Cloud data fusion platform for medical wearable devices

Abstract

The present application provides a cloud data fusion platform for medical wearable devices, which relates to the field of data processing technology, including: an indicator interference analysis module; a medical monitoring data division module, which is used to divide the medical monitoring data set into a trusted and abnormal monitoring data set; an abnormal monitoring data compensation module, which is used to predict abnormal monitoring data based on the trusted monitoring data set, and map and compensate the abnormal monitoring data set based on the predicted abnormal data set; a health status assessment module, which is used to input a standard environmental data set, a standard motion data set, and a standard monitoring data set into a health assessment model, and output a health status assessment result. This application can solve the technical problem that the accuracy and stability of the monitoring data are poor due to the fact that the vital sign monitoring data is easily interfered by external factors, resulting in low accuracy and reliability of the user's health status assessment, and can achieve the effect of accurately assessing the user's health status and reducing the probability of misdiagnosis.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a cloud data fusion platform for medical wearable devices. Background Art

[0002] Portable medical wearable devices are wearable electronic devices designed to monitor and record the user's physiological and motion data for remote monitoring and health analysis. Portable medical wearable devices have a wide range of applications in the fields of telemedicine, health monitoring and preventive medicine, helping users better understand their health status and providing valuable data support for medical professionals.

[0003] At present, the existing methods for monitoring and evaluating user health status based on medical wearable devices have technical problems such as poor accuracy and stability of monitoring data due to the fact that vital sign monitoring data are easily interfered by external factors, resulting in low accuracy and reliability of user health status assessment. Summary of the invention

[0004] The purpose of this application is to provide a cloud data fusion platform for medical wearable devices to solve the technical problem that the existing method of monitoring and evaluating user health status based on medical wearable devices is easy to be interfered by external factors, resulting in poor accuracy and stability of monitoring data, resulting in low accuracy and reliability of user health status assessment.

[0005] In view of the above problems, the present application provides a cloud data fusion platform for medical wearable devices, including: a data monitoring module, which is used to read the monitoring data of a medical acquisition module, an environmental acquisition module and a motion acquisition module respectively under a preset monitoring node, obtain a medical monitoring data set, an environmental monitoring data set and a motion monitoring data set, and transmit the medical monitoring data set, the environmental monitoring data set and the motion monitoring data set to a cloud analysis platform, wherein the medical acquisition module and the medical wearable device are communicatively connected, and the medical wearable device has a device ID; an indicator interference analysis module is used to pre-process the environmental monitoring data set and the motion monitoring data set according to a data pre-processing strategy in the cloud analysis platform to obtain a standard environmental data set and a standard motion data set, and transmit the standard environmental data set and the standard motion data set to a cloud analysis platform. The data set is input into an indicator interference database for matching to obtain multiple indicator credibility coefficients; a medical monitoring data division module is used to obtain predetermined coefficient constraints, and divide the medical monitoring data set into a credible monitoring data set and an abnormal monitoring data set according to multiple indicator credibility coefficients and predetermined coefficient constraints; an abnormal monitoring data compensation module is used to predict abnormal monitoring data based on the credible monitoring data set based on indicator mapping association to obtain a predicted abnormal data set, and map and compensate the abnormal monitoring data set based on the predicted abnormal data set, and generate a standard monitoring data set after fusion; a health status assessment module is used to call an associated health assessment model based on a device ID, input the standard environment data set, the standard motion data set and the standard monitoring data set into the associated health assessment model, and output a user health status assessment result.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] By reading the monitoring data of the medical acquisition module, the environment acquisition module and the motion acquisition module respectively under the preset monitoring node, a medical monitoring data set, an environmental monitoring data set and a motion monitoring data set are obtained, and the medical monitoring data set, the environmental monitoring data set and the motion monitoring data set are transmitted to the cloud analysis platform, wherein the medical acquisition module is communicatively connected with the medical wearable device, and the medical wearable device has a device ID identification; then, in the cloud analysis platform, the environmental monitoring data set and the motion monitoring data set are preprocessed according to the data preprocessing strategy to obtain a standard environment data set and a standard motion data set, and the standard environment data set and the standard motion data set are input into the indicator interference data The medical monitoring data set is matched with the database to obtain multiple indicator credibility coefficients; the predetermined coefficient constraints are further obtained, and the medical monitoring data set is divided into a credible monitoring data set and an abnormal monitoring data set according to the multiple indicator credibility coefficients and the predetermined coefficient constraints; then based on the indicator mapping association, the abnormal monitoring data is predicted according to the credible monitoring data set to obtain a predicted abnormal data set, and the abnormal monitoring data set is mapped and compensated according to the predicted abnormal data set, and a standard monitoring data set is generated after fusion; finally, the associated health assessment model is called based on the device ID, the standard environment data set, the standard motion data set and the standard monitoring data set are input into the associated health assessment model, and the user health status assessment result is output. That is to say, by collecting the medical monitoring data set, environmental monitoring data set and motion monitoring data set of the target user, and then performing indicator interference analysis based on the environmental monitoring data set and the motion monitoring data set, the indicator credibility coefficients of multiple monitoring indicators are obtained; then the medical monitoring data with a credibility coefficient greater than the predetermined coefficient constraint is divided into credible monitoring data, and the medical monitoring data with a credibility coefficient less than or equal to the predetermined coefficient constraint is divided into abnormal monitoring data, and a credible monitoring data set and an abnormal monitoring data set are obtained; then based on the mapping association between indicators, the abnormal monitoring data is predicted according to the credible monitoring data set, and the predicted abnormal data set is mapped and compensated for the abnormal monitoring data set, and a standard monitoring data set is generated after fusion; finally, the standard environmental data set, the standard motion data set and the standard monitoring data set are input into the associated health assessment model for evaluation, and the user health status assessment result is obtained. By determining the credible monitoring data and the abnormal monitoring data based on the external interference factor analysis, and using the credible monitoring data to perform targeted mapping correction on the abnormal monitoring data, the accuracy and reliability of the vital sign monitoring data can be improved, thereby achieving the technical effect of accurately assessing the user's health status and reducing the probability of misdiagnosis.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0010] Figure 1 This is a schematic diagram of the structure of a cloud data fusion platform for medical wearable devices in this application;

