Cardiovascular tumor emergency monitoring and early warning system based on wearable device

Abstract

The invention relates to the technical field of emergency monitoring, in particular to a cardiovascular tumor emergency monitoring and early warning system based on wearable equipment, which comprises a data acquisition module for acquiring multi-modal physiological data of a patient in real time; the information setting module is used for setting verification information between the standby physiological data time sequence and the actual physiological data time sequence; the data verification module is used for verifying the to-be-used physiological data based on the verification information and taking the to-be-used physiological data passing the verification as effective physiological data; and the monitoring and early warning module continuously acquires effective physiological data at the medical care terminal to evaluate emergency risk indexes, and performs early warning on the multi-modal physiological data corresponding to the emergency risk indexes exceeding a preset risk index threshold to the supervision terminal and / or the medical care terminal, so that the authenticity and reliability of the physiological data for analysis and early warning can be ensured, and the accuracy of analysis and early warning is improved. And the accuracy of emergency judgment is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of emergency monitoring, and in particular to a cardiovascular tumor emergency monitoring and early warning system based on wearable devices. Background Art

[0002] Cardiovascular cancer, a major disease that poses a serious threat to human health, is often sudden and dangerous. Without timely and effective monitoring and early warning of emergencies, patients face significant life risks. With the rapid advancement of technology, wearable devices, due to their portability and wearability, have become a crucial tool in healthcare monitoring. Wearable devices can integrate a variety of biosensors to collect multimodal physiological data such as heart rate, blood pressure, blood oxygen saturation, and body temperature in real time, providing a rich data source for monitoring cardiovascular cancer emergencies.

[0003] However, the current physiological data monitoring system based on wearable devices is susceptible to transmission delays or other conditions when transmitting the data collected by the wearable devices to the medical end, resulting in inaccurate transmitted data. It is difficult to ensure the authenticity and reliability of the physiological data used for analysis and early warning, which in turn affects the accuracy of emergency judgment.

[0004] To this end, we propose a cardiovascular tumor emergency monitoring and early warning system based on wearable devices to solve the above problems. Summary of the Invention

[0005] The purpose of the present invention is to provide a cardiovascular tumor emergency monitoring and early warning system based on a wearable device to solve the problems raised in the above background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a cardiovascular tumor emergency monitoring and early warning system based on a wearable device, comprising:

[0007] The data acquisition module integrates multimodal biosensors on wearable devices to collect multimodal physiological data of patients in real time to generate actual physiological data time series. It determines the abnormal cause corresponding to the abnormal data based on the patient's body position change information in the corresponding environment, and annotates the abnormal data in the actual physiological data time series based on the abnormal cause. The annotated actual physiological data time series is then transmitted between the monitoring end, the supervised end, and the medical end.

[0008] An information setting module obtains a time series of standby physiological data and a time series of actual physiological data; sets verification information between the time series of standby physiological data and the time series of actual physiological data, wherein the verification information includes multiple pairs of sliding windows and corresponding synchronization chains;

[0009] The data verification module verifies the pending physiological data based on the verification information, and treats the pending physiological data that passes the verification as valid physiological data; invalidates the pending physiological data that fails the verification, and re-acquires the actual physiological data of the supervised end for verification until the verification passes;

[0010] The monitoring and early warning module continuously obtains effective physiological data from the medical side to evaluate the emergency risk index, and warns the supervisory side and / or the medical side of the multimodal physiological data corresponding to the emergency risk index that exceeds the preset risk index threshold.

[0011] Preferably, the step of transmitting the patient's multimodal data between the monitoring end, the supervised end, and the medical end includes:

[0012] Collect the patient's identity information and register to get the supervised end; collect the patient's family's identity information and register to get the supervisory end; collect the patient's corresponding hospital's identity information and register to get the medical end;

[0013] The multimodal physiological data of the supervised end is transmitted from the supervised end to the supervisory end and the medical end, and early warning limit information corresponding to the supervisory end and the medical end is set respectively, wherein the early warning limit information is the abnormal level of the actual physiological data.

[0014] Preferably, the step of setting verification information between the standby physiological data time series and the actual physiological data time series, wherein the verification information includes multiple pairs of sliding windows and synchronization chains between each pair of sliding windows, comprises:

[0015] The actual physiological data of the monitored end is pre-processed and transmitted to the medical end to obtain multiple types of stand-by physiological data, and the actual physiological data time series and the stand-by physiological data time series are generated based on the collection time point of each type of physiological data;

[0016] A plurality of pairs of sliding windows are respectively set corresponding to the actual physiological data time series and the standby physiological data time series, and a synchronization chain is set between each pair of sliding windows;

[0017] Set the moving distance and selected sites for each pair of sliding windows, and assign a unique identity to each selected site. The last acquisition time point in the selected site corresponding to the previous sliding window is the first acquisition time point in the selected site corresponding to the current sliding window.

[0018] Multiple pairs of sliding windows and their corresponding synchronization chains are used as verification information.

[0019] Preferably, the step of pre-processing the actual physiological data of the monitored end and transmitting it to the medical end to obtain multiple types of physiological data to be used includes:

[0020] Acquire multimodal physiological data collected by the supervised end through the wearable device as multiple types of actual physiological data;

[0021] Preset a standard threshold corresponding to the actual physiological data; obtain the actual physiological data time series corresponding to each type of actual physiological data, and preliminarily mark the actual physiological data that does not meet the standard threshold in the actual physiological data time series as abnormal data to obtain preliminary marked data and the corresponding abnormal level;

[0022] The position information and body position change information of the supervised terminal at the collection time point corresponding to the preliminary marked data are obtained, and the preliminary marked data corresponding to the body position change information exceeding the preset conditions are classified and marked to obtain the actual physiological data time series after marking.

