A wireless vital sign monitoring method, device, system, equipment and storage medium

Abstract

The present invention discloses a wireless vital sign monitoring method, device, system, computer equipment and storage medium based on millimeter waves, which are applied to a vital sign monitoring system, wherein the wireless vital sign monitoring method comprises: receiving raw data sent by a millimeter wave sleep recorder in real time, wherein the raw data comprises millimeter wave data recorded by the millimeter wave sleep recorder and monitoring data monitored by a wearable vital sign monitoring device; performing data screening on the millimeter wave data based on signal energy in the millimeter wave data to obtain vital sign data; performing data compression on the vital sign data and the monitoring data; pushing the monitoring data to a user end in real time, and storing the compressed data in real time; calling the compressed data within a preset time period as data to be analyzed, performing data analysis and cross-validation on the data to be analyzed using a pre-trained deep neural network model and a random forest model, and generating a vital sign monitoring report based on the cross-validated analysis data to complete the vital sign monitoring.

Description

Technical Field

[0001] The present invention relates to the field of information processing, and in particular to a method, device, system, computer equipment and storage medium for wireless vital sign monitoring based on millimeter waves. Background Art

[0002] When monitoring the vital signs of the monitored object, in order to improve the accuracy of vital sign monitoring, vital sign monitoring is usually carried out from multiple aspects, such as breathing, sleep, electrocardiogram, blood oxygen, electroencephalogram, etc., but traditional vital sign monitoring requires the monitored object to wear wired monitoring equipment, such as electrocardiograph, polysomnography (PSG), etc., but this wired electrode connection method will affect the daily activities of the monitored object, the comfort is not high, and the displacement of the monitoring equipment caused by the monitored object during the activity will also affect the accuracy of the monitoring results. In addition, millimeter-wave radar can also be used for vital sign monitoring, but the data of millimeter-wave radar is easily affected by environmental factors, and the amount of millimeter-wave radar data is also large and complex. Especially in some scenarios of vital sign monitoring, the monitored object may not be able to wear all the monitoring equipment, which also leads to the overall low accuracy of vital sign monitoring.

[0003] Therefore, how to improve the accuracy of vital sign monitoring while improving the convenience of vital sign monitoring has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] Based on the above situation, the main purpose of the present invention is to provide a millimeter-wave-based wireless vital sign monitoring method, device, system, computer equipment and storage medium, so as to improve the convenience of vital sign monitoring while improving the accuracy of vital sign monitoring.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0006] In a first aspect, an embodiment of the present invention discloses a wireless vital sign monitoring method based on millimeter waves, which is applied to a vital sign monitoring system, wherein the vital sign monitoring system includes a wearable vital sign monitoring device and a millimeter wave sleep recorder, wherein the millimeter wave sleep recorder is wirelessly connected to the wearable vital sign monitoring device to receive monitoring data of the wearable vital sign monitoring device, and the method includes:

[0007] Step S100, receiving raw data sent by the millimeter wave sleep recorder in real time, wherein the raw data includes millimeter wave data of the millimeter wave sleep recorder and monitoring data of the wearable vital sign monitoring device;

[0008] Step S200, filtering the millimeter wave data based on signal energy in the millimeter wave data to obtain vital sign data, where the vital sign data includes chest and abdominal data;

[0009] Step S300, compressing the vital sign data and the monitoring data to obtain compressed data;

[0010] Step S400, pushing the monitoring data to the user end in real time, and storing the compressed data in real time;

[0011] Step S500, calling the compressed data within a preset time period as data to be analyzed;

[0012] Step S600, using a pre-trained deep neural network model to perform data analysis on the data to be analyzed, to obtain first analysis data and a preferred data type of each vital sign indicator, wherein the preferred data type is monitoring data or vital sign data, and the first analysis data includes visualization data corresponding to each of the vital sign indicators;

[0013] Step S700, based on the preferred data type of each vital sign indicator, use a pre-trained random forest model to cross-validate the first analysis data, obtain a vital sign monitoring report based on the verified first analysis data, and push the vital sign monitoring report to the user terminal.

[0014] Optionally, the step S200 includes:

[0015] Step S210, screening the millimeter wave data according to the ratio of target energy to background noise to obtain primary screening data;

[0016] Step S220, processing the initial screening data to obtain vital sign data, wherein the data processing includes static interference removal and dynamic target point screening, and the vital sign data includes chest and abdominal data.

[0017] Optionally, the step S220 further includes:

[0018] Step S221, performing data processing on the primary screening data to obtain secondary screening data, wherein the data processing includes static interference removal and moving target point screening;

[0019] Step S222, using a pre-trained vital sign model to screen the secondary screening data to obtain vital sign data, wherein the vital sign data includes chest and abdominal data.

[0020] Optionally, between step S500 and step S600, the method further includes:

[0021] Step S510, performing data cleaning on the data to be analyzed to obtain cleaned data, wherein the data cleaning includes at least one of removing noise and removing environmental interference;

[0022] Step S520, performing time series processing on the cleaned data to obtain model data.

[0023] Optionally, the step S700 includes:

[0024] Step S710, based on the preferred data type of each physical sign indicator, using a preset random forest model to perform data analysis on the data to be analyzed to obtain second analysis data, wherein the second analysis data includes visualization data corresponding to each physical sign indicator;

[0025] Step S720, performing cross-validation based on the first analysis data and the second analysis data to obtain third analysis data corresponding to each of the physical sign indicators;

[0026] Step S730: Generate a vital sign monitoring report based on the third analysis data, and push the vital sign monitoring report to the user terminal.