[0011] Figure 2 A schematic diagram of a process for generating an indicator interference database in a cloud data fusion platform for medical wearable devices in this application;

[0012] Description of reference numerals:

[0013] 11 data monitoring module, 12 indicator interference analysis module, 13 medical monitoring data division module, 14 abnormal monitoring data compensation module, 15 health status assessment module. DETAILED DESCRIPTION

[0014] This application provides a cloud data fusion platform for medical wearable devices, which solves the existing method of monitoring and evaluating the health status of users based on medical wearable devices. Because the vital sign monitoring data is easily interfered by external factors, the accuracy and stability of the monitoring data are poor, resulting in low accuracy and reliability of the user's health status evaluation. By determining the credible monitoring data and abnormal monitoring data based on external interference factor analysis, and using the credible monitoring data to perform targeted mapping and correction on the abnormal monitoring data, the accuracy and reliability of the vital sign monitoring data can be improved, thereby achieving the technical effect of accurately evaluating the user's health status and reducing the probability of misdiagnosis.

[0015] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0016] For examples, please see the attached Figure 1 , the present application provides a cloud data fusion platform for medical wearable devices, including:

[0017] The data monitoring module 11 is used to read the monitoring data of the medical acquisition module, the environmental acquisition module and the motion acquisition module respectively under the preset monitoring node, obtain the medical monitoring data set, the environmental monitoring data set and the motion monitoring data set, and transmit the medical monitoring data set, the environmental monitoring data set and the motion monitoring data set to the cloud analysis platform, wherein the medical acquisition module and the medical wearable device are communicatively connected, and the medical wearable device has a device ID.

[0018] Specifically, first, under the preset monitoring node, the monitoring data of the medical acquisition module, the environmental acquisition module and the motion acquisition module are read respectively to obtain the medical monitoring data set, the environmental monitoring data set and the motion monitoring data set; wherein, the preset monitoring node is a data acquisition time node, which can be set according to actual needs, wherein the medical acquisition module is connected in communication with the medical wearable device, and is used to obtain the vital sign monitoring data of the target user collected by the medical wearable device, the medical wearable device is a portable medical wearable device, such as: smart watch, health monitoring belt, etc., and the medical wearable device has a device ID identification, wherein the device ID is unique, and the vital sign monitoring data includes body temperature, heart rate, blood pressure, blood oxygen saturation and other data. The environmental acquisition module is used to collect the environmental interference factors of the user's current location, and the environmental interference factors refer to the factors that affect the accuracy of the vital sign monitoring data, such as: temperature, humidity, lighting conditions, electromagnetic intensity, etc., and the environmental acquisition module is connected in communication with the environmental monitoring array, wherein the environmental monitoring array includes temperature sensors, humidity sensors and other devices, which can be set according to actual conditions. The motion collection module is used to collect the user's current motion status information, including data such as motion type, exercise intensity, and exercise time. Different motion states will interfere with the accuracy of vital sign monitoring data.

[0019] The medical monitoring data set, environmental monitoring data set and motion monitoring data set are then encrypted and transmitted to the cloud analysis platform to protect user privacy and data security, wherein an adaptive encryption transmission algorithm can be selected according to the actual application scenario, such as using a key for data transmission.

[0020] The indicator interference analysis module 12 is used to preprocess the environmental monitoring data set and the motion monitoring data set in the cloud analysis platform according to the data preprocessing strategy to obtain a standard environmental data set and a standard motion data set, and input the standard environmental data set and the standard motion data set into the indicator interference database for matching to obtain multiple indicator credibility coefficients.

[0021] Specifically, in the cloud analysis platform, first, the environmental monitoring data set and the motion monitoring data set are preprocessed according to the data preprocessing strategy to ensure data quality and facilitate subsequent data analysis, wherein the data preprocessing strategy includes steps such as data cleaning, data conversion, and data integration. For example, through data cleaning, outliers or noise points in the data set can be identified and removed to ensure data accuracy, and interpolation, average filling or other methods can be used to fill missing values ​​in the data set; through data integration, data from different sources can be integrated, wherein the environmental monitoring data and the motion monitoring data come from different devices or sensors, and data integration is required to ensure the uniformity and consistency of the data; a standard environmental data set and a standard motion data set are obtained.

[0022] The standard environment data set and the standard motion data set are further input into an indicator interference database for matching, wherein the indicator interference database includes multiple interference parameter combinations and multiple indicator credibility coefficient sets corresponding to the multiple interference parameter combinations, which are constructed after testing with sample data, wherein under different interference parameter combinations, that is, under different interference factors, the credibility coefficient corresponding to each indicator is different, wherein the indicator is a vital sign monitoring indicator, and the credibility coefficient represents the credibility of the vital sign monitoring indicator, which is determined based on the accuracy and stability of the monitoring indicator in the current environment. The higher the accuracy and stability, the higher the credibility of the monitoring indicator, and the larger the credibility coefficient; multiple indicator credibility coefficients are obtained based on the matching results.