[0023] Preferably, the step of obtaining the position information and body position change information of the supervised terminal at the collection time point corresponding to the preliminary marked data, and classifying and marking the preliminary marked data corresponding to the body position change information exceeding the preset conditions to obtain the marked actual physiological data time series includes:

[0024] Obtaining a video image of the patient's body movements, obtaining the joint positions of the person's body movements in the video image, and constructing a virtual skeletal model of the patient based on the joint positions;

[0025] The patient's virtual skeleton model is labeled with limb names, and multiple identification points are set according to the patient's corresponding limb names. The change information of each identification point is monitored in real time, where the change information includes the change speed, change direction and change frequency;

[0026] Based on the change information of the identification points, the patient's body position change information is determined as the abnormal cause corresponding to the preliminary marking data and a secondary marking is performed;

[0027] The parameter difference between the preliminary marked data after secondary marking and the preset conditions is determined, and the corresponding abnormal level is matched based on the parameter difference to perform tertiary marking to obtain the actual physiological data time series after marking.

[0028] Preferably, the step of monitoring the change information of each identification point in real time, wherein the change information includes the change speed, change direction, and change frequency; and determining the patient's body position change information as the abnormal cause corresponding to the preliminary marking data based on the change information of the identification point and performing secondary marking comprises:

[0029] Collect the patient's environmental data in real time, including relative friction and relative tilt angle. The formula for calculating the ground friction coefficient μ is: Among them, P j,x represents the component of the jth pressure sensor on the x-axis, P j,yrepresents the component of the jth pressure sensor on the y-axis, P j,z represents the component of the j-th pressure sensor on the z-axis, m represents the number of sensors, and j represents the serial number of the sensor;

[0030] The body inclination angle θ is calculated by the three-axis accelerometer, and the corresponding formula is in, represents the accelerometer reading on the x-axis, Indicates the accelerometer reading on the y-axis, a z Represents the accelerometer reading on the z-axis;

[0031] Reconstruct the three-dimensional coordinates of the joint points. The joint coordinates are represented by Jk = (xk, yk, zk), where Jk represents the three-dimensional coordinates of the identification point corresponding to the k-th joint, and xk, yk, and zk represent the coordinate values ​​of the identification point corresponding to the joint on the x, y, and z axes, respectively. Calculate the change speed, change direction, and change frequency of the identification point.

[0032] Establish the correlation between environmental data, joint dynamic characteristics and physiological parameters, calculate the dynamic characteristic deviation, preset the deviation change threshold, compare the dynamic characteristic deviation with the preset threshold to determine whether the position change is the abnormal cause corresponding to the abnormal data, and determine the dynamic characteristic change information corresponding to the actual deviation exceeding the deviation change threshold as the position state change information exceeding the preset change index threshold. The calculation formula corresponding to the dynamic characteristic deviation is: Among them, ω1, ω2, and ω3 represent the weight coefficients corresponding to the change speed, change direction, and change frequency respectively, vi represents the change speed of the identification point corresponding to the current joint, and vz represents the average speed of the identification point corresponding to the current joint in the normal state. Indicates the motion direction vector of the identification point corresponding to the current joint, represents the mean direction vector of the identification point corresponding to the current joint in the normal state, fi represents the main frequency of the movement of the identification point corresponding to the current joint, and fz represents the mean main frequency of the identification point corresponding to the current joint in the normal state.

[0033] Preferably, the step of verifying the physiological data to be used based on the verification information includes:

[0034] Obtaining a time series of standby physiological data and an actual physiological data corresponding to the standby physiological data and the actual physiological data, respectively, and determining abnormal data sites in the time series of the actual physiological data as reference points, each reference point corresponding to at least two pairs of sliding windows;

[0035] Each pair of sliding windows is moved by the same moving distance based on the synchronization chain, and the standby physiological data corresponding to the selected site in the standby physiological data time series is compared to see whether it is consistent with the actual physiological data at the selected site in the other sliding window;

[0036] If they are consistent, the standby physiological data at the selected site is determined to have passed the verification. Otherwise, it is determined to have failed the verification, and the collection time point corresponding to the selected site is obtained and sent to the supervised end. The actual physiological data corresponding to the collection time point is re-obtained and replaced with the standby physiological data.

[0037] Preferably, the step of determining abnormal data sites in the actual physiological data time series as reference points includes:

[0038] Obtaining a collection time point corresponding to abnormal data in the actual physiological data time series, and determining whether the abnormal data at the collection time point is consistent with the abnormal data at the same collection time point in the standby physiological data time series;

[0039] If they are consistent, the collection time point is used as the reference point;

[0040] If they are inconsistent, the abnormal data corresponding to the collection time point are selected from the actual physiological data time series to replace the standby physiological data at the same collection time point, and the time point at which the abnormal data with continuous abnormalities starts and the time point at which the abnormal data ends are used as the reference points corresponding to the two pairs of sliding windows.

[0041] Preferably, the step of continuously acquiring effective physiological data at the medical end to evaluate the emergency risk index and alerting the supervisory end and / or the medical end of the multimodal physiological data corresponding to the emergency risk index exceeding a preset risk index threshold includes:

[0042] Continuously acquiring valid physiological data to generate a valid physiological data time series, wherein the valid physiological data includes valid physiological abnormal data and valid physiological normal data;

[0043] Acquiring annotation information based on the annotated valid physiological data time series, wherein the annotation information includes the abnormal location, abnormal cause, and abnormal degree of the abnormal data;

[0044] Determine abnormal data in the valid physiological data time series based on the annotation information, and set different emergency risk indexes according to the abnormal levels of different abnormal data;

[0045] The emergency risk index is obtained by evaluating the risk index based on the labeled information of multiple abnormal data;

[0046] Abnormal data that requires early warning will be alerted to both the supervisory and medical ends at the same time. The medical end will conduct a comprehensive risk assessment based on other valid physiological data, and exchange information with the supervisory end to formulate corresponding treatment measures for the supervised end.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] 1. By setting multiple pairs of sliding windows, synchronization chains between each pair of sliding windows, and the moving distance and selected locations of the sliding windows, each pair of sliding windows is moved synchronously according to a certain moving distance, and the actual physiological data at the same collection time point is selected for comparison with the stand-by physiological data, thereby verifying the accuracy of the stand-by physiological data obtained by the medical side, ensuring the accuracy of the stand-by physiological data obtained by the medical side, thereby ensuring the accuracy of subsequent emergency risk assessment, and improving the monitoring effect of cardiovascular cancer patients.