[0027] Optionally, the step S710 includes:

[0028] Step S711, selecting data corresponding to the physical sign indicator from the data to be analyzed according to the preferred data type of the physical sign indicator;

[0029] Step S712, using a random forest model to perform data analysis on the data corresponding to the selected physical sign indicators to obtain visualization data corresponding to the physical sign indicators.

[0030] In a second aspect, an embodiment of the present invention discloses a millimeter wave-based wireless vital sign monitoring device, which is applied to a vital sign monitoring system. The vital sign monitoring system includes a wearable vital sign monitoring device and a millimeter wave sleep recorder. The millimeter wave sleep recorder is wirelessly connected to the wearable vital sign monitoring device to receive monitoring data of the wearable vital sign monitoring device. The device includes:

[0031] A data receiving module, used for receiving in real time the raw data sent by the millimeter wave sleep recorder, wherein the raw data includes the millimeter wave data of the millimeter wave sleep recorder and the monitoring data of the wearable vital sign monitoring device;

[0032] A data screening module, configured to screen the millimeter wave data based on signal energy in the millimeter wave data to obtain vital sign data, wherein the vital sign data includes chest and abdominal data;

[0033] A data compression module, used for compressing the vital sign data and the monitoring data to obtain compressed data;

[0034] A push storage module, used to push the monitoring data to the user end in real time, and to store the compressed data in real time;

[0035] A data calling module, used for calling the compressed data within a preset time period as data to be analyzed;

[0036] A data analysis module, used to perform data analysis on the data to be analyzed using a pre-trained deep neural network model to obtain first analysis data and a preferred data type of each vital sign indicator, wherein the preferred data type is monitoring data or vital sign data, and the first analysis data includes visualization data corresponding to each of the vital sign indicators;

[0037] A verification generation module is used to cross-validate the first analysis data based on the preferred data type of each vital sign indicator using a pre-trained random forest model, obtain a vital sign monitoring report based on the verified first analysis data, and push the vital sign monitoring report to the user terminal.

[0038] In a third aspect, an embodiment of the present invention discloses a millimeter wave-based wireless vital sign monitoring system, including a cloud, a wearable vital sign monitoring device, and a millimeter wave sleep recorder, wherein:

[0039] The millimeter wave sleep recorder is wirelessly connected to the wearable vital sign monitoring device, and the millimeter wave sleep recorder is used to obtain millimeter wave data of the monitored object and receive monitoring data of the monitored object by the wearable vital sign monitoring device;

[0040] The cloud is wirelessly connected to the millimeter wave sleep recorder, and the cloud is used to execute the method as described in any one of the first aspects to perform a vital sign analysis on the monitored object and generate a vital sign monitoring report.

[0041] In a fourth aspect, an embodiment of the present invention discloses a computer device, characterized in that it includes: adopting the wireless vital sign monitoring method based on millimeter waves as described in the first aspect; or, including the device as described in the third aspect.

[0042] In a fifth aspect, an embodiment of the present invention discloses a computer-readable storage medium having a computer program stored thereon, wherein the computer program stored in the storage medium is used to be executed by a processor to implement the method described in the first aspect.

[0043] Beneficial Effects

[0044] A millimeter-wave-based wireless vital sign monitoring method, device, system, computer equipment and storage medium disclosed in an embodiment of the present invention are applied to a vital sign monitoring system, wherein the vital sign monitoring system includes a wearable vital sign monitoring device and a millimeter-wave sleep recorder, wherein the wireless vital sign monitoring method includes: receiving raw data sent by the millimeter-wave sleep recorder in real time, the raw data including millimeter-wave data recorded by the millimeter-wave sleep recorder and monitoring data monitored by the wearable vital sign monitoring device; filtering the millimeter-wave data based on the signal energy in the millimeter-wave data to obtain vital sign data; compressing the vital sign data and the monitoring data; pushing the monitoring data to a user end in real time, and storing the compressed data in real time; calling the compressed data within a preset time period as the data to be analyzed, using a pre-trained deep neural network model and a random forest model to perform data analysis and cross-validation on the data to be analyzed, and generating a vital sign monitoring report based on the cross-validated analysis data, and pushing the vital sign monitoring report to the user end to complete the vital sign monitoring. By setting up a wearable vital sign monitoring device to monitor vital signs, it replaces the monitoring device that uses wired electrodes to monitor vital signs in the prior art, which can ensure the comfort and flexibility of the monitored object during vital sign monitoring, thereby improving the convenience of vital sign monitoring while also ensuring the accuracy of the monitoring data. In addition, using two models, the deep neural network model and the random forest model, to analyze and cross-validate the data to be analyzed can improve the accuracy of the generated vital sign monitoring report, and can also ensure the accuracy of the generated vital sign monitoring report when some monitoring data is missing.

[0045] Other beneficial effects of the present invention will be explained in the specific implementation manner through the introduction of specific technical features and technical solutions. Through the introduction of these technical features and technical solutions, those skilled in the art should be able to understand the beneficial technical effects brought about by the technical features and technical solutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The following will describe a preferred embodiment of a millimeter wave-based wireless vital sign monitoring method, device, system, computer equipment and storage medium of the present invention with reference to the accompanying drawings.