[0023] The medical monitoring data division module 13 is used to obtain a predetermined coefficient constraint and divide the medical monitoring data set into a credible monitoring data set and an abnormal monitoring data set according to multiple indicator credibility coefficients and the predetermined coefficient constraint.

[0024] Specifically, a predetermined coefficient constraint is obtained, and the predetermined coefficient constraint is a standard indicator credibility coefficient, which is set based on the monitoring accuracy requirement. The higher the monitoring accuracy requirement, the larger the predetermined coefficient constraint. Then, the plurality of indicator credibility coefficients are judged based on the predetermined coefficient constraint, and the medical monitoring data whose indicator credibility coefficient is greater than the predetermined coefficient constraint is set as credible monitoring data, and the medical monitoring data whose indicator credibility coefficient is less than or equal to the predetermined coefficient constraint is set as abnormal monitoring data, so as to obtain a credible monitoring data set and an abnormal monitoring data set.

[0025] The abnormal monitoring data compensation module 14 is used to predict the abnormal monitoring data according to the trusted monitoring data set based on the indicator mapping association to obtain a predicted abnormal data set, and to map and compensate the abnormal monitoring data set according to the predicted abnormal data set to generate a standard monitoring data set after fusion.

[0026] Specifically, based on the Pearson correlation coefficient, correlation analysis is performed on multiple medical monitoring indicators in the medical monitoring indicator set to identify and quantify the linear relationship between the indicators, and an indicator mapping association is established based on the correlation analysis results. Then, based on the indicator mapping association, the associated credible monitoring data set corresponding to the abnormal monitoring indicator in the abnormal monitoring data set is obtained to obtain multiple associated credible monitoring data sets; then, through the BP neural network model, abnormal monitoring data prediction is performed based on the multiple associated credible monitoring data sets to obtain a predicted abnormal data set.

[0027] Then, mapping compensation is performed on the abnormal monitoring data set according to the predicted abnormal data set, wherein mapping compensation refers to weighted calculation of the predicted abnormal data and the abnormal monitoring data with a mapping relationship, wherein the weight of the abnormal monitoring data is determined based on the indicator credibility coefficient, and the weighted calculation result is used as the standard compensation data of the abnormal monitoring data to obtain a standard compensation data set; finally, a standard monitoring data set is generated according to the credible monitoring data set and the standard compensation data set. By using the credible monitoring data set to perform compensation analysis on the abnormal monitoring data with a mapping relationship, the accuracy of the abnormal monitoring data can be improved, thereby improving the accuracy and reliability of the vital sign monitoring data.

[0028] The health status assessment module 15 is used to call the associated health assessment model based on the device ID, input the standard environment data set, the standard motion data set and the standard monitoring data set into the associated health assessment model, and output the user health status assessment result.

[0029] Specifically, the associated health assessment model is called based on the device ID, wherein the device ID and the associated health assessment model correspond one-to-one, and the associated health assessment model is constructed based on the user vital sign data set corresponding to the device ID to provide personalized health analysis and assessment, thereby further improving the accuracy and reliability of the user's health status assessment. The standard environmental data set, the standard motion data set, and the standard monitoring data set are then input into the associated health assessment model for evaluation to obtain a user health status assessment result, wherein the user health status assessment result is a user health assessment coefficient, and the larger the user health assessment coefficient, the better the user's health status. Finally, comprehensive and personalized health management and intervention strategies can be provided to users based on the user health status assessment results to help users improve their health.

[0030] Furthermore, the health status assessment module 15 is also used for:

[0031] A user vital sign data set is obtained based on the device ID, wherein the device ID and the user vital sign data set correspond one-to-one, and the user vital sign data includes age, gender, height, weight and disease characteristics; with the user vital sign data as a constraint, a predetermined monitoring indicator data set is matched and obtained, and a deviation is calculated for the standard monitoring data set based on the predetermined monitoring indicator data set to obtain a monitoring deviation data set; the monitoring deviation data set, the standard environment data set and the standard motion data set are input into the associated health assessment model for assessment, and a user health status assessment result is output.

[0032] Specifically, first, a user vital sign data set is obtained based on the device ID, where the device ID and the user vital sign data set correspond one-to-one, and the user vital sign data includes age, gender, height, weight and disease characteristics, and the disease characteristics include symptom description, medical history, disease severity, complications, etc. Then, with the user vital sign data as a constraint, a predetermined monitoring indicator data set is matched and obtained, where the predetermined monitoring indicator data is the monitoring indicator data under an ideal state, that is, the standard monitoring data. Then, a mapping deviation calculation is performed on the standard monitoring data set according to the predetermined monitoring indicator data set, and a monitoring deviation data set is obtained based on the deviation calculation result.

[0033] Finally, the monitoring deviation data set, the standard environment data set and the standard motion data set are input into the associated health assessment model to perform user health status assessment, wherein the associated health assessment model is constructed based on a BP neural network and outputs a user health status assessment result.

[0034] Furthermore, the health status assessment module 15 is also used for:

[0035] Taking user vital sign data as constraints, a sample evaluation data set is obtained based on big data retrieval, wherein the sample evaluation data set includes multiple sample monitoring deviation data sets, multiple sample environment data sets, multiple sample motion data sets and a sample health assessment coefficient set; based on the sample evaluation data set, supervised training and cross-validation are performed on the BP neural network to generate an associated health assessment model, and a mapping relationship between the device ID and the associated health assessment model is established.