[0049] 2. By setting corresponding identification points for the patient's limbs and monitoring the correlation changes between the identification points, the patient's posture change information at the continuous time points where abnormal data occurs can be determined, thereby determining the abnormal cause corresponding to the abnormal data, and marking the abnormal data, abnormal cause and abnormal collection time point corresponding to the abnormal data in the actual physiological data time series. The marked content is transmitted to the medical end, which can reduce the analysis time of the medical end, thereby improving the efficiency of the medical end in analyzing the abnormal conditions of the patient's vital signs based on the physiological data, and improving the safety monitoring effect of the patient. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0051] Figure 1 This is a system structure diagram of the present invention. DETAILED DESCRIPTION

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0053] Example

[0054] See also Figure 1The present invention provides a technical solution for a cardiovascular tumor emergency monitoring and early warning system based on a wearable device: a cardiovascular tumor emergency monitoring and early warning system based on a wearable device, comprising:

[0055] The data acquisition module integrates multimodal biosensors on wearable devices to collect multimodal physiological data of patients in real time to generate actual physiological data time series. It determines the abnormal cause corresponding to the abnormal data based on the patient's body position change information in the corresponding environment, and annotates the abnormal data in the actual physiological data time series based on the abnormal cause. The annotated actual physiological data time series is then transmitted between the monitoring end, the supervised end, and the medical end.

[0056] The steps of transmitting the annotated actual physiological data time series between the supervisory end, the supervised end, and the medical end include: collecting the patient's identity information and registering it to obtain the supervised end, collecting the patient's family member's identity information and registering it to obtain the supervisory end, collecting the patient's corresponding hospital identity information and registering it to obtain the medical end, transmitting the multimodal physiological data of the supervised end from the supervised end to the supervisory end and the medical end, and setting warning limit information corresponding to the supervisory end and the medical end, respectively, wherein the warning limit information is the abnormality level of the actual physiological data;

[0057] Specifically, multimodal physiological data includes at least cardiovascular-related physiological data and tumor marker-related physiological data. Cardiovascular-related physiological data include but are not limited to heart rate, blood pressure, electrocardiogram waveform, blood oxygen saturation, and pulse wave transit time. Tumor marker-related physiological data is obtained through a microfluidic chip detection device integrated into the wearable device. The microfluidic chip detection device can specifically identify and detect the concentrations of at least three tumor markers related to cardiovascular tumors in the blood. The three tumor markers are carcinoembryonic antigen (CEA), carbohydrate antigen 125 (CA125), and cytokeratin 19 fragment (CYFRA21-1). For patients with acute cardiovascular tumors, wearable devices need to focus on monitoring parameters such as heart rate, blood pressure, SpO2, and respiratory rate, and issue warnings based on the following standards: Heart rate: Continuously too fast (≥120 beats / minute) or too slow (≤50 beats / minute). Blood pressure: Hypertension (≥180 / 110 mmHg) or hypotension (≤90 / 60 mmHg). Blood oxygen saturation ≤ 90% or rapidly decreasing ≥ 5%; respiratory rate: ≥ 25 breaths / minute or ≤ 10 breaths / minute. Multi-parameter combined monitoring and dynamic trend analysis can improve the early identification rate of cardiovascular and oncological emergencies, and data exceeding the corresponding standard thresholds are considered abnormal.

[0058] An information setting module obtains a time series of standby physiological data and a time series of actual physiological data; sets verification information between the time series of standby physiological data and the time series of actual physiological data, wherein the verification information includes multiple pairs of sliding windows and corresponding synchronization chains;

[0059] The steps of setting verification information between the standby physiological data time series and the actual physiological data time series, wherein the verification information includes multiple pairs of sliding windows and a synchronization chain between each pair of sliding windows, include: pre-processing the actual physiological data of the supervised end and transmitting it to the medical end to obtain multiple types of standby physiological data, generating the actual physiological data time series and the standby physiological data time series based on the collection time point of each type of physiological data, setting multiple pairs of sliding windows corresponding to the actual physiological data time series and the standby physiological data time series, and setting a synchronization chain between each pair of sliding windows; setting the moving distance and the selected site of each pair of sliding windows, and assigning a unique identity to each selected site, wherein the last collection time point in the selected site corresponding to the previous sliding window is the first collection time point in the selected site corresponding to the current sliding window; and using multiple pairs of sliding windows and the corresponding synchronization chains as verification information.

[0060] Specifically, by setting multiple pairs of sliding windows, synchronization chains between each pair of sliding windows, and the moving distance and selection sites of the sliding windows, based on the collection time point of the actual physiological data as the reference point, arbitrarily select the standby physiological data corresponding to the same time point for comparison with the actual physiological data, move each pair of sliding windows at the same time according to a certain moving distance, select the actual physiological data at the same time point for comparison with the standby physiological data, and when the data are consistent, it means that the verification is successful. Multiple selection sites are set in the sliding window, each selection site is given a unique identity, and each selection site corresponds to the position of a collection time point, that is, in the process of moving the sliding window, the distance between two adjacent selection sites corresponds to the distance between two adjacent collection time points. Multiple selection sites can be set in a sliding window to compare a section Whether the corresponding stand-by physiological data at the collection time point is consistent with the actual physiological data can facilitate the subsequent determination of the location of the inconsistent data when inconsistent data is monitored, and then retransmit the data at the collection time point to ensure the accuracy of subsequent risk assessment. The correspondence between the actual physiological data and the stand-by physiological data is guaranteed by the collection time point to prevent the data at the previous collection time point from being corresponded to the data at the current collection time point during verification. At least two pairs of sliding windows are set for each abnormal data in each actual physiological data time series, and a synchronization chain is set between each pair of sliding windows to ensure that the two sliding windows in each pair of sliding windows can move synchronously. The two pairs of sliding windows start to move toward each other from the position corresponding to the collection time point where the abnormal data is located, thereby improving the verification efficiency of the stand-by physiological data.

