarm module; The Vital Signs Detection Module collects vital sign data, including respiratory rate, respiratory depth, and blood oxygen saturation, from patients with lung diseases during both daytime and nighttime sleep. Using highly sensitive sensors, it can collect these key vital signs in real time. The module has been specially optimized to address the complex respiratory characteristics of patients with lung diseases, ensuring stable and accurate data acquisition even while the patient is asleep.

[0023] The data processing module processes the data collected by the vital signs detection module. It possesses powerful data processing capabilities and performs real-time processing of the data collected by the vital signs detection module. This processing includes data denoising and feature extraction, providing high-quality data support for subsequent data analysis and anomaly detection.

[0024] The communication module transmits processed data to external devices. Using wireless communication technologies like Bluetooth and Wi-Fi, the processed data is transmitted to the patient's mobile phone, home medical terminal, or hospital server. This enables real-time data synchronization, allowing medical staff and patients' families to keep informed of their health status.

[0025] The alarm module is designed to issue an alert when abnormalities are detected in a patient's vital signs. When a patient's vital signs exceed normal ranges or fluctuate abnormally, an alert is immediately issued. Various alert methods are available, including sound, vibration, flashing lights, text messages, and push notifications to relevant personnel, ensuring that abnormal patient conditions are promptly detected and addressed.

[0026] By designing a multifunctional medical bracelet and a corresponding real-time vital sign monitoring method, the present invention can accurately monitor the respiratory status of patients with lung diseases during sleep, both day and night. The bracelet's built-in high-sensitivity vital sign detection module collects respiratory data. The data processing module undergoes a complex and precise processing flow, including denoising, feature extraction, and comparison with a standard curve. This allows for the timely detection of respiratory abnormalities, and prompts an alarm through the alarm module. This effectively addresses the difficulty in detecting respiratory distress during sleep in patients with lung diseases, preventing undetected respiratory abnormalities from leading to decreased sleep quality, impaired physical recovery, organ damage, and even life-threatening conditions.

[0027] Example 2: Figure 2 As shown, the present invention relates to a method for real-time monitoring of vital signs, comprising the following steps: S1: Acquisition and preprocessing of respiratory waveforms: first, denoising the respiratory waveforms, and then obtaining the respiratory waveforms of the patient during sleep; As another embodiment of the present invention, step S1 specifically includes the following steps: S101: De-noising processing, using a wavelet transform denoising algorithm to process the respiratory waveform.

[0028] Among them, first select a suitable wavelet basis function, such as db4 wavelet basis, to analyze the respiratory waveform signal Perform wavelet decomposition to obtain wavelet coefficients at different scales. Let the decomposition scale be , the wavelet coefficients are , and its calculation formula is: ; Where, , Wavelet mother function, is the scaling factor, Used to control the position of the wavelet function, different scales and location It can capture the characteristics of different details and frequency components of the signal; Then, the wavelet coefficients are processed according to the threshold rule to remove the wavelet coefficients caused by noise. Common threshold rules include hard threshold and soft threshold rules. Here, the soft threshold rule is used. According to the empirical formula Calculate, where is the noise standard deviation, is the signal length. The wavelet coefficients after processing The calculation formula is: ; Finally, the denoised respiratory waveform is obtained through wavelet reconstruction.

[0029] S102: Obtain respiratory waveforms. Utilizing the wristband's built-in high-precision respiratory sensor, the wristband continuously collects respiratory waveforms during the patient's day and nighttime sleep. The frequency of acquisition is based on the respiratory changes of patients with lung diseases to ensure accurate capture of detailed changes in the respiratory waveform. S2: Respiratory signal processing: collecting, calibrating and judging respiratory waveform data, waveform fitting, calculating the mean square error, determining the standard value and establishing a standard curve, and generating a respiratory signal processing graph; As another embodiment of the present invention, step S2 specifically includes the following steps: S201: Data collection: The denoised respiratory waveform data is stored in chronological order to form a data sequence. The storage format uses an efficient data structure to facilitate subsequent rapid reading and processing.