[0047] Figure 1 A schematic diagram of the structure of a millimeter wave-based wireless vital sign monitoring system disclosed in this embodiment;

[0048] Figure 2 A schematic diagram of a flow chart of a millimeter wave-based wireless vital sign monitoring method disclosed in this embodiment;

[0049] Figure 3 A schematic diagram of a process for screening millimeter wave data disclosed in this embodiment;

[0050] Figure 4 This is a schematic diagram of a process for cross-validation and generating a physical sign monitoring report using a random forest model disclosed in this embodiment;

[0051] Figure 5 The present invention is a schematic diagram of a framework structure of a millimeter wave-based wireless vital sign monitoring device disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The present invention is described below based on embodiments, but the present invention is not limited to these embodiments. In the following detailed description of the present invention, some specific details are described in detail. In order to avoid confusing the essence of the present invention, known methods, processes, procedures, and components are not described in detail.

[0053] In addition, persons of ordinary skill in the art will appreciate that the drawings provided herein are for illustration purposes and are not necessarily drawn to scale.

[0054] Unless the context clearly requires otherwise, throughout the specification and claims, the words "include", "comprising" and similar words should be interpreted in an inclusive sense rather than an exclusive or exhaustive sense; that is, in the sense of "including but not limited to".

[0055] In the description of the present invention, it should be understood that the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0056] In order to improve the convenience and accuracy of vital sign monitoring, this embodiment discloses a wireless vital sign monitoring method, device, system, computer equipment and storage medium based on millimeter wave. The wireless vital sign monitoring method based on millimeter wave is applied to the vital sign monitoring system, please refer to Figure 1 , Figure 1 This is a schematic diagram of the structure of a millimeter wave-based wireless vital sign monitoring system disclosed in this embodiment.

[0057] like Figure 1As shown in , the vital sign monitoring system 10 includes a wearable vital sign monitoring device 11, a millimeter wave sleep recorder 12 and a cloud 13, wherein the millimeter wave sleep recorder 12 is wirelessly connected to the wearable vital sign monitoring device 11 to receive the monitoring data of the wearable vital sign monitoring device 11, and the cloud 13 is wirelessly connected to the millimeter wave sleep recorder 12. In this embodiment, the wearable vital sign monitoring device 11 can be, for example, an electrocardiogram patch, an electroencephalogram patch, an oximeter, or the like, which does not require wired electrode connection and only requires a patch to be worn. The millimeter wave sleep recorder 12 can be wirelessly connected to the wearable vital sign monitoring device 11 through a Bluetooth module to achieve data reception. The wearable vital sign monitoring device 11 sends the measured monitoring data to the millimeter wave sleep recorder 12 via Bluetooth, and the millimeter wave sleep recorder 12 sends the monitoring data measured by the wearable vital sign monitoring device 11 and the millimeter wave data measured by the millimeter wave sleep recorder 12 to the cloud 13. The cloud 13 processes and analyzes these data to generate a vital sign monitoring report, and sends the vital sign monitoring report to the user terminal 20, which is presented to the user through the user terminal 20.

[0058] Before performing vital sign monitoring, first configure the equipment for vital sign monitoring, install the millimeter wave sleep breathing recorder 12 on the bedside or bracket, turn it on, and face the front of the user's chest. The bracket should be 5 cm away from the bed to avoid the impact of bed vibration; attach the wearable vital sign monitoring device 11 to the corresponding position on the monitored object, for example, wear the EEG electrode on the center of the forehead, wear the ECG electrode on the chest, fix the blood oximeter on the wrist and turn it on. Then make a one-to-one connection through the Bluetooth address to distinguish different Bluetooth devices and devices with different numbers.

[0059] When the device detects an ECG signal, it automatically enters the data collection state, broadcasts data via Bluetooth, establishes a connection with the millimeter-wave sleep apnea recorder 12, and reports data to the millimeter-wave sleep apnea recorder 12; when the device detects an EEG signal, the device turns on Bluetooth, establishes a connection with the millimeter-wave sleep apnea recorder 12, and reports the real-time EEG signal to the millimeter-wave sleep apnea recorder 12; wear the blood oximeter, turn it on, and the Bluetooth starts broadcasting. After the millimeter-wave sleep apnea recorder 12 scans the blood oximeter, it establishes a connection and requests data, and the blood oximeter starts reporting data.

[0060] The millimeter-wave sleep apnea recorder 12 receives ECG data, EEG data, and blood oxygen data via Bluetooth. All link data adopt the DH encryption algorithm for key exchange. Each cloud service device is matched with an independent security certificate and key. Each data exchange link ensures data security, and the data monitored by the wearable vital sign monitoring device 11 is aligned with the data of the millimeter-wave sleep apnea recorder 12 in time, and then compressed, and uploaded to the cloud 13 through a network module (such as a WIFI module or a 4G / 5G module).

[0061] After receiving the data sent by the millimeter wave sleep apnea recorder 12, the cloud 13 performs data decryption and data decompression operations on the data, and then saves the received data into the physical model according to the data format. Then, based on the needs of the user end 20, the data of the preset time period is called to generate a vital sign monitoring report for the monitored object.