[0061] The steps of preprocessing the actual physiological data of the supervised end and transmitting it to the medical end to obtain multiple types of stand-by physiological data include: obtaining multimodal physiological data collected by the supervised end through a wearable device as multiple types of actual physiological data; presetting standard thresholds corresponding to the actual physiological data; obtaining actual physiological data time series corresponding to each type of actual physiological data, and preliminarily marking the actual physiological data that does not meet the standard threshold in the actual physiological data time series as abnormal data to obtain preliminary marked data and corresponding abnormality levels; obtaining the position information and body position change information of the supervised end at the collection time point corresponding to the preliminary marked data, and classifying and marking the preliminary marked data corresponding to the body position change information that exceeds the preset conditions to obtain the marked actual physiological data time series.

[0062] The steps of obtaining the position information and posture change information of the supervised end at the collection time point corresponding to the preliminary marking data, classifying and marking the preliminary marking data corresponding to the posture change information exceeding the preset conditions to obtain the marked actual physiological data time series include: obtaining a video image of the body movement of the patient at the supervised end, obtaining the joint positions of the person's body movement in the video image, constructing a patient virtual skeletal model according to the joint positions, labeling the limb names of the corresponding patient virtual skeletal model, setting multiple identification points according to the patients corresponding to the limb names, and monitoring the change information of each identification point in real time, wherein the change information includes the change speed, change direction and change frequency; determining the patient's posture change information as the abnormal cause corresponding to the preliminary marking data based on the change information of the identification point and performing secondary marking, determining the parameter difference between the preliminary marking data after the secondary marking and the preset conditions, performing tertiary marking based on the abnormal level matching the parameter difference, and obtaining the marked actual physiological data time series.

[0063] Real-time monitoring of change information of each identification point, wherein the change information includes change speed, change direction and change frequency; the steps of determining the patient's posture change information as the abnormal cause corresponding to the preliminary marking data based on the change information of the identification point and performing secondary marking include: real-time monitoring of the change information of each identification point, wherein the change information includes change speed, change direction and change frequency; establishing a characteristic model of the change speed, change direction and change frequency of each joint identification point corresponding to different posture states of the patient in different environments, calculating the similarity between the real-time collected identification point change information and the pre-established posture state model, and selecting the posture state model with the highest similarity as the initial posture change information; continuously collecting identification point change information and environmental information, updating the patient's posture change information in real time, and obtaining the posture state information corresponding to the preliminary marking data that exceeds a preset change index threshold as the abnormal cause corresponding to the preliminary marking data and performing secondary marking;

[0064] The specific content of the similarity calculation between the real-time collected recognition point change information and the pre-established body posture state model is as follows: the real-time collected recognition point change information constitutes a vector X = (x1, x2, ..., x n ), the pre-established posture state model constitutes a vector Y = (y1, y2, ..., y n ), the Euclidean distance d(X, Y) between them is calculated as: Among them, X represents the real-time collected identification point change information vector, Y represents the pre-established posture state model vector, and x i represents the i-th component of vector X, i.e. the i-th eigenvalue of the real-time acquired identification point change information, y i represents the i-th component of vector Y, representing the corresponding i-th eigenvalue in the body position state model, n represents the dimension of the vector, that is, the number of features of the recognition point change information, representing different features of the recognition point change information (such as change speed, change direction component, change frequency, etc.), d(X, Y) represents the Euclidean distance between vector X and vector Y, and a threshold T is set in advance. When d(X, Y) ≤ T, it is considered that the real-time collected recognition point change information is similar to the body position state model, that is, the patient may be in this body position state; when d(X, Y) > T, it is considered dissimilar.

[0065] Update the patient's body position change information in real time, and obtain the body position status information corresponding to the preliminary marked data that exceeds the preset change index threshold as the abnormal cause corresponding to the preliminary marked data and perform secondary marking. Specific content: preset the deviation change threshold, and use the dynamic characteristics corresponding to the actual deviation that exceeds the deviation change threshold as the abnormal cause corresponding to the abnormal data and perform secondary marking;

[0066] The steps of establishing a characteristic model of the change speed, change direction and change frequency of each joint identification point corresponding to different body postures of the patient in different environments include: collecting the patient's environmental data in real time, wherein the environmental data includes relative friction and relative tilt angle, and calculating the ground friction coefficient μ through a plantar pressure sensor array (such as an insole sensor) as follows: Among them, P j,x represents the component of the jth pressure sensor on the x-axis, P j,yRepresents the component of the j-th pressure sensor on the y-axis, Pj,z represents the component of the j-th pressure sensor on the z-axis, m represents the number of sensors, and j represents the serial number of the sensor. Wet ground (μ<0.3): The patient may excessively flex the knee joint due to insufficient friction. This gait adjustment will change the movement speed, direction, and frequency of the joint identification point, thereby increasing the deviation of the dynamic characteristics. Dry ground (μ>0.5): The patient may cause the ankle joint to invert due to excessive friction. This gait adjustment will change the movement characteristics of the joint identification point and increase the deviation of the dynamic characteristics. The body inclination angle (θ) is calculated by the three-axis accelerometer, and the corresponding formula is in, represents the accelerometer reading on the x-axis, Represents the accelerometer reading on the y-axis, a z Represents the reading of the accelerometer on the z-axis. Uphill (θ>10°): The patient may increase the flexion angle of the hip joint due to tilt. Downhill (θ<-10°): The patient may increase the load on the knee joint due to tilt. Gait changes will change the motion trajectory and speed of the joint identification point, thereby affecting the dynamic feature deviation; reconstruct the three-dimensional joint point coordinates, the joint coordinates are represented by Jk=(xk, yk, zk), Jk represents the three-dimensional coordinates of the identification point corresponding to the kth joint, xk, yk, zk represent the coordinate values ​​of the identification point corresponding to the joint on the x, y, and z axes respectively, calculate the change speed, change direction and change frequency of the identification point; establish the correlation between environmental data, joint dynamic characteristics and physiological parameters, calculate the dynamic feature deviation, preset the deviation change threshold, compare the dynamic feature deviation with the preset threshold to determine whether the posture change is the abnormal cause corresponding to the abnormal data, and determine the dynamic feature change information corresponding to the actual deviation that exceeds the deviation change threshold as the posture state change information that exceeds the preset change index threshold. The calculation formula corresponding to the dynamic feature deviation is Among them, ω1, ω2, and ω3 represent the weight coefficients corresponding to the change speed, change direction, and change frequency respectively, vi represents the change speed of the identification point corresponding to the current joint, and vz represents the average speed of the identification point corresponding to the current joint in the normal state. Indicates the motion direction vector of the identification point corresponding to the current joint, The ground friction coefficient indirectly affects the movement speed and direction of the joint identification points by affecting the patient's gait stability, thereby increasing the dynamic feature deviation. The body inclination angle affects the movement frequency and speed of the joint identification points by changing the patient's walking posture and joint load, which also increases the dynamic feature deviation. By comparing the current dynamic feature deviation with the preset threshold, it is possible to determine whether the patient's body position change is caused by abnormal environmental data (such as slippery ground and steep slope), and further determine whether the abnormal data is caused by the body position change brought about by the abnormal environmental data. For example, slippery ground + abnormal increase in knee joint speed + sharp increase in heart rate → slip risk triggers a stress response, downhill + downward tilt of the ankle joint + decreased blood oxygen saturation → abnormal joint pressure distribution leads to physiological stress. Through three-dimensional analysis of speed, direction, and frequency, the abnormal body position and abnormal physiological parameters can be traced, and the patient's emergency risk index can be further analyzed based on the body position status information.