[0030] S202: Calibration and judgment, using a calibration method based on template matching. First, a set of respiratory waveform templates suitable for patients with lung diseases is established. The templates are established by statistically analyzing the normal respiratory waveforms of a large number of patients with lung diseases, including features such as average respiratory frequency, average respiratory depth, and standard waveform shape. For the collected respiratory waveform data, calculate its similarity with each template. The similarity calculation uses the dynamic time warping (DTW) algorithm, assuming that the respiratory waveform sequence to be calibrated is , the template waveform sequence is , DTW distance The calculation process is as follows: ; Where, is the set of all possible regular paths, and The paths are Previous Points in the sequence and The index in is the distance between two corresponding points, usually using Euclidean distance. , if the calculated DTW distance is less than , then the respiratory waveform data is considered to meet the calibration parameters and proceed to the next step; otherwise, reacquire the data and calibrate again; S203: Waveform fitting: polynomial fitting is used to fit the respiratory curve that meets the calibration parameters. Assume that the respiratory curve data point is , ,use polynomial Perform fitting and determine the polynomial coefficients by the least squares method , the objective function is: ; Where, Represents the actual value of all data points The sum of squares of the errors from the fitted values is minimized by , the polynomial coefficients can be determined , so that the fitting curve is as close as possible to the actual respiratory curve, thereby better describing the changing trend of the respiratory curve.

[0031] right about Find the partial derivative and set it to 0, and you will get a set of linear equations. Solving the system of equations will give you the polynomial coefficients. , after fitting, we get the respiratory monitor waveform fitting value , the formula is: ; Where, Generate a fitting curve graph for time to more intuitively observe the changing trend of the respiratory curve. The polynomial coefficients are determined by the least squares method. This formula calculates the polynomial coefficients at different times based on the fitted polynomial. The respiratory waveform fitting value under is used to generate a fitting curve graph; S204: Calculate the mean square error and the mean square error of the waveforms of adjacent cycles in the fitting curve. Assume that the respiratory waveform data of two adjacent cycles are and , the cycle length is , mean square error and The calculation formula is: ; By analyzing the mean square error and the mean square error, the stability and regularity of the respiratory curve can be evaluated; S205: Determine the standard value, sort the calculated periodic waveform mean square error and select the top The mean square error of the parts is taken as the standard value, where the front The standard values are partially expressed as the mean square error and the top 10%-20% of the total data set. This allows for the determination of reasonable standard values based on a large amount of data, allowing the standard curve to more accurately reflect normal respiratory conditions, thereby improving the sensitivity of abnormality detection. The standard values were selected based on statistical analysis of respiratory data from a large number of patients with lung diseases, reflecting the fluctuation range of the respiratory curves of patients with lung diseases under normal respiratory conditions.

[0032] S206: Establish a standard curve, establish standard waveform data based on the standard value, and use the spline interpolation method to construct a standard respiratory curve suitable for patients with lung diseases. Spline interpolation is to construct a smooth curve between given data points. , , through the cubic spline interpolation function In each subinterval The above is a cubic polynomial with continuity of function value, first-order derivative, and second-order derivative at the nodes. A respiratory signal processing graph is generated for subsequent comparative analysis with actual monitoring data.