[0036] Specifically, firstly, with the user's vital sign data as a constraint, a sample evaluation data set is obtained based on big data retrieval, where the sample evaluation data set includes multiple sample monitoring deviation data sets, multiple sample environment data sets, multiple sample motion data sets and sample health assessment coefficient sets. Then, an associated health assessment model is constructed based on the BP neural network, where the input data of the associated health assessment model is the monitoring deviation data set, the standard environment data set and the standard motion data set, and the output data is the user health assessment coefficient, where the larger the user health assessment coefficient, the better the user's health status.

[0037] The associated health assessment model is further supervised and cross-validated based on the sample evaluation data set, wherein the training method of the associated health assessment model is the same as the training method of the above-mentioned first indicator prediction model, which will not be described in detail here. The training method of the above-mentioned first indicator prediction model can be referred to to obtain an associated health assessment model that meets the expected convergence constraints, and establish a mapping relationship between the device ID and the associated health assessment model, that is, the device ID and the associated health assessment model correspond one-to-one.

[0038] A cloud data fusion platform for medical wearable devices is provided, comprising: a data monitoring module, which is used to read the monitoring data of a medical acquisition module, an environmental acquisition module and a motion acquisition module respectively under a preset monitoring node to obtain a medical monitoring data set, an environmental monitoring data set and a motion monitoring data set, and transmit the medical monitoring data set, the environmental monitoring data set and the motion monitoring data set to a cloud analysis platform, wherein the medical acquisition module is communicatively connected to the medical wearable device, and the medical wearable device has a device ID; an indicator interference analysis module, which is used to pre-process the environmental monitoring data set and the motion monitoring data set according to a data pre-processing strategy in the cloud analysis platform to obtain a standard environmental data set and a standard motion data set, and input the standard environmental data set and the standard motion data set into the cloud analysis platform. The indicator interference database is matched to obtain multiple indicator credibility coefficients; the medical monitoring data division module is used to obtain the predetermined coefficient constraints, and the medical monitoring data set is divided into a credible monitoring data set and an abnormal monitoring data set according to the multiple indicator credibility coefficients and the predetermined coefficient constraints; the abnormal monitoring data compensation module is used to predict the abnormal monitoring data based on the indicator mapping association, obtain the predicted abnormal data set, and map and compensate the abnormal monitoring data set according to the predicted abnormal data set, and generate a standard monitoring data set after fusion; the health status assessment module is used to call the associated health assessment model based on the device ID, input the standard environment data set, the standard motion data set and the standard monitoring data set into the associated health assessment model, and output the user health status assessment result. It can solve the existing method of monitoring and evaluating the health status of users based on medical wearable devices. Because the vital sign monitoring data is easily interfered by external factors, the accuracy and stability of the monitoring data are poor, resulting in the technical problem of low accuracy and reliability of the user health status assessment. By determining credible monitoring data and abnormal monitoring data based on external interference factor analysis, and using credible monitoring data to perform targeted mapping and correction of abnormal monitoring data, the accuracy and reliability of vital sign monitoring data can be improved, thereby achieving the technical effect of accurately assessing the user's health status and reducing the probability of misdiagnosis.

[0039] Further, as attached Figure 2 As shown, the indicator interference analysis module 12 is also used for:

[0040] Obtain environmental interference factors and motion interference factors, wherein the environmental interference factors include at least ambient temperature, ambient humidity, lighting conditions and electromagnetic interference, and the motion interference factors include at least motion type, motion intensity and motion duration; divide the environmental interference factors and the motion interference factors according to a preset step size to obtain an ambient temperature threshold set, an ambient humidity threshold set, a lighting condition threshold set, an electromagnetic interference intensity threshold set, a motion type set, a motion intensity threshold set and a motion duration threshold set; in the ambient temperature threshold set, the ambient humidity threshold set, the lighting condition threshold set, the electromagnetic interference intensity threshold set, the motion type set, the motion intensity threshold set and the motion duration threshold set A parameter is randomly selected from the exercise duration threshold set to be combined to generate an interference parameter combination set; with the interference parameter combination set as the test condition, the medical monitoring indicator set is respectively tested for credibility to generate multiple indicator credibility coefficient sets, wherein the interference parameter combination corresponds to the indicator credibility coefficient set one by one, and the medical monitoring indicator set includes at least body temperature, blood sugar, blood pressure, blood oxygen saturation, heart rate and respiratory rate; based on the classification decision principle, with the interference parameter combination as a sub-node and the corresponding indicator credibility coefficient set as the leaf node of the sub-node, an indicator interference database is generated according to the interference parameter combination set and the multiple indicator credibility coefficient sets.

[0041] Specifically, first, environmental interference factors and motion interference factors are obtained, among which environmental interference factors refer to environmental factors that interfere with the accuracy of vital sign monitoring data, including at least ambient temperature, ambient humidity, lighting conditions and electromagnetic interference. For example, ambient temperature will affect the accuracy of body temperature monitoring, and external light will affect blood oxygen saturation readings, etc.; motion interference factors refer to factors that interfere with the accuracy of vital sign monitoring data due to the user's exercise status, including at least exercise type, exercise intensity and exercise duration, among which exercise types include running, swimming, etc.