[0067] Specifically, when the speed change of the identification point exceeds the set fall threshold and the environmental acceleration fluctuates violently in an instant, the patient is judged to be in a fall state; when the speed change of the identification point fluctuates within a certain range and the environmental data is relatively stable, the patient is judged to be in a state of movement based on the historical movement pattern. Based on the state judgment result, the cause of the abnormality of the collected physiological data is marked. If it is judged to be an abnormality caused by a fall, the physiological data at the corresponding time point will be marked as "fall abnormality"; if it is judged to be a change in physiological data caused by exercise, the type of exercise and intensity and other information will be further marked.

[0068] By marking the abnormal data collected by wearable devices and transmitting the marked information to the medical end during transmission, the medical end can conduct a preliminary analysis of the patient's condition based on the abnormal data in the physiological parameters and provide a preliminary solution. While implementing the preliminary solution, it can further provide a subsequent treatment plan by comprehensively analyzing other effective physiological data to be used. This allows the medical end to formulate corresponding treatment plans based on physiological data more quickly, without having to browse all of them before providing a preliminary plan, thereby improving the safety monitoring effect of patients. Existing medical ends generally need to obtain all physiological parameters first and then find abnormal data for analysis when conducting an evaluation. Here, the abnormal data, abnormal causes and abnormal collection time points corresponding to the abnormal data are directly marked in the actual physiological data time series, and the marked content is transmitted to the medical end. This can reduce the analysis time of the medical end, thereby improving the efficiency of the medical end in analyzing the abnormal conditions of the patient's vital signs based on the physiological data, thereby improving the safety monitoring effect of the patient.

[0069] Abnormal data here refers to data that exceeds the standard threshold as abnormal data, but it does not mean that it is not actual physiological data. It just means that this parameter exceeds the patient's normal standard threshold. Therefore, it refers to abnormal data when various parameters of physiological data are abnormal, that is, when the patient is prone to symptoms, the corresponding physiological data is abnormal data;

[0070] Specifically, multimodal physiological data collected by wearable devices at the supervised end (patient) in real time is obtained, and then classified according to the type of physiological data, such as heart rate as one category and blood pressure as another category. For each type of physiological data, an actual physiological data time series is generated according to the collection time point, and the standard threshold corresponding to the preset actual physiological data is screened out. At the same time, the abnormal data is marked with an abnormal level according to the degree of non-satisfaction, and the marked data is further marked according to the cause of the abnormal data, that is, the abnormal position, abnormal level and abnormal cause of the abnormal data are marked on the actual physiological data time series. When determining the abnormal cause of the abnormal data, the body movement video image of the patient at the supervised end is monitored to obtain the joint position in the person's body movement video image, and a patient virtual model is constructed according to the joint position. The limb name is marked on the corresponding patient virtual model, and multiple identification points are set for the patient corresponding to the limb name (the identification point corresponds to the point set for the direction of each limb part), and corresponding restriction conditions are formulated for the patient virtual model corresponding to the limb name. The restriction conditions include the corresponding patient virtual model based on the limb name. The distribution positions of multiple identification points in the model and the change speed, direction and frequency of the identification points. For example, when a patient is running, the legs and limbs will perform the same action in a cycle, and when exercising, the changes of the limbs will have certain rules. Here, the change speed, direction and frequency of an identification point are limited, which can identify whether the supervised end is performing a certain exercise and whether the change of its physiological data is caused by exercise. After collecting the status information of the supervised end, the status information is transmitted to the management center corresponding to the wearable device. The management center marks the abnormal causes of the physiological data according to the patient's status. It can also be used to monitor whether the patient has fallen. When a fall occurs, the speed of the identification point will change rapidly. When the abnormal data is caused by a fall, the physiological data corresponding to the collection time point will be marked as caused by a fall, thereby marking the physiological data according to the abnormal cause. It can facilitate the subsequent medical care end to analyze the patient's historical physiological data according to the abnormal cause, so as to better analyze the patient's physical condition and formulate a corresponding treatment plan.