[0033] S3: Establish a respiratory waveform curve library, set up a respiratory waveform library, screen high-quality data, and generate a respiratory waveform sub-library; As another embodiment of the present invention, step S3 specifically includes the following steps: S301: Set up a respiratory waveform library and build a time-indexed database to store respiratory waveforms and respiratory signal processing diagrams. The database uses a relational database, such as MySQL, to facilitate data management and query. Each data record contains information such as respiratory waveform data, respiratory signal processing diagram data, acquisition time, and patient identification. S302: Filter high-quality data and mark the respiratory waveform graph and respiratory signal processing graph as . Count the number of respiratory waveforms and respiratory signal processing graphs within the monitoring range every day and sort them, and select the ones with the highest number. The data is used as high-quality monitoring data, among which The data selection range is 60%-80% of the daily data volume. This range can ensure that the amount of screened data is sufficient without affecting the processing efficiency due to excessive data. The selection of high-quality data is based on factors such as data completeness, accuracy, and the stability of the patient's condition at the time of collection; S303: Generate respiratory waveform sub-library, establish high-quality monitoring database, extract high-quality monitoring data from the database for analysis and processing. Use cluster analysis algorithm, such as K-Means algorithm, to perform cluster analysis on high-quality monitoring data. Suppose the data point set is , the number of clusters is , the goal of the algorithm is to minimize the sum of the squares of the distances from each data point to the center of the cluster to which it belongs. The objective function is: ; Where, It is clusters, It is The center of the cluster, Represents a data point To the cluster center The algorithm continuously adjusts the cluster center so that the sum of the squares of the distances from each data point to its cluster center is Minimum, thus clustering similar data points into one category, generating a respiratory waveform sub-library, namely the respiratory waveform curve library. This curve library contains different types of normal and abnormal respiratory waveform patterns, providing a reference for subsequent abnormality judgment.

[0034] By designing and constructing a respiratory waveform library and screening high-quality data to generate a respiratory waveform sub-library structure, a more accurate reference is provided for abnormal breathing judgment. Based on a large amount of monitoring data, high-quality data is selected for cluster analysis, covering different types of normal and abnormal respiratory waveform patterns. This allows for more accurate judgment of respiratory abnormalities when comparing and analyzing patient respiratory waveform curves, reducing misjudgments and further improving the accuracy of abnormal breathing monitoring, thereby more reliably protecting patients' health.

[0035] S4: Respiratory abnormality processing process, obtains respiratory waveform curve, extracts curve of specific period and compares and analyzes it with normal respiratory curve, marks abnormal curve, and traverses and marks all curves; As another embodiment of the present invention, step S4 specifically includes the following steps: S401: Acquire a respiratory waveform curve, and acquire stored respiratory waveform curves from a respiratory waveform curve library in chronological order to achieve real-time monitoring of the patient's respiratory condition; S402: Extract the specific time period curve, extract the patient's sleep process, to The respiratory waveform curve within the time period is recorded as , and simultaneously obtain the corresponding normal breathing curve. The normal breathing curve is selected from the respiratory waveform curve library to represent the patient's respiratory waveform in a stable and healthy state; S403: Comparison analysis and marking, using a feature vector-based comparison method. First, extract The characteristic vector of the normal breathing curve includes respiratory frequency, respiratory depth, waveform shape features, etc. The waveform shape features are extracted by Fourier descriptor. The Fourier transform of , take the front Fourier coefficients as descriptors .calculate Euclidean distance from the normal breathing curve eigenvector : ; Where, and They are and the normal breathing curve feature vector elements. According to the preset distance threshold ,like , then mark the respiratory waveform curve of this section as abnormal; S404: Traverse the mark and perform the operations of steps S402-S403 on all the respiratory waveform curves in the respiratory waveform curve library in sequence to ensure comprehensive monitoring of the patient's respiratory condition. By traversing the mark, abnormal changes in the patient's breathing during the entire sleep process can be discovered in a timely manner.

[0036] S5: Marking curve analysis and processing, obtaining the marking curve and related information, judging whether to alarm, storing the relevant information if no alarm is needed, and sending the alarm information if an alarm is needed; As another embodiment of the present invention, step S5 specifically includes the following steps: S501: Obtain the marked curve and related information, obtain the respiratory waveform curve marked in step S4, and extract the corresponding respiratory waveform graph, respiratory signal processing graph, and patient location information. The patient location information is obtained through the built-in positioning module of the wristband, such as GPS or Bluetooth positioning technology.