[0042] Then, the environmental interference factors and the motion interference factors are divided according to the preset step length, wherein the preset step length can be set according to the actual situation, and the step length of each interference factor is different, for example: the environmental temperature step length can be set to 0.1 degrees Celsius, the motion time step length can be set to 1 minute, etc.; obtain the environmental temperature threshold set, environmental humidity threshold set, lighting condition threshold set, electromagnetic interference intensity threshold set, motion type set, motion intensity threshold set and motion duration threshold set, wherein the environmental temperature threshold set includes multiple environmental temperature thresholds, and the interval range of each environmental temperature threshold is the preset environmental temperature step length. Further, a parameter is randomly selected from the environmental temperature threshold set, environmental humidity threshold set, lighting condition threshold set, electromagnetic interference intensity threshold set, motion type set, motion intensity threshold set and motion duration threshold set for combination to generate a first interference parameter combination, and multiple interference parameter combinations are obtained in sequence using the same method to construct an interference parameter combination set.

[0043] Then, multiple interference test conditions are set based on the interference parameter combination set, and under the multiple interference test conditions, the medical monitoring indicator set is respectively tested for credibility, wherein the medical monitoring indicator set includes at least body temperature, blood sugar, blood pressure, blood oxygen saturation, heart rate and respiratory rate, and additional vital sign monitoring indicators can also be added according to actual conditions. Based on the credibility test results, multiple indicator credibility coefficient sets are obtained, wherein each interference parameter combination corresponds to an indicator credibility coefficient set.

[0044] The decision tree is a supervised learning algorithm for classification and regression. It divides data points according to features and attributes to form a tree structure, in which each node represents a feature or attribute, and each branch represents a decision. Finally, based on the principle of classification decision-making, the interference parameter combination is used as a sub-node, and the corresponding indicator credibility coefficient set is used as the leaf node of the sub-node. According to the interference parameter combination set and the multiple indicator credibility coefficient sets, an indicator interference database is constructed, in which each node of the indicator interference database represents an interference parameter combination, and the leaf node represents the indicator credibility coefficient under the combination. By constructing the indicator interference database, the mapping association between interference factors and medical monitoring indicator credibility coefficients can be better understood, and the accuracy and efficiency of indicator credibility coefficient matching can be improved.

[0045] Furthermore, the indicator interference analysis module 12 is also used for:

[0046] A first interference parameter combination is randomly selected from the interference parameter combination set, and a first medical monitoring indicator is randomly selected from the medical monitoring indicator set; with the first interference parameter combination and the first medical monitoring indicator as constraints, a first sample indicator deviation set is obtained based on big data retrieval; based on the first sample indicator deviation set, error and volatility analysis are respectively performed to obtain a first error rate and a first fluctuation coefficient, wherein the first error rate is the ratio of the number of indicators whose indicator deviation is greater than a predetermined deviation threshold to the total number of indicators in the first sample indicator deviation set, and the first fluctuation coefficient is proportional to the variance of the first sample indicator deviation set; based on the first error rate and the first fluctuation coefficient, a credibility analysis is performed to obtain a first credibility coefficient, and the first credibility coefficient is added to the first indicator credibility coefficient set, and multiple indicator credibility coefficient sets are obtained in sequence.

[0047] Specifically, first, a first interference parameter combination is randomly selected from the interference parameter combination set, wherein the first interference parameter combination is any one of the interference parameter combination set; a first medical monitoring indicator is randomly selected from the medical monitoring indicator set, wherein the first medical monitoring indicator is any one of the medical monitoring indicator set. Then, taking the first interference parameter combination and the first medical monitoring indicator as constraints, a first sample indicator deviation set is obtained based on big data retrieval, wherein in order to evaluate the impact of different interference parameter combinations on specific medical monitoring indicators, a large amount of historical data needs to be collected for analysis, and this process is called big data retrieval, the purpose of which is to retrieve data samples that match specific constraints from the database; the first sample indicator deviation set includes multiple first sample indicator deviation data, wherein the indicator deviation refers to the deviation value between the actual indicator monitoring data and the ideal indicator monitoring data, that is, the deviation value caused by external factors.

[0048] Then, based on the first sample indicator deviation set, error and volatility analysis are respectively performed to obtain the first error rate and the first volatility coefficient, wherein the first error rate is the ratio of the number of indicators whose indicator deviation is greater than the predetermined deviation threshold to the total number of indicators in the first sample indicator deviation set. The larger the first error rate, the greater the deviation degree of the first medical monitoring indicator; the first volatility coefficient is proportional to the variance of the first sample indicator deviation set, that is, the variance of the first sample indicator deviation set. The larger the first volatility coefficient, the worse the reliability of the first medical monitoring indicator. Then, based on the first error rate and the first volatility coefficient, credibility analysis is performed, that is, the first error rate and the first volatility coefficient are weighted and summed, and the inverse of the weighted summation result is set as the first credibility coefficient, wherein the weight value is set based on the degree of influence of the indicator on the overall credibility, wherein the larger the first credibility coefficient, the greater the credibility degree of the first medical monitoring indicator under the first interference parameter combination. Finally, the first credibility coefficient is added to the first indicator credibility coefficient set, and multiple indicator credibility coefficient sets are obtained in sequence using the same method.

[0049] Furthermore, the abnormal monitoring data compensation module 14 is also used for:

[0050] A first medical monitoring indicator is randomly selected from the medical monitoring indicator set, wherein the first medical monitoring indicator is any one of the medical monitoring indicator set; using the Pearson correlation coefficient, correlation analysis is performed on the first medical monitoring indicator and other medical monitoring indicators in the medical monitoring indicator set, and other medical monitoring indicators with correlation are set as first associated medical monitoring indicators to obtain a first associated medical monitoring indicator set; and the indicator mapping association is constructed based on the first medical monitoring indicator and the first associated medical monitoring indicator set.