[0071] The data verification module verifies the pending physiological data based on the verification information, and treats the pending physiological data that passes the verification as valid physiological data; invalidates the pending physiological data that fails the verification, and re-acquires the actual physiological data of the supervised end for verification until the verification passes;

[0072] The step of verifying the standby physiological data based on the verification information includes: obtaining the standby physiological data time series and the actual physiological data time series corresponding to the standby physiological data and the actual physiological data respectively, determining the abnormal data site in the actual physiological data time series as a reference point, each reference point corresponds to at least two pairs of sliding windows, moving each pair of sliding windows according to the same moving distance based on the synchronization chain, and comparing whether the standby physiological data corresponding to the selected site in the standby physiological data time series is consistent with the actual physiological data at the selected site in another sliding window. If they are consistent, it is determined that the standby physiological data at the selected site is verified. Otherwise, it is determined that the verification has failed, and the collection time point corresponding to the selected site is obtained and sent to the supervised end, and the actual physiological data corresponding to the collection time point is re-obtained and replaced with the standby physiological data.

[0073] The step of determining the abnormal data site in the actual physiological data time series as the reference point includes: obtaining the collection time point corresponding to the abnormal data in the actual physiological data time series, and judging whether the abnormal data at the collection time point is consistent with the abnormal data at the same collection time point in the stand-by physiological data time series; if they are consistent, the collection time point is used as the reference point; if they are inconsistent, the abnormal data corresponding to the collection time point is selected from the actual physiological data time series to replace the stand-by physiological data at the same collection time point; the time point at which the abnormal data with continuous abnormalities starts and the time point at which the abnormal data ends are respectively used as the reference points corresponding to the two pairs of sliding windows.

[0074] Specifically, on the standby physiological data, abnormal data is determined, and the corresponding selected site is determined as the reference point based on the abnormal data, and then the position of the reference point is used as the sliding starting point of the sliding window. The two pairs of sliding windows are moved back to back, and the standby physiological data are verified to be consistent with the actual physiological data. When they are consistent, they are directly used as the valid standby data on the medical side. When they are inconsistent, the actual physiological data corresponding to the standby physiological data is determined according to the collection time point corresponding to the selected site, and the inconsistent standby physiological data is re-acquired, and the inconsistent standby physiological data is destroyed at the same time. This is used to judge whether the standby physiological data is valid, that is, whether it is abnormal or not. It is necessary to ensure that the physiological data to be used is consistent with the actual physiological data at the corresponding collection time point in order to ensure the validity of the physiological data to be used; before setting the benchmark point, it is necessary to first determine whether the abnormal data in the physiological data to be used is consistent with the abnormal data in the actual physiological data, and use the consistent data as the benchmark point. If they are inconsistent, the abnormal data of the corresponding collection time point will be re-obtained from the actual physiological data, and the abnormal data in the physiological data to be used will be replaced and destroyed. This can be used to ensure that the physiological data to be used obtained by the medical end is completely consistent with the actual physiological data of the supervised end, thereby ensuring the accuracy of subsequent emergency risk assessment and improving the monitoring effect of patients.

[0075] The monitoring and early warning module continuously obtains effective physiological data from the medical side to evaluate the emergency risk index, and warns the supervisory side and / or the medical side of the multimodal physiological data corresponding to the emergency risk index that exceeds the preset risk index threshold.

[0076] The steps of continuously acquiring effective physiological data at the medical end to evaluate the emergency risk index, and alerting the multimodal physiological data corresponding to the emergency risk index that exceeds the preset risk index threshold to the supervision end and / or the medical end include: continuously acquiring effective physiological data to generate an effective physiological data time series, wherein the effective physiological data includes effective physiological abnormal data and effective physiological normal data; acquiring annotation information based on the annotated effective physiological data time series, wherein the annotation information includes the abnormal position, abnormal cause and abnormal degree of the abnormal data; determining the abnormal data in the effective physiological data time series according to the annotation information, setting different emergency risk indices according to the abnormal levels of different abnormal data, and performing risk index assessment based on the annotation information of multiple abnormal data to obtain the emergency risk index; alerting the abnormal data that needs to be warned to the supervision end and the medical end at the same time, and the medical end performs a comprehensive risk assessment based on other effective physiological data, and exchanges information with the supervision end to formulate treatment measures corresponding to the supervised end.

[0077] The process of information exchange between the medical and nursing ends and the regulatory ends includes data review, discussion of risk assessment results, and input and confirmation of treatment methods. The comprehensive risk assessment results of the medical and nursing ends are transmitted to the regulatory end, and corresponding treatment methods are given. The regulatory end makes specific decisions based on the treatment methods given by the medical and nursing ends, determines whether to implement the corresponding treatment methods, and thus jointly formulates the corresponding treatment methods for the regulated end.

[0078] The risk index is evaluated based on the labeled information of multiple abnormal data to obtain the formula corresponding to the specific content of the emergency risk index RI. Among them, ω i It represents the weight of the abnormal cause corresponding to the i-th abnormal data, reflecting the contribution of the abnormal cause to the risk of cardiovascular tumor emergency. l represents the number of abnormal data within the effective time, i represents the sequence number of the abnormal data, and λ i represents the abnormality weight of the i-th abnormal data, indicating the severity of the abnormal data; T i represents the continuous duration of the i-th abnormal data (unit: hours / minutes, etc., determined according to the actual data collection frequency), T 总 Indicates the total duration of the effective time (unit is the same as T i same), It represents the proportion of the i-th abnormal data in the effective time, reflecting its persistent impact. a represents the basic risk value: a preset constant used to adjust the baseline level of the overall risk index, which can be set based on clinical experience and data verification.

[0079] Specifically, since the actual physiological data time series is marked, after the physiological data time series to be used is verified, it can be guaranteed that the entire sequence of the physiological data time series to be used is valid, and the abnormal data in the actual physiological data is marked. During transmission, the marked information needs to be synchronously transmitted to the corresponding medical end. The marked information includes the abnormal position of the abnormal data (the position of the collection time point corresponding to the abnormal data), the cause of the abnormality and the degree of abnormality; the abnormal data in the effective physiological data time series is determined according to the marked information, and different emergency risk indexes are corresponding to different abnormal levels of abnormal data. The emergency risk index of multiple abnormal data within the effective time is preferentially evaluated, involving the continuous duration of the abnormal data, the cause of the abnormality and the degree of the abnormality. The risk index is evaluated according to the marked information, the risk index is determined, and then an early warning is issued. The data that needs to be warned is warned to the supervision end and the medical end at the same time. The medical end then combines other effective physiological data for a comprehensive risk assessment, and then interacts with the supervision end to formulate corresponding treatment measures.