[0037] S502: Determine whether to issue an alarm. The marked respiratory waveform curve is analyzed to determine whether an alarm is necessary. If an alarm is not necessary, the curve is identified as a high-quality respiratory waveform curve, its corresponding information is stored, and added to the respiratory waveform curve library. If an alarm is necessary, an alarm message is sent to the patient, medical staff, and relevant guardians.

[0038] As another embodiment of the present invention, in step S502, analyzing the marked respiratory waveform curve to determine whether an alarm is required specifically includes the following steps: S502a: Establish an analysis model, obtain the labeled respiratory waveform curve and the type of lung disease the corresponding patient suffers from, and establish a targeted analysis model through a deep learning large model. The deep learning large model uses a convolutional neural network (CNN), and the network structure includes multiple convolutional layers, pooling layers, and fully connected layers. The input data is the labeled respiratory waveform curve and its corresponding respiratory signal processing graph. The convolution layer extracts features, the pooling layer performs downsampling, and the fully connected layer performs classification and judgment. The model is trained using the backpropagation algorithm, and the loss function is the cross-entropy loss function: ; Where, is the number of categories, is the true label, is the probability value predicted by the model. The model is trained using a large amount of respiratory data from patients with lung diseases, enabling it to accurately identify different types of respiratory abnormalities. As another embodiment of the present invention, the establishment of the deep learning large model in S502a specifically includes the following steps: A: Data labeling: We collect and label a large number of respiratory waveforms, respiratory signal processing diagrams, and detailed information about lung diseases from patients with lung diseases. Labeling includes information such as disease type, severity, patient age, and gender, so that the model can learn how different factors affect respiratory characteristics. B: Correlation data acquisition: obtain the respiratory waveform, respiratory signal processing curve, and vital signs data corresponding to the above data, including blood oxygen content, heart rate, etc. These data are correlated and integrated to form a complete data set; C: Data correspondence: Respiration waveforms and respiratory signal processing curves are mapped to vital sign data to generate vital sign data charts. The charts use time as the horizontal axis and display different types of data in the form of curves or bar graphs, making it easier to observe the correlation between the data. D: Curve comparison and marking: Compare the respiratory waveform curve and the respiratory signal processing curve, and decide whether to mark the curve based on the differences in curve shape, frequency, amplitude, and other characteristics. If the curve has obvious abnormal characteristics, such as too fast or too slow respiratory rate, irregular waveform shape, etc., it will be marked as an abnormal sample for model training; E: Analysis and processing of marked vital signs.

[0039] By designing a large deep learning model to establish a targeted analysis model structure, more accurate identification of respiratory abnormalities in patients with different types of lung diseases is achieved. The model takes the labeled respiratory waveform curve and its corresponding respiratory signal processing graph as input, and after processing through convolutional layers, pooling layers, and fully connected layers, it can deeply learn the differences in respiratory characteristics under different disease types. This helps to comprehensively consider the patient's disease type when judging respiratory abnormalities, improve the specificity of abnormality judgment, and provide more targeted and accurate monitoring and early warning for different patients, further improving the monitoring effect of the respiratory status of patients with lung diseases and better serving patients' treatment and rehabilitation.