[0051] Specifically, in the field of medical monitoring, correlation analysis is performed to understand the relationship between different medical monitoring indicators in order to better interpret and utilize these data. First, a first medical monitoring indicator is randomly selected from the medical monitoring indicator set, wherein the first medical monitoring indicator is any one of the medical monitoring indicator set; then the Pearson correlation coefficient is calculated for the selected first medical monitoring indicator and other indicators in the medical monitoring indicator set, wherein the Pearson correlation coefficient is used to measure the linear correlation between two variables, and its value is between -1 and 1; then the correlation between the two indicators is judged according to the value of the Pearson correlation coefficient. Generally, a correlation coefficient greater than 0.5 or less than -0.5 is considered to be an indicator with significant correlation; further, other medical monitoring indicators with significant correlation with the first medical monitoring indicator are set as first associated medical monitoring indicators to obtain a first associated medical monitoring indicator set. Then, the same method is used for iterative analysis to obtain multiple medical monitoring indicators and multiple associated medical monitoring indicator sets, and the indicator mapping association is constructed according to multiple medical monitoring indicators and multiple associated medical monitoring indicator sets.

[0052] Furthermore, the abnormal monitoring data compensation module 14 is also used for:

[0053] A first abnormal monitoring indicator is selected from the abnormal monitoring data set, and a first associated indicator set of the first abnormal monitoring indicator is matched based on the indicator mapping association; based on the first associated indicator set, extraction is performed in the trusted monitoring data set to obtain a first trusted data set; based on the first abnormal monitoring indicator, a first indicator prediction model is called, the first trusted data set is input into the first indicator prediction model, first predicted abnormal data is output, and the data is added to the predicted abnormal data set.

[0054] Specifically, first, a first abnormal monitoring indicator is selected from the abnormal monitoring data set, wherein the first abnormal monitoring indicator is any one of the corresponding multiple abnormal monitoring indicators in the abnormal monitoring data set. Then, based on the indicator mapping association, a first associated indicator set of the first abnormal monitoring indicator is matched. Then, mapping and extraction are performed in the trusted monitoring data set according to the first associated indicator set to obtain a first trusted data set.

[0055] Then, based on the first abnormal monitoring indicator, a first indicator prediction model is called, wherein the first indicator prediction model is constructed based on a BP neural network and corresponds to the first abnormal monitoring indicator one by one. The first credible data set is further input into the first indicator prediction model to predict the abnormal monitoring indicator, obtain the first predicted abnormal data, and add the first predicted abnormal data to the predicted abnormal data set to obtain the predicted abnormal data set.

[0056] Furthermore, the abnormal monitoring data compensation module 14 is also used for:

[0057] Taking the first abnormal monitoring indicator and the first associated indicator set as constraints, a sample data set is retrieved online to obtain the sample data set, wherein the sample data set includes multiple sample first associated data sets and sample first abnormal data sets; the sample data set is divided into N equal parts, and selected N times without replacement to generate a first sample data set, and the selection is continued iteratively N times to obtain N sample data sets; taking the first associated data set as input and the first abnormal data as output, the BP neural network is supervised trained and cross-validated using the N sample data sets to obtain N convergence indicator prediction units; based on the N convergence indicator prediction units, a first indicator prediction model is built in parallel, wherein the output of the first indicator prediction model is the mode of the output results of the N convergence indicator prediction units.

[0058] Specifically, first, with the first abnormal monitoring indicator and the first associated indicator set as constraints, a sample data set is retrieved online based on big data technology, wherein the sample data set includes multiple sample first associated data sets and sample first abnormal data sets. Then, the sample data set is divided into N parts, each of which contains the same amount of sample data, while ensuring that each part has a representative sample distribution to facilitate subsequent model training and evaluation; then, N times are selected from the N parts without replacement to generate the first sample data set, and the same method is used to continue to iterate and select N times to obtain N sample data sets.

[0059] An indicator prediction unit is constructed based on a BP neural network. The indicator prediction unit is a neural network model that can be iteratively optimized in machine learning. The input data of the indicator prediction unit is a related data set, and the output data is abnormal data. Then, the first related data set is used as input, the first abnormal data is used as output, and N sample data sets are used to perform supervised training and cross-validation on the BP neural network. The supervised training process is to input the training set data, calculate the network output through forward propagation, and adjust the weight through back propagation, and then repeat this process until the performance of the network on the validation set reaches a preset threshold or reaches a preset number of iterations. The cross-validation process is to randomly exchange the N sample data sets as validation data sets, and then perform validation training until the preset convergence index is met, and then output N convergence indicator prediction units.

[0060] Finally, a first indicator prediction model is built in parallel based on N convergence indicator prediction units, wherein the output of the first indicator prediction model is the mode of the output results of the N convergence indicator prediction units, so as to improve the accuracy and reliability of the output results of the first indicator prediction model.

[0061] Furthermore, the abnormal monitoring data compensation module 14 is also used for:

[0062] Obtain the indicator credibility coefficient of the abnormal monitoring data in the abnormal monitoring data set, set the credibility weight of the abnormal monitoring data based on the indicator credibility coefficient, and obtain multiple monitoring credibility weights, wherein the monitoring credibility weight corresponds to the abnormal monitoring data one-to-one and is positively correlated with the indicator credibility coefficient; obtain multiple prediction credibility weights based on the multiple monitoring credibility weights, wherein the sum of the monitoring credibility weight and the prediction credibility weight of the same abnormal monitoring data is 1; perform mapping compensation according to the multiple monitoring credibility weights and the multiple prediction credibility weights.