[0062] Please refer to Figure 2 , Figure 2 FIG. 1 is a flow chart of a method for wireless vital sign monitoring based on millimeter waves disclosed in this embodiment. Figure 2 As shown in , the millimeter wave-based wireless vital sign monitoring method includes:

[0063] Step S100, receiving the raw data sent by the millimeter wave sleep recorder in real time, the raw data including the millimeter wave data of the millimeter wave sleep recorder and the monitoring data of the wearable vital signs monitoring device. In this embodiment, the wearable vital signs monitoring device sends the measured monitoring data to the millimeter wave sleep recorder via Bluetooth, and the millimeter wave sleep recorder sends the received monitoring data and its own measured millimeter wave data to the cloud, and the cloud receives the raw data sent by the millimeter wave sleep recorder in real time, so as to facilitate the cloud to perform subsequent data processing and analysis.

[0064] Step S200, based on the signal energy in the millimeter wave data, the millimeter wave data is screened to obtain vital sign data, which includes chest and abdominal data. In this embodiment, since the amount of millimeter wave data is large and complex, and the millimeter wave radar signal is susceptible to reflection and scattering of environmental factors (such as furniture, fans, walls, etc.), the millimeter wave data can be screened based on the energy of the radar signal in the millimeter wave data to obtain vital sign data that can reflect the vital sign information of the monitored object. Among them, the millimeter wave data includes radar data at the head position, radar data at the chest and abdomen position, and radar data at the leg position. After data screening, only the radar data at the chest and abdomen position can be retained as vital sign data. In addition, in some embodiments, the radar data at the chest and abdomen position and the radar data at the leg position can also be retained together as vital sign data.

[0065] In an alternative embodiment, see Figure 3 , Figure 3 FIG. 1 is a flow chart of the process of screening millimeter wave data disclosed in this embodiment. Figure 3 As shown in , step S200 includes:

[0066] Step S210, the millimeter wave data is screened according to the ratio of the target energy to the background noise to obtain the initial screening data. In this embodiment, the background noise refers to the signal inherent in the radar signal in the millimeter wave data during the vital sign monitoring process. For example, when the monitored object is not lying in bed, the millimeter wave sleep apnea recorder will also generate a radar signal, and this radar signal is the background noise. Since the ratio (dynamic range) of the strongest target energy to the background noise in the radar signal itself is usually limited and much smaller than the dynamic range corresponding to the number of bits set during storage, the ratio between the target energy and the background noise in the radar signal in the millimeter wave data can be used to screen the millimeter wave data, thereby reducing the amount of millimeter wave data without affecting the data processing effect, and then reducing the data storage requirements through the compression algorithm.

[0067] Step S220, data processing is performed on the initial screening data to obtain vital sign data, the data processing includes static interference removal and moving target point screening, and the vital sign data includes chest and abdominal data. In this embodiment, the initial screening data can be obtained after the millimeter wave data is screened according to the ratio of target energy to background noise, but because the millimeter wave radar signal is easily affected by environmental factors (such as furniture, fans, walls, etc.), the initial screening data obtained may still include noise signals. Therefore, a time-space compression feature representation learning algorithm can be used to pre-process the distance-Doppler map of the millimeter wave radar echo signal, including static interference removal and moving target point screening, so as to remove static interference data from the initial screening data, such as interference radar signals caused by bedside tables, fans, etc. fixed in the environment, and use moving target point screening to screen out the monitoring radar signals for the monitored object, thereby obtaining vital sign data.

[0068] In an optional embodiment, step S220 further includes:

[0069] Step S221, data processing is performed on the primary screening data to obtain secondary screening data, and the data processing includes static interference removal and moving target point screening. In this embodiment, the primary screening data can be obtained after the millimeter wave data is screened according to the ratio of target energy to background noise, but because the millimeter wave radar signal is easily affected by environmental factors (such as furniture, fans, walls, etc.), the primary screening data obtained may still include noise signals. Therefore, a time-space compression feature representation learning algorithm can be used to pre-process the distance-Doppler map of the millimeter wave radar echo signal, including static interference removal and moving target point screening, so as to remove static interference data from the primary screening data, such as interference radar signals caused by bedside tables, fans, etc. fixed in the environment, and use moving target point screening to screen out the monitoring radar signals for the monitored objects, thereby obtaining secondary screening data.

[0070] Step S222, use the pre-trained vital signs model to screen the secondary screening data to obtain vital signs data, which includes chest and abdominal data. In this embodiment, the machine learning method can be used to process the secondary screening data again, and the pre-trained vital signs model can be used to extract key vital signs features, that is, vital signs data, from the secondary screening data, thereby reducing the amount of data while maintaining the data quality of the millimeter wave data. In the specific implementation process, only the radar data at the chest and abdomen position can be retained as the vital signs data. In addition, in some embodiments, the radar data at the chest and abdomen position and the radar data at the leg position can also be retained as the vital signs data.

[0071] Step S300, compress the vital sign data and the monitoring data to obtain compressed data. In this embodiment, the hardware computing acceleration module can be modularly designed and unified into an interface form, and support cross-platform applications. When verifying the accuracy of the algorithm, it can be quickly verified by module. With the help of the hardware accelerator, the compression algorithm fully utilizes the chip performance to achieve efficient data compression.