[0040] S502b: Curve matching comparison, the marked respiratory waveform curve is matched and compared with the respiratory curve in the respiratory waveform curve library. The dynamic time warping (DTW) algorithm is used to calculate the similarity between the two. If the similarity is greater than the set threshold, If the respiratory waveform does not match, the respiratory waveform is preliminarily alarmed and the unmatched respiratory waveform is sent to medical staff for further determination of whether to formally alarm. S502c: Generate marker points, find mismatching points between the marked respiratory waveform curve and the respiratory curves in the respiratory waveform curve library, and generate respiratory waveform marker points. The marker points can be generated by comparing the difference points between the two curves in the time series. For example, when the amplitude difference between the two curves exceeds a certain threshold, the time point is marked as a mismatching point. S502d: Compare and analyze the respiratory signal processing curves, compare and analyze the respiratory signal processing graph corresponding to the respiratory waveform curve with the respiratory signal processing curves in the respiratory signal processing curve library, and use the structural similarity index (SSIM) for measurement. Suppose the two respiratory signal processing graphs are and , the calculation formula of SSIM is: ; Where, and is an image and The mean of and is an image and The standard deviation of is an image and The covariance of and Is a constant used to avoid the denominator being 0. The closer the calculated SSIM value is to 1, the more similar the two respiratory signal processing graphs are. Set the threshold , if the SSIM value is less than , then calculate the mismatch points The respiratory signal processing curve points in the respiratory signal processing curve library Error : ; S502e: Set threshold judgment, set threshold , according to the calculated Value judgment. , mark the point with an alarm; if , no marking is required; S502f: Comprehensive Error Calculation and Marking: This step sequentially calculates the error between the respiratory waveform curve and all respiratory signal processing curves in the respiratory signal processing curve library, and then issues an alarm or non-alarm based on the result. This comprehensive error calculation allows for a more accurate assessment of the degree of abnormality in the respiratory waveform curve.

[0041] S502g: Continuous abnormality judgment, judge whether there are three consecutive abnormalities in the respiratory waveform curve If it exists, an alarm message is sent to the respiratory waveform curve; if it does not exist, normal breathing information is sent.

[0042] The embodiments disclosed in the present invention are preferred embodiments, but are not limited to them. Ordinary technicians in this field can easily understand the spirit of the present invention based on the above embodiments and make different extensions and changes. As long as they do not deviate from the spirit of the present invention, they are all within the scope of protection of the present invention.

Claims

1. A method for real-time monitoring of vital signs, characterized in that: The steps include: S1: First, perform denoising on the respiratory waveform, and then obtain the patient's respiratory waveform during sleep; S2: Collect, calibrate and judge the respiratory waveform data, perform waveform fitting, calculate the mean square error, determine the standard value and establish a standard curve to generate a respiratory signal processing diagram; S3: Set up a respiratory waveform library, screen high-quality data, and generate a respiratory waveform sub-library, i.e., a respiratory waveform curve library; S4: Obtain respiratory waveform curves, extract curves for specific periods and compare and analyze them with normal respiratory curves, mark abnormal curves, and traverse and mark all curves; S5: Analyze and process the marking curve, obtain the marking curve and related information, determine whether to alarm, store the related information if no alarm is needed, and send the alarm information if an alarm is needed.

2. A method for real-time monitoring of vital signs according to claim 1, characterized in that: The step S3 specifically includes the following steps: S301: Constructing a database indexed by time to store respiratory waveforms and respiratory signal processing diagrams, i.e., a respiratory waveform library; S302: Count the number of respiratory waveforms and respiratory signal processing graphs within the monitoring range every day and sort them, and select the graphs with the highest number. The data are used as high-quality monitoring data; S303: Perform cluster analysis on the high-quality monitoring data using the K-Means algorithm to generate a respiratory waveform sub-library, namely a respiratory waveform curve library.

3. A method for real-time monitoring of vital signs according to claim 1, characterized in that: The step S4 specifically includes the following steps: S401: Acquire stored respiratory waveform curves from a respiratory waveform curve library in chronological order; S402: Extract the patient's sleep process to The respiratory waveform curve within the time period is recorded as , and simultaneously obtain the corresponding normal breathing curve; S403: Using the feature vector-based comparison method, calculate Euclidean distance from the normal breathing curve eigenvector If the distance is greater than the set value, the respiratory waveform curve of this section is marked as abnormal; S404: performing the operations of steps S402 to S403 in sequence on all the respiratory waveform curves in the respiratory waveform curve library.

4. A method for real-time monitoring of vital signs according to claim 3, characterized in that: The S403 is specifically performed by the following method for comparison analysis and marking: First, extract and the characteristic vector of the normal breathing curve, let the breathing waveform curve The Fourier transform of , take the front Fourier coefficients as descriptors ,calculate Euclidean distance from the normal breathing curve eigenvector : ； Where, and They are and the normal breathing curve feature vector elements; Then according to the preset distance threshold ,like , then mark this respiratory waveform curve as abnormal.