[0063] Specifically, the indicator credibility coefficient of the abnormal monitoring data in the abnormal monitoring data set is obtained, and the corresponding credibility weight is set according to the indicator credibility coefficient, wherein the credibility weight reflects the reliability and accuracy of the data quality, corresponds to the abnormal monitoring data one by one, and is positively correlated with the indicator credibility coefficient, that is, the larger the indicator credibility coefficient is, the higher the credibility of the characterization data is, and the larger the corresponding credibility weight is. The credibility weight can be set by the existing coefficient of variation method to obtain multiple monitoring credibility weights. Then, multiple prediction credibility weights are calculated based on multiple monitoring credibility weights, wherein the sum of the monitoring credibility weight and the prediction credibility weight of the same abnormal monitoring data is 1. Finally, the predicted abnormal data set and the abnormal monitoring data set are mapped and weighted according to multiple monitoring credibility weights and multiple prediction credibility weights to obtain a standard compensation data set.

[0064] In summary, the cloud data fusion platform for medical wearable devices provided by this application has the following technical effects:

[0065] By collecting the medical monitoring data set, environmental monitoring data set and motion monitoring data set of the target user, and then performing indicator interference analysis based on the environmental monitoring data set and the motion monitoring data set, the indicator credibility coefficients of multiple monitoring indicators are obtained; then the medical monitoring data with a credibility coefficient greater than the predetermined coefficient constraint is divided into credible monitoring data, and the medical monitoring data with a credibility coefficient less than or equal to the predetermined coefficient constraint is divided into abnormal monitoring data, and a credible monitoring data set and an abnormal monitoring data set are obtained; then based on the mapping association between indicators, the abnormal monitoring data is predicted according to the credible monitoring data set, and the predicted abnormal data set is mapped and compensated for the abnormal monitoring data set, and a standard monitoring data set is generated after fusion; finally, the standard environmental data set, the standard motion data set and the standard monitoring data set are input into the associated health assessment model for assessment, and the user health status assessment result is obtained. By determining the credible monitoring data and the abnormal monitoring data based on the external interference factor analysis, and using the credible monitoring data to perform targeted mapping correction on the abnormal monitoring data, the accuracy and reliability of the vital sign monitoring data can be improved, thereby achieving the technical effect of accurately assessing the user's health status and reducing the probability of misdiagnosis.

[0066] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0067] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. A cloud data fusion platform for medical wearable devices, characterized by: include: A data monitoring module is used to read the monitoring data of the medical acquisition module, the environmental acquisition module and the motion acquisition module respectively under a preset monitoring node, obtain a medical monitoring data set, an environmental monitoring data set and a motion monitoring data set, and transmit the medical monitoring data set, the environmental monitoring data set and the motion monitoring data set to a cloud analysis platform, wherein the medical acquisition module is communicatively connected with a medical wearable device, and the medical wearable device has a device ID; The indicator interference analysis module is used to pre-process the environmental monitoring data set and the motion monitoring data set in the cloud analysis platform according to the data pre-processing strategy to obtain a standard environmental data set and a standard motion data set, and input the standard environmental data set and the standard motion data set into the indicator interference database for matching to obtain multiple indicator credibility coefficients, wherein the indicator interference database includes multiple interference parameter combinations and multiple indicator credibility coefficient sets corresponding to the multiple interference parameter combinations, which are constructed after testing with sample data; under different interference parameter combinations, the credibility coefficient corresponding to each indicator is different, wherein the indicator is a vital sign monitoring indicator; the indicator credibility coefficient represents the credibility of the vital sign monitoring indicator, which is determined based on the accuracy and stability of the monitoring indicator in the current environment, the higher the accuracy and stability, the higher the credibility of the monitoring indicator, and the larger the credibility coefficient; multiple indicator credibility coefficients are obtained based on the matching results; A medical monitoring data division module, used for obtaining a predetermined coefficient constraint, and dividing the medical monitoring data set into a credible monitoring data set and an abnormal monitoring data set according to a plurality of indicator credible coefficients and the predetermined coefficient constraint; An abnormal monitoring data compensation module is used to predict abnormal monitoring data based on the trusted monitoring data set based on the indicator mapping association to obtain a predicted abnormal data set, and to perform mapping compensation on the abnormal monitoring data set based on the predicted abnormal data set, and to generate a standard monitoring data set after fusion, wherein correlation analysis is performed on multiple medical monitoring indicators in the medical monitoring indicator set based on the Pearson correlation coefficient to identify and quantify the linear relationship between the indicators, and to establish an indicator mapping association based on the correlation analysis result; A health status assessment module, used to calculate a monitoring deviation data set according to the standard monitoring data set, call an associated health assessment model based on a device ID, input the standard environment data set, the standard motion data set and the monitoring deviation data set into the associated health assessment model, and output a user health status assessment result; Wherein, the health status assessment module is also used for: Obtaining a user's vital signs data set based on the device ID, wherein the device ID corresponds to the user's vital signs data set one by one, and the user's vital signs data includes age, gender, height, weight, and disease characteristics; Taking the user's vital sign data as a constraint, matching and acquiring a predetermined monitoring indicator data set, and performing deviation calculation on the standard monitoring data set based on the predetermined monitoring indicator data set to obtain a monitoring deviation data set; Inputting the monitoring deviation data set, the standard environment data set and the standard motion data set into the associated health assessment model for assessment, and outputting a user health status assessment result; Furthermore, the health status assessment module is also used for: Taking the user's vital sign data as a constraint, a sample evaluation data set is obtained based on big data retrieval, wherein the sample evaluation data set includes multiple sample monitoring deviation data sets, multiple sample environment data sets, multiple sample motion data sets and a sample health assessment coefficient set; Based on the sample evaluation data set, supervised training and cross-validation are performed on the BP neural network to generate an associated health evaluation model, and a mapping relationship between the device ID and the associated health evaluation model is established; The indicator interference analysis module is also used for: Obtaining environmental interference factors and motion interference factors, wherein the environmental interference factors at least include ambient temperature, ambient humidity, lighting conditions, and electromagnetic interference, and the motion interference factors at least include motion type, motion intensity, and motion duration; The environmental interference factors and the motion interference factors are divided according to a preset step size to obtain an environmental temperature threshold set, an environmental humidity threshold set, a lighting condition threshold set, an electromagnetic interference intensity threshold set, a motion type set, a motion intensity threshold set, and a motion duration threshold set; Randomly select a parameter from the ambient temperature threshold set, the ambient humidity threshold set, the illumination condition threshold set, the electromagnetic interference intensity threshold set, the motion type set, the motion intensity threshold set, and the motion duration threshold set to combine and generate an interference parameter combination set; Taking the interference parameter combination set as the test condition, respectively performing credibility tests on the medical monitoring indicator set to generate multiple indicator credibility coefficient sets, wherein the interference parameter combination corresponds to the indicator credibility coefficient set one by one, and the medical monitoring indicator set at least includes body temperature, blood sugar, blood pressure, blood oxygen saturation, heart rate and respiratory rate; Based on the classification decision principle, the interference parameter combination is used as a child node, and the corresponding indicator credibility coefficient set is used as the leaf node of the child node. The indicator interference database is generated according to the interference parameter combination set and the multiple indicator credibility coefficient sets.