[0072] Through data screening and data compression, not only can the accuracy of subsequent vital signs monitoring be improved, but also the algorithm and network model can be simplified. While maintaining signal quality, the size of data can be reduced, thereby improving data processing efficiency, compressing MB-level data to KB level. Under the premise of ensuring radar data quality, it relies on smaller network bandwidth resources to achieve the expansion and integration of more other dimensional data.

[0073] It is understandable that step S200 and step S300 do not have a fixed order, and step S200 can be executed first and then step S300; data compression can also be performed while data screening in step S200, that is, step S200 and step S300 can also be executed in parallel.

[0074] Step S400, push the monitoring data to the user end in real time, and store the compressed data in real time. In this embodiment, part or all of the monitoring data can be pushed to the user end, for example, the heart rate data and blood oxygen data in the monitoring data can be pushed to the user end; and the compressed data is serialized and stored in the cloud for subsequent model training and data analysis. In the specific implementation process, the device operation status can also be pushed to the user end in real time, where the device operation status refers to the operation status of the wearable vital sign monitoring device and the millimeter wave sleep recorder in the vital sign monitoring system. By pushing the device operation status in real time, it is convenient for the user end to monitor the operation status of each device in the vital sign monitoring system, so that the user end can promptly discover the operation abnormality of the device (such as disconnection, no network, etc.) and handle it in time.

[0075] Step S500, calling the compressed data within the preset time period as the data to be analyzed. In this embodiment, before generating the vital sign monitoring report, the compressed data within the preset time period can be retrieved from the storage module as the data to be analyzed, so as to generate the vital sign monitoring report based on the data to be analyzed later. In the specific implementation process, the preset time period can be the whole night, or it can be other time periods selected by the user. The cloud can call the data after receiving the report generation request sent by the user, or it can actively call the data to generate a vital sign monitoring report according to the monitoring cycle. For example, the monitoring cycle is from 09:59 on the nth day to 09:58 on the n+1th day Beijing time, and the data is automatically called after one cycle to generate a vital sign monitoring report. It can be understood that the monitoring cycle can be set according to actual needs.

[0076] In an optional embodiment, between step S500 and step S600, the millimeter wave-based wireless vital sign monitoring method further includes:

[0077] Step S510, data cleaning is performed on the data to be analyzed to obtain cleaned data, where data cleaning includes at least one of removing noise and removing environmental interference. In this embodiment, since the millimeter wave radar data, that is, the vital sign data, is susceptible to environmental noise interference, the data to be analyzed may be preprocessed first, including removing noise, removing environmental interference, etc., to improve the signal-to-noise ratio.

[0078] Step S520, time-series processing is performed on the cleaned data to obtain model data. In this embodiment, the cleaned data after data cleaning is regularized, rearranged, and sorted to normalize and time-series the cleaned data, complete data processing, and obtain model data. When generating a vital sign monitoring report, the processed model data can be used to generate a vital sign monitoring report.

[0079] Step S600, using a pre-trained deep neural network model to perform data analysis on the data to be analyzed, and obtain the first analysis data and the preferred data type of each vital sign indicator, the preferred data type is monitoring data or vital sign data, and the first analysis data includes visualization data corresponding to each vital sign indicator. In this embodiment, a pre-trained deep neural network model can be used to perform data analysis on the data to be analyzed, and visualization data corresponding to each vital sign indicator can be obtained, wherein the vital sign indicators may include sleep staging, EEG, EOG, EMG, ECG, body movement waveform, respiratory spectrogram, respiratory rate (bpm), pulse rate (bpm), blood oxygen (%), respiratory waveform, etc.; the blood oxygen index may also include the average SpO2 during the recording period, the lowest SpO2 during the recording period, and the proportion of SpO2 <90% in TRT; the sleep index and the respiratory index may also include other more detailed indicators.

[0080] In the specific implementation process, deep neural networks (DNNs) can integrate multimodal data, improve the accuracy of detecting abnormal vital signs by learning normal and abnormal patterns, and use feature-level fusion, decision-level fusion or model-level fusion to improve the accuracy and robustness of the overall model and the model's ability to understand context. The data to be analyzed is input into a pre-trained deep neural network model, which performs data analysis on the data to be analyzed, so that the preferred data type for different vital signs can be obtained. The preferred data type refers to the optimal data type for vital signs. For example, since millimeter-wave radar relies on speed and distance information to identify respiratory mode, sleeping posture, body shape, etc., it is optimal to select vital signs data for data analysis for respiratory indicators, sleeping posture, body shape and other vital signs, the best is to use EEG to identify sleep stages, the best is to use ECG to identify heart conditions, and the best is to use blood oxygen to identify blood oxygen content. However, within the performance boundary of a single technical field, performance improvement often leads to exponential consumption of computing power and poor results. Therefore, the data to be analyzed, including multi-dimensional data, is jointly input into the deep neural network model. With the advantage of multi-connected data, cross-interpretation of data of different dimensions can be performed, and the advantages of multi-dimensional data in different application scenarios can be fully utilized. The algorithm model calculation can be performed with data of seconds in length to realize health risk interpretation, and the patient's vital signs data can be analyzed in real time, possible health risks can be identified in time, and early warnings can be provided, which is helpful for clinical decision-making and early intervention. It is also suitable for processing irregular and disordered data structures such as human physiological information, and can capture complex relationships between entities.