[0080] The patient's physiological data is collected, stored in the management center corresponding to the wearable device, and transmitted to the medical end in real time. The data received by the medical end is compared with the data of the management center to ensure the accuracy of the data received by the medical end. During the comparison process, abnormal data is marked in the management center and classified according to the abnormal cause (sports, symptoms, etc.). The accuracy of the marked points is compared using a sliding window (window, synchronization chain for verification), and the marked points are used as a reference to expand outward on both sides for piece-by-piece comparison until all data comparison results are obtained. Normal data is retained, and the interval corresponding to the abnormal data is determined, so as to determine the location of the abnormal data and perform abnormal data analysis. It is updated to ensure the accuracy of the physiological data used by the medical staff in analyzing and evaluating the risk index, thereby improving the accuracy and timeliness of emergency judgments. By marking the abnormal data collected by wearable devices, the marked information is transmitted to the medical staff during transmission. The medical staff can conduct a preliminary analysis of the patient's condition based on the abnormal data in the physiological parameters and provide a preliminary solution. While implementing the preliminary solution, other effective physiological data to be used are comprehensively analyzed to further provide subsequent treatment plans. The medical staff can formulate corresponding treatment plans based on the physiological data more quickly, without having to browse all the data before providing a preliminary plan, thereby improving the safety monitoring effect of the patients.

[0081] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0082] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. Cardiovascular tumor emergency monitoring and early warning system based on wearable devices, characterized by: include: The data acquisition module integrates multimodal biosensors on wearable devices to collect multimodal physiological data of patients in real time to generate actual physiological data time series. It also determines the abnormal causes corresponding to abnormal data based on the patient's body position change information in the corresponding environment, and annotates the abnormal data in the actual physiological data time series based on the abnormal causes. The annotated actual physiological data time series is transmitted between the supervisory end, the supervised end, and the medical end; An information setting module obtains a time series of physiological data to be used and a time series of actual physiological data; Setting verification information between the standby physiological data time series and the actual physiological data time series, wherein the verification information includes multiple pairs of sliding windows and corresponding synchronization chains; A data verification module verifies the to-be-used physiological data based on the verification information, and takes the to-be-used physiological data that passes the verification as valid physiological data; The unused physiological data that fails the verification will be invalidated, and the actual physiological data of the supervised terminal will be re-obtained for verification until the verification passes; The monitoring and early warning module continuously obtains effective physiological data from the medical side to evaluate the emergency risk index, and warns the supervisory side and / or the medical side of the multimodal physiological data corresponding to the emergency risk index that exceeds the preset risk index threshold.

2. The wearable device-based cardiovascular tumor emergency monitoring and early warning system according to claim 1, characterized in that: The steps for transmitting the patient's multimodal data between the supervisor, the supervised, and the medical staff include: Collect the patient's identity information and register to get the supervised end; collect the patient's family's identity information and register to get the supervisory end; collect the patient's corresponding hospital's identity information and register to get the medical end; The multimodal physiological data of the supervised end is transmitted from the supervised end to the supervisory end and the medical end, and early warning limit information corresponding to the supervisory end and the medical end is set respectively, wherein the early warning limit information is the abnormal level of the actual physiological data.

3. The cardiovascular tumor emergency monitoring and early warning system based on a wearable device according to claim 1, characterized in that: The step of setting verification information between the standby physiological data time series and the actual physiological data time series, wherein the verification information includes multiple pairs of sliding windows and synchronization chains between each pair of sliding windows, comprises: The actual physiological data of the monitored end is pre-processed and transmitted to the medical end to obtain multiple types of stand-by physiological data, and the actual physiological data time series and the stand-by physiological data time series are generated based on the collection time point of each type of physiological data; A plurality of pairs of sliding windows are respectively set corresponding to the actual physiological data time series and the standby physiological data time series, and a synchronization chain is set between each pair of sliding windows; Set the moving distance and selected sites for each pair of sliding windows, and assign a unique identity to each selected site. The last acquisition time point in the selected site corresponding to the previous sliding window is the first acquisition time point in the selected site corresponding to the current sliding window. Multiple pairs of sliding windows and their corresponding synchronization chains are used as verification information.

4. The wearable device-based cardiovascular tumor emergency monitoring and early warning system according to claim 1, characterized in that: The steps of pre-processing the actual physiological data of the monitored terminal and transmitting the data to the medical terminal to obtain multiple types of physiological data to be used include: Acquire multimodal physiological data collected by the supervised end through the wearable device as multiple types of actual physiological data; Preset a standard threshold corresponding to the actual physiological data; obtain the actual physiological data time series corresponding to each type of actual physiological data, and preliminarily mark the actual physiological data that does not meet the standard threshold in the actual physiological data time series as abnormal data to obtain preliminary marked data and the corresponding abnormal level; The position information and body position change information of the supervised terminal at the collection time point corresponding to the preliminary marked data are obtained, and the preliminary marked data corresponding to the body position change information exceeding the preset conditions are classified and marked to obtain the actual physiological data time series after marking.

5. The cardiovascular tumor emergency monitoring and early warning system based on a wearable device according to claim 4 is characterized by: The steps of obtaining the position information and body position change information of the supervised terminal at the collection time point corresponding to the preliminary marked data, and classifying and marking the preliminary marked data corresponding to the body position change information exceeding the preset conditions to obtain the marked actual physiological data time series include: Obtaining a video image of the patient's body movements, obtaining the joint positions of the person's body movements in the video image, and constructing a virtual skeletal model of the patient based on the joint positions; The patient's virtual skeleton model is labeled with limb names, and multiple identification points are set according to the patient's corresponding limb names. The change information of each identification point is monitored in real time, where the change information includes the change speed, change direction and change frequency; Based on the change information of the identification points, the patient's body position change information is determined as the abnormal cause corresponding to the preliminary marking data and a secondary marking is performed; The parameter difference between the preliminary marked data after secondary marking and the preset conditions is determined, and the corresponding abnormal level is matched based on the parameter difference to perform tertiary marking to obtain the actual physiological data time series after marking.