2. The cloud data fusion platform for medical wearable devices according to claim 1, characterized in that: The indicator interference analysis module is also used for: Randomly select a first interference parameter combination from the interference parameter combination set, and randomly select a first medical monitoring indicator from the medical monitoring indicator set; Taking the first interference parameter combination and the first medical monitoring indicator as constraints, retrieving data samples matching the constraints from historical data to obtain a first sample indicator deviation set; Based on the first sample indicator deviation set, error and volatility analysis are respectively performed to obtain a first error rate and a first volatility coefficient, wherein the first error rate is the ratio of the number of indicators whose indicator deviation is greater than a predetermined deviation threshold to the total number of indicators in the first sample indicator deviation set, and the first volatility coefficient is proportional to the variance of the first sample indicator deviation set; A credibility analysis is performed based on the first error rate and the first fluctuation coefficient to obtain a first credibility coefficient, which is added to a first indicator credibility coefficient set to obtain multiple indicator credibility coefficient sets in sequence.

3. The cloud data fusion platform for medical wearable devices according to claim 1, characterized in that: The abnormal monitoring data compensation module is also used for: Randomly selecting a first medical monitoring indicator from the medical monitoring indicator set, wherein the first medical monitoring indicator is any one of the medical monitoring indicator set; Using the Pearson correlation coefficient, respectively perform correlation analysis on the first medical monitoring indicator and other medical monitoring indicators in the medical monitoring indicator set, and set other medical monitoring indicators with correlation as first associated medical monitoring indicators to obtain a first associated medical monitoring indicator set; The indicator mapping association is constructed based on the first medical monitoring indicator and the first associated medical monitoring indicator set.

4. The cloud data fusion platform for medical wearable devices according to claim 3, characterized in that: The abnormal monitoring data compensation module is also used for: Selecting a first abnormal monitoring indicator from the abnormal monitoring data set, and matching based on the indicator mapping association to obtain a first associated indicator set of the first abnormal monitoring indicator; Based on the first associated indicator set, extraction is performed in the trusted monitoring data set to obtain a first trusted data set; Based on the first abnormal monitoring indicator, a first indicator prediction model is called, the first credible data set is input into the first indicator prediction model, first predicted abnormal data is output, and added to the predicted abnormal data set.

5. The cloud data fusion platform for medical wearable devices according to claim 4, characterized in that: The abnormal monitoring data compensation module is also used for: Taking the first abnormal monitoring indicator and the first associated indicator set as constraints, searching online to obtain a sample data set, wherein the sample data set includes a plurality of sample first associated data sets and a sample first abnormal data set; Divide the sample data set into N equal parts, select N parts without replacement to generate a first sample data set, and continue to iteratively select N parts to obtain N sample data sets; Taking the first associated data set as input and the first abnormal data as output, using N sample data sets to perform supervised training and cross-validation on the BP neural network to obtain N convergence index prediction units; A first indicator prediction model is built in parallel based on N convergence indicator prediction units, wherein the output of the first indicator prediction model is the mode of the output results of the N convergence indicator prediction units.

6. The cloud data fusion platform for medical wearable devices according to claim 1, characterized in that: The abnormal monitoring data compensation module is also used for: Obtaining an indicator credibility coefficient of the abnormal monitoring data in the abnormal monitoring data set, setting a credibility weight of the abnormal monitoring data based on the indicator credibility coefficient, and obtaining a plurality of monitoring credibility weights, wherein the monitoring credibility weights correspond to the abnormal monitoring data one by one and are positively correlated with the indicator credibility coefficient; Based on the multiple monitoring credible weights, multiple prediction credible weights are calculated, wherein the sum of the monitoring credible weight and the prediction credible weight of the same abnormal monitoring data is 1; Mapping compensation is performed according to the plurality of monitoring trust weights and the plurality of prediction trust weights.

Images