[0081] Step S700, based on the preferred data type of each vital sign indicator, the first analysis data is cross-validated using a pre-trained random forest model, and a vital sign monitoring report is obtained based on the validated first analysis data, and the vital sign monitoring report is pushed to the user end.

[0082] In this embodiment, after determining the preferred data type of each vital sign indicator, the corresponding data set is selected for different vital sign indicators, and then a pre-trained random forest model is used for prediction. The prediction results of the random forest model are cross-validated with the first analysis data, and a vital sign monitoring report is obtained based on the verified first analysis data, and the vital sign monitoring report is pushed to the user end.

[0083] The random forest model is composed of multiple decision trees, each of which is trained with training samples with replacement. Different sample data will determine different prediction directions. Through cross-validation using an integrated learning method, the prediction results are finally determined and a visual report is generated. The random forest model is used to cross-validate the first analysis data, which not only takes advantage of multi-type sample data, but also makes up for the shortcomings of insufficient verification of single-type data when monitoring vital signs, thereby improving the accuracy of the obtained vital sign monitoring report.

[0084] The random forest algorithm can effectively balance categorical data sets, evaluate the importance of features, and help select the most relevant features, thereby improving the interpretability and accuracy of the model. The random forest model is used for clinical risk prediction, processing survival data, longitudinal data, and multivariate data, providing personalized estimates to predict the development of the disease and the prognosis of patients; it provides an assessment of the importance of features, which helps to understand which factors most affect the risk of disease; it can effectively handle high-dimensional, time-series, and noise problems in medical data, and provide high-quality data input for model training through data preprocessing steps; it relies on multi-connection, different-dimensional vital sign data input, trained algorithm models, and the data performance of the same vital signs in different dimensions, which improves the performance of the model under a single data set, increases the universality of the model, establishes the correlation of different vital sign data, and gives full play to the clinical significance of multi-connection data.

[0085] In an alternative embodiment, see Figure 4 , Figure 4 FIG. 1 is a flow chart of cross-validation and generation of a vital sign monitoring report using a random forest model disclosed in this embodiment. Figure 4 As shown in , step S700 includes:

[0086] Step S710, based on the preferred data type of each vital sign indicator, the data to be analyzed is analyzed using a preset random forest model to obtain second analysis data, and the second analysis data includes visualization data corresponding to each of the vital sign indicators. In this embodiment, based on the preferred data type of each vital sign indicator, corresponding data is selected from the data to be analyzed according to the preferred data type to perform data analysis to obtain second analysis data. For example, for the electrocardiogram indicator, the monitoring data provided by the electrocardiogram patch is selected for data analysis to obtain the second analysis data of the electrocardiogram indicator.

[0087] In an optional embodiment, step S710 further includes:

[0088] Step S711, select data corresponding to the physical sign indicator from the data to be analyzed according to the preferred data type of the physical sign indicator. In this embodiment, for different physical sign indicators, the data corresponding to the physical sign indicator is selected from the data to be analyzed according to the preferred data type of the physical sign indicator. For example, for the electrocardiogram indicator, the monitoring data provided by the electrocardiogram patch is selected.

[0089] Step S712, using the random forest model to analyze the data corresponding to the selected physical sign indicators, and obtain visualization data corresponding to the physical sign indicators. In this embodiment, after the data corresponding to the physical sign indicators are selected, the random forest model is used to analyze the data, so as to obtain visualization data corresponding to the physical sign indicators.

[0090] Step S720, cross-validation is performed based on the first analysis data and the second analysis data to obtain third analysis data corresponding to each of the physical sign indicators. In this embodiment, cross-validation is performed based on the first analysis data and the second analysis data to correct part of the data in the first analysis data to obtain third analysis data, wherein the third analysis data is data obtained by combining the deep neural network model and the random forest model. For example, after obtaining the second analysis data of the ECG indicator using the random forest model, the second analysis data of the ECG indicator can be used to correct the data corresponding to the ECG indicator in the first analysis data.

[0091] Step S730, generating a vital sign monitoring report based on the third analysis data, and pushing the vital sign monitoring report to the user terminal. In this embodiment, the vital sign monitoring report is generated based on the third analysis data, and the third analysis data is visualized, for example, a vital sign monitoring report can be generated in the form of a report or a picture, and the vital sign monitoring report is pushed to the user terminal.

[0092] This millimeter-wave-based wireless vital sign monitoring method is based on the usability of millimeter-wave radar. It uses a wireless method and combines the data with higher monitoring accuracy collected by wearable vital sign monitoring equipment to form a high-precision, simple and easy-to-use vital sign monitoring system. It can increase the features with higher weight items in a convenient way on the basis of ensuring the compliance of the monitored object, thereby improving the accuracy of monitoring. In addition, the wearable vital sign monitoring device and the millimeter-wave sleep recorder transmit data through the ultra-low power Bluetooth communication mode, which removes the constraints of the traditional electrode connection line on the monitored object, greatly improving the usability and popularity of the vital sign monitoring system; it can also perform long-term continuous monitoring, and supports remote monitoring, and can obtain the real-time vital sign data of the monitored object from a remote end, which is conducive to the development of home care and telemedicine services, so that high-quality medical resources can serve more users, reduce the number of medical visits and hospitalization time, and reduce medical costs.