5. A method for real-time monitoring of vital signs according to claim 1, characterized in that: The step S5 specifically includes the following steps: S501: Acquire a marked respiratory waveform curve and its corresponding respiratory waveform graph, respiratory signal processing graph, and patient position information; S502: Analyze the marked respiratory waveform curve to determine whether an alarm is required. If an alarm is not required, store the corresponding information; if an alarm is required, send an alarm message.

6. A method for real-time monitoring of vital signs according to claim 5, characterized in that: The step S502 specifically includes the following steps: S502a: Obtain the marked respiratory waveform curve and the type of lung disease the corresponding patient suffers from, and establish a targeted analysis model through a deep learning model; S502b: matching and comparing the marked respiratory waveform curve with respiratory curves in a respiratory waveform curve library; S502c: Finding mismatching points between the marked respiratory waveform curve and the respiratory curves in the respiratory waveform curve library, and generating respiratory waveform curve marking points; S502d: Compare and analyze the respiratory signal processing graph corresponding to the respiratory waveform curve with the respiratory signal processing curves in the respiratory signal processing curve library using the structural similarity index (SSIM), and set a threshold , if the SSIM value is less than , then calculate the mismatch points The respiratory signal processing curve points in the respiratory signal processing curve library Error : ； S502e: Setting Thresholds ,like , mark the point with an alarm; if , no marking is required; S502f: sequentially calculating the errors between the respiratory waveform curve and all respiratory signal processing curves in the respiratory signal processing curve library, and setting an alarm or non-alarm mark based on the results; S502g: Determine whether there are three consecutive If it exists, an alarm message is sent to the respiratory waveform curve; otherwise, normal breathing information is sent.

7. A method for real-time monitoring of vital signs according to claim 6, characterized in that: The deep learning model uses a convolutional neural network (CNN). The network structure includes multiple convolutional layers, pooling layers, and fully connected layers. The input data is a labeled respiratory waveform curve and its corresponding respiratory signal processing graph. The convolution layer extracts features, the pooling layer performs downsampling, and the fully connected layer performs classification and judgment. The model is trained using the backpropagation algorithm, and the loss function is the cross-entropy loss function: ； Where, is the number of categories, is the true label, is the probability value predicted by the model.

8. A method for real-time monitoring of vital signs according to claim 7, characterized in that: The step S502b matches and compares the marked respiratory waveform curve with the respiratory curves in the respiratory waveform curve library, specifically including: using a dynamic time warping (DTW) algorithm to calculate the similarity between the two; If the similarity is greater than the set threshold , it is considered a match and proceed to the next step; Otherwise, a preliminary alarm will be processed for the respiratory waveform curve, and the unmatched respiratory waveform curve will be sent to medical staff for further determination whether to issue a formal alarm.

9. A method for real-time monitoring of vital signs according to claim 8, characterized in that: The structural similarity index (SSIM) is used for comparative analysis in step S502d, specifically including: assuming that the two respiratory signal processing graphs are and , the calculation formula of SSIM is: ； Where, and is an image and The mean of and is an image and The standard deviation of is an image and The covariance of and is a constant; The closer the calculated SSIM value is to 1, the more similar the two respiratory signal processing graphs are.

10. A multifunctional medical wristband, which uses the real-time vital sign monitoring method according to any one of claims 1 to 9, comprising a medical wristband body and a silicone strap, characterized in that: The medical bracelet body is built with: The vital signs detection module is used to collect vital signs data of patients with lung diseases during their daytime and nighttime sleep; A data processing module, used to process the data collected by the vital signs detection module; a communication module for transmitting processed data to an external device; The alarm module is used to issue an alarm when abnormalities are detected in the patient's vital signs data.

Images