6. The cardiovascular tumor emergency monitoring and early warning system based on a wearable device according to claim 5, characterized in that: The step of monitoring the change information of each identification point in real time, wherein the change information includes the change speed, change direction, and change frequency; determining the patient's body position change information as the abnormal cause corresponding to the preliminary marking data based on the change information of the identification point and performing secondary marking includes: Real-time monitoring of the change information of each identification point, where the change information includes the change speed, change direction and change frequency; Establish a characteristic model of the change speed, change direction and change frequency of each joint identification point corresponding to different body postures of the patient in different environments, calculate the similarity between the real-time collected identification point change information and the pre-established body posture state model, and select the body posture state model with the highest similarity as the initial body posture change information; Continuously collect identification point change information and environmental information, update the patient's body position change information in real time, and obtain the body position status information corresponding to the preliminary marking data that exceeds the preset change index threshold as the abnormal cause corresponding to the preliminary marking data and perform secondary marking.

7. The wearable device-based cardiovascular tumor emergency monitoring and early warning system according to claim 6, characterized in that: The step of establishing a characteristic model of the change speed, change direction and change frequency of each joint identification point corresponding to different body postures of the patient in different environments includes: Collect the patient's environmental data in real time, including relative friction and relative tilt angle. The formula for calculating the ground friction coefficient μ is: Among them, P j,x represents the component of the jth pressure sensor on the x-axis, P j,y represents the component of the jth pressure sensor on the y-axis, P j,z represents the component of the j-th pressure sensor on the z-axis, m represents the number of sensors, and j represents the serial number of the sensor; The body inclination angle θ is calculated by the three-axis accelerometer, and the corresponding formula is in, represents the accelerometer reading on the x-axis, Represents the accelerometer reading on the y-axis, a z Represents the accelerometer reading on the z-axis; Reconstruct the three-dimensional coordinates of the joint points. The joint coordinates are represented by Jk = (xk, yk, zk), where Jk represents the three-dimensional coordinates of the identification point corresponding to the k-th joint, and xk, yk, and zk represent the coordinate values ​​of the identification point corresponding to the joint on the x, y, and z axes, respectively. Calculate the change speed, change direction, and change frequency of the identification point. Establish the correlation between environmental data, joint dynamic characteristics and physiological parameters, calculate the dynamic characteristic deviation, preset the deviation change threshold, compare the dynamic characteristic deviation with the preset threshold to determine whether the position change is the abnormal cause corresponding to the abnormal data, and determine the dynamic characteristic change information corresponding to the actual deviation exceeding the deviation change threshold as the position state change information exceeding the preset change index threshold. The calculation formula corresponding to the dynamic characteristic deviation is: Among them, ω1, ω2, and ω3 represent the weight coefficients corresponding to the change speed, change direction, and change frequency respectively, vi represents the change speed of the identification point corresponding to the current joint, and vz represents the average speed of the identification point corresponding to the current joint in the normal state. Indicates the motion direction vector of the identification point corresponding to the current joint, represents the mean direction vector of the identification point corresponding to the current joint in the normal state, fi represents the main frequency of the movement of the identification point corresponding to the current joint, and fz represents the mean main frequency of the identification point corresponding to the current joint in the normal state.

8. The cardiovascular tumor emergency monitoring and early warning system based on a wearable device according to claim 1 is characterized by: The step of verifying the physiological data to be used based on the verification information includes: Obtaining a time series of standby physiological data and an actual physiological data corresponding to the standby physiological data and the actual physiological data, respectively, and determining abnormal data sites in the time series of the actual physiological data as reference points, each reference point corresponding to at least two pairs of sliding windows; Each pair of sliding windows is moved by the same moving distance based on the synchronization chain, and the standby physiological data corresponding to the selected site in the standby physiological data time series is compared to see whether it is consistent with the actual physiological data at the selected site in the other sliding window; If they are consistent, the standby physiological data at the selected site is determined to have passed the verification. Otherwise, it is determined to have failed the verification, and the collection time point corresponding to the selected site is obtained and sent to the supervised end. The actual physiological data corresponding to the collection time point is re-obtained and replaced with the standby physiological data.

9. The cardiovascular tumor emergency monitoring and early warning system based on a wearable device according to claim 8, characterized in that: The step of determining abnormal data sites in the actual physiological data time series as reference points includes: Obtaining a collection time point corresponding to abnormal data in the actual physiological data time series, and determining whether the abnormal data at the collection time point is consistent with the abnormal data at the same collection time point in the standby physiological data time series; If they are consistent, the collection time point is used as the reference point; If they are inconsistent, the abnormal data corresponding to the collection time point are selected from the actual physiological data time series to replace the standby physiological data at the same collection time point, and the time point at which the abnormal data with continuous abnormalities starts and the time point at which the abnormal data ends are used as the reference points corresponding to the two pairs of sliding windows.

10. The cardiovascular tumor emergency monitoring and early warning system based on a wearable device according to claim 1, characterized in that: The steps of continuously acquiring effective physiological data at the medical end to evaluate the emergency risk index and alerting the supervisory end and / or the medical end of the multimodal physiological data corresponding to the emergency risk index exceeding a preset risk index threshold include: Continuously acquiring valid physiological data to generate a valid physiological data time series, wherein the valid physiological data includes valid physiological abnormal data and valid physiological normal data; Acquiring annotation information based on the annotated valid physiological data time series, wherein the annotation information includes the abnormal location, abnormal cause, and abnormal degree of the abnormal data; Determine abnormal data in the valid physiological data time series based on the annotation information, and set different emergency risk indexes according to the abnormal levels of different abnormal data; The emergency risk index is obtained by evaluating the risk index based on the labeled information of multiple abnormal data; Abnormal data that requires early warning will be alerted to both the supervisory and medical ends at the same time. The medical end will conduct a comprehensive risk assessment based on other valid physiological data, and exchange information with the supervisory end to formulate corresponding treatment measures for the supervised end.

Images