[0093] Please refer to Figure 5 , Figure 5 The present invention discloses a millimeter wave-based wireless vital sign monitoring device. The millimeter wave-based wireless vital sign monitoring device is applied to a vital sign monitoring system, which includes a wearable vital sign monitoring device and a millimeter wave sleep recorder. The millimeter wave sleep recorder is wirelessly connected to the wearable vital sign monitoring device to receive monitoring data from the wearable vital sign monitoring device. Figure 5 As shown in , the millimeter wave-based wireless vital signs monitoring device includes:

[0094] The data receiving module 100 is used to receive the original data sent by the millimeter wave sleep recorder in real time, and the original data includes the millimeter wave data of the millimeter wave sleep recorder and the monitoring data of the wearable vital sign monitoring device.

[0095] The data screening module 200 is used to screen the millimeter wave data based on the signal energy in the millimeter wave data to obtain vital sign data, which includes chest and abdominal data.

[0096] The data compression module 300 is used to compress the vital sign data and the monitoring data to obtain compressed data.

[0097] The push storage module 400 is used to push the monitoring data to the user end in real time and store the compressed data in real time.

[0098] The data calling module 500 is used to call the compressed data within a preset time period as the data to be analyzed.

[0099] The data analysis module 600 is used to perform data analysis on the data to be analyzed using a pre-trained deep neural network model to obtain first analysis data and a preferred data type for each vital sign indicator. The preferred data type is monitoring data or vital sign data. The first analysis data includes visualization data corresponding to each vital sign indicator.

[0100] The verification generation module 700 is used to cross-validate the first analysis data based on the preferred data type of each vital sign indicator using a pre-trained random forest model, obtain a vital sign monitoring report based on the verified first analysis data, and push the vital sign monitoring report to the user end.

[0101] An embodiment of the present invention further provides a computer device, comprising: adopting the wireless vital sign monitoring method based on millimeter waves disclosed in the above embodiment; or comprising the device disclosed in the above embodiment.

[0102] In addition, the present invention also provides a computer-readable storage medium, such as a chip, a CD, etc., on which an execution program is stored, and when the execution program is executed, any of the methods described above is implemented.

[0103] It should be noted that the computer-readable storage medium described in the embodiments of the present disclosure is not limited to the above-mentioned embodiments, and may also be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the embodiments of the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, device, or device.

[0104] A millimeter-wave-based wireless vital sign monitoring method, device, system, computer equipment and storage medium disclosed in an embodiment of the present invention are applied to a vital sign monitoring system, wherein the vital sign monitoring system includes a wearable vital sign monitoring device and a millimeter-wave sleep recorder, wherein the vital sign monitoring method includes: receiving raw data sent by the millimeter-wave sleep recorder in real time, the raw data including millimeter-wave data recorded by the millimeter-wave sleep recorder and monitoring data monitored by the wearable vital sign monitoring device; filtering the millimeter-wave data based on the signal energy in the millimeter-wave data to obtain vital sign data; compressing the vital sign data and the monitoring data; pushing the monitoring data to a user end in real time, and storing the compressed data in real time; calling the compressed data within a preset time period as the data to be analyzed, using a pre-trained deep neural network model and a random forest model to perform data analysis and cross-validation on the data to be analyzed, and generating a vital sign monitoring report based on the cross-validated analysis data, and pushing the vital sign monitoring report to the user end to complete the vital sign monitoring. By setting up a wearable vital sign monitoring device to monitor vital signs, it replaces the monitoring device that uses wired electrodes to monitor vital signs in the prior art, which can ensure the comfort and flexibility of the monitored object during vital sign monitoring, thereby improving the convenience of vital sign monitoring while also ensuring the accuracy of the monitoring data. In addition, using two models, the deep neural network model and the random forest model, to analyze and cross-validate the data to be analyzed can improve the accuracy of the generated vital sign monitoring report, and can also ensure the accuracy of the generated vital sign monitoring report when some monitoring data is missing.

[0105] It will be appreciated by those skilled in the art that, under the premise of no conflict, the above-mentioned preferred solutions can be freely combined and superimposed. Among them, the flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment, or a part of a code, and the module, program segment, or a part of a code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings, for example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions. The numbering of each step in this article is only for the convenience of description and reference, and is not used to limit the order of execution. The specific execution order is determined by the technology itself, and those skilled in the art can determine various allowable and reasonable orders based on the technology itself.

[0106] Those skilled in the art will appreciate that, without conflict, the above-mentioned preferred solutions can be freely combined and superimposed.

[0107] It should be understood that the above-mentioned embodiments are merely illustrative and not restrictive. Without departing from the basic principles of the present invention, various obvious or equivalent modifications or substitutions that can be made by those skilled in the art to the above-mentioned details will all be included in the scope of the claims of the present invention.

Claims

1. A wireless vital sign monitoring method based on millimeter waves, characterized in that: Applied to a vital sign monitoring system, the vital sign monitoring system includes a wearable vital sign monitoring device and a millimeter wave sleep recorder, the millimeter wave sleep recorder is wirelessly connected to the wearable vital sign monitoring device to receive monitoring data of the wearable vital sign monitoring device, the method includes: Step S100, receiving raw data sent by the millimeter wave sleep recorder in real time, wherein the raw data includes millimeter wave data of the millimeter wave sleep recorder and monitoring data of the wearable vital sign monitoring device; Step S200, filtering the millimeter wave data based on signal energy in the millimeter wave data to obtain vital sign data, where the vital sign data includes chest and abdominal data; Step S300, compressing the vital sign data and the monitoring data to obtain compressed data; Step S400, pushing the monitoring data to the user end in real time, and storing the compressed data in real time; Step S500, calling the compressed data within a preset time period as data to be analyzed; Step S600, using a pre-trained deep neural network model to perform data analysis on the data to be analyzed, to obtain first analysis data and a preferred data type of each vital sign indicator, wherein the preferred data type is monitoring data or vital sign data, and the first analysis data includes visualization data corresponding to each of the vital sign indicators; Step S700, based on the preferred data type of each vital sign indicator, use a pre-trained random forest model to cross-validate the first analysis data, obtain a vital sign monitoring report based on the verified first analysis data, and push the vital sign monitoring report to the user terminal.

2. The method for wireless vital sign monitoring based on millimeter waves according to claim 1, characterized in that: The step S200 includes: Step S210, screening the millimeter wave data according to the ratio of target energy to background noise to obtain primary screening data; Step S220, processing the initial screening data to obtain vital sign data, wherein the data processing includes static interference removal and dynamic target point screening, and the vital sign data includes chest and abdominal data.

3. The method for wireless vital sign monitoring based on millimeter waves according to claim 2, characterized in that: The step S220 further includes: Step S221, performing data processing on the primary screening data to obtain secondary screening data, wherein the data processing includes static interference removal and moving target point screening; Step S222, using a pre-trained vital sign model to screen the secondary screening data to obtain vital sign data, wherein the vital sign data includes chest and abdominal data.

4. The method for wireless vital sign monitoring based on millimeter waves according to claim 1, characterized in that: Between step S500 and step S600, the method further includes: Step S510, performing data cleaning on the data to be analyzed to obtain cleaned data, wherein the data cleaning includes at least one of removing noise and removing environmental interference; Step S520, performing time series processing on the cleaned data to obtain model data.

5. The method for wireless vital sign monitoring based on millimeter waves according to claim 1, characterized in that: The step S700 includes: Step S710, based on the preferred data type of each physical sign indicator, using a preset random forest model to perform data analysis on the data to be analyzed to obtain second analysis data, wherein the second analysis data includes visualization data corresponding to each physical sign indicator; Step S720, performing cross-validation based on the first analysis data and the second analysis data to obtain third analysis data corresponding to each of the physical sign indicators; Step S730: Generate a vital sign monitoring report based on the third analysis data, and push the vital sign monitoring report to the user terminal.

6. The method for wireless vital sign monitoring based on millimeter waves according to claim 5, characterized in that: The step S710 includes: Step S711, selecting data corresponding to the physical sign indicator from the data to be analyzed according to the preferred data type of the physical sign indicator; Step S712, using a random forest model to perform data analysis on the data corresponding to the selected physical sign indicators to obtain visualization data corresponding to the physical sign indicators.

7. A millimeter wave based wireless vital sign monitoring device, characterized in that: Applied to a vital sign monitoring system, the vital sign monitoring system includes a wearable vital sign monitoring device and a millimeter wave sleep recorder, the millimeter wave sleep recorder is wirelessly connected to the wearable vital sign monitoring device to receive monitoring data of the wearable vital sign monitoring device, the device includes: A data receiving module (100) is used to receive in real time the raw data sent by the millimeter wave sleep recorder, wherein the raw data includes the millimeter wave data of the millimeter wave sleep recorder and the monitoring data of the wearable vital sign monitoring device; A data screening module (200) is used to screen the millimeter wave data based on signal energy in the millimeter wave data to obtain vital sign data, wherein the vital sign data includes chest and abdominal data; A data compression module (300), used for compressing the vital sign data and the monitoring data to obtain compressed data; A push storage module (400) is used to push the monitoring data to a user terminal in real time, and to store the compressed data in real time; A data calling module (500), used for calling the compressed data within a preset time period as data to be analyzed; A data analysis module (600) is used to perform data analysis on the data to be analyzed using a pre-trained deep neural network model to obtain first analysis data and a preferred data type of each vital sign indicator, wherein the preferred data type is monitoring data or vital sign data, and the first analysis data includes visualization data corresponding to each of the vital sign indicators; A verification generation module (700) is used to cross-validate the first analysis data based on the preferred data type of each vital sign indicator using a pre-trained random forest model, obtain a vital sign monitoring report based on the verified first analysis data, and push the vital sign monitoring report to the user terminal.

8. A millimeter wave-based wireless vital sign monitoring system, characterized in that: Including cloud, wearable vital signs monitoring equipment and millimeter wave sleep recorder, including: The millimeter wave sleep recorder is wirelessly connected to the wearable vital sign monitoring device, and the millimeter wave sleep recorder is used to obtain millimeter wave data of the monitored object and receive monitoring data of the monitored object by the wearable vital sign monitoring device; The cloud is wirelessly connected to the millimeter wave sleep recorder, and the cloud is used to execute the method described in any one of claims 1-6 to perform physical sign analysis on the monitored object and generate a physical sign monitoring report.

9. A computer device, characterized in that: include: Adopt the wireless vital sign monitoring method based on millimeter wave as described in any one of claims 1-6; or include the device as described in claim 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer program stored in the storage medium is used to be executed by a processor to implement the method according to any one of claims 1 to 6.

Images