Abnormal breathing analysis system for monitoring and analyzing breathing data

Abstract

The invention discloses a respiration abnormity analysis system for respiration data monitoring analysis, and relates to the technical field of data processing, the system comprises a monitoring data sequence obtaining module used for obtaining a respiration monitoring data sequence and an environment monitoring data sequence; the respiration monitoring data sub-sequence obtaining module is used for obtaining K respiration monitoring data sub-sequences; the environment monitoring data sub-sequence obtaining module is used for obtaining K environment monitoring data sub-sequences; the feature vector obtaining module is used for obtaining a correlation fusion multi-dimensional monitoring feature vector; the abnormal record set obtaining module is used for retrieving and obtaining a matched monitoring abnormal record set; an asynchronous monitoring step length set obtaining module; a monitoring data set obtaining module; and the breathing abnormity analysis result obtaining module is used for obtaining a breathing abnormity analysis result. The technical problems that in the prior art, breathing anomaly analysis is low in accuracy and long in data analysis period are solved, and the technical effect of improving anomaly analysis reliability is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a respiratory abnormality analysis system for monitoring and analyzing respiratory data. Background Art

[0002] Abnormal breathing analysis has a wide range of applications in the fields of medical health, environmental monitoring, and personalized health management, such as the diagnosis of sleep apnea syndrome, monitoring of chronic respiratory diseases, and health assessment of specific occupational groups. However, the collection of respiratory data is usually interfered by external factors such as ambient temperature, humidity, and air pressure, and the respiratory monitoring equipment in the existing technology often operates alone, resulting in low accuracy in the extraction of abnormal breathing features, and the analysis results are difficult to truly reflect the actual situation. In addition, it relies on a fixed acquisition frequency and a linear processing flow, making it difficult to quickly respond to complex scenario requirements during data collection and analysis, and the analysis results are significantly delayed.

[0003] The existing technology has technical problems of low accuracy in abnormal breathing analysis and long data analysis cycle. Summary of the invention

[0004] The present application provides a respiratory abnormality analysis system for respiratory data monitoring and analysis, which is used to solve the technical problems of low accuracy of respiratory abnormality analysis and long data analysis cycle in the prior art.

[0005] In view of the above problems, the present application provides a respiratory abnormality analysis system for respiratory data monitoring and analysis, the system comprising:

[0006] A monitoring data sequence acquisition module is used to interact with the target monitoring device and the environmental monitoring module to extract the monitoring data of the preset monitoring window, and obtain a respiratory monitoring data sequence and an environmental monitoring data sequence, wherein the respiratory monitoring data corresponds to the environmental monitoring data one by one;

[0007] A respiratory monitoring data subsequence obtaining module, used to conditionally select K-1 respiratory monitoring data from the respiratory monitoring data sequence as K-1 starting division points, divide the respiratory monitoring data sequence with the K-1 starting division points, and obtain K respiratory monitoring data subsequences, wherein K is an integer greater than or equal to 3;

[0008] An environment monitoring data subsequence obtaining module, used for performing subsequence division on the environment monitoring data sequence based on the K respiratory monitoring data subsequences to obtain K environment monitoring data subsequences;

[0009] A feature vector acquisition module is used to use the K respiratory monitoring data subsequences as foreground data and the K environmental monitoring data subsequences as background data, perform association fusion analysis in sequence, and obtain an association fusion multi-dimensional monitoring feature vector;

[0010] The abnormal record set acquisition module is used to obtain the basic user characteristics of the target user, perform data mining in the big data in combination with the associated fusion multi-dimensional monitoring feature vector, and retrieve and obtain the matching monitoring abnormal record set;

[0011] An asynchronous monitoring step set acquisition module is used to perform cluster analysis on the matching monitoring anomaly record set with the anomaly type as an index to determine an asynchronous monitoring step set, wherein the asynchronous monitoring step set includes a first monitoring step and a second monitoring step, and the first monitoring step is less than or equal to the second monitoring step;

[0012] a monitoring data set acquisition module, configured to control the target monitoring device to perform asynchronous breathing monitoring on the target user within a feedback monitoring window according to the first monitoring step length and the second monitoring step length, respectively, to obtain a first breathing monitoring data set and a second breathing monitoring data set;

[0013] The breathing abnormality analysis result obtaining module is used to use a breathing abnormality analyzer to perform abnormality analysis on the first breathing monitoring data set and the second breathing monitoring data set to obtain a breathing abnormality analysis result.

[0014] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0015] The present application extracts monitoring data of a preset monitoring window through an interactive target monitoring device and an environmental monitoring module to obtain a respiratory monitoring data sequence and an environmental monitoring data sequence, wherein the respiratory monitoring data corresponds to the environmental monitoring data one by one, and then conditionally selects K-1 respiratory monitoring data from the respiratory monitoring data sequence as K-1 starting division points, divides the respiratory monitoring data sequence with the K-1 starting division points, and obtains K respiratory monitoring data subsequences, wherein K is an integer greater than or equal to 3, and then performs subsequence division on the environmental monitoring data sequence based on the K respiratory monitoring data subsequences to obtain K environmental monitoring data subsequences, uses the K respiratory monitoring data subsequences as foreground data, and uses the K environmental monitoring data subsequences as background data, and sequentially performs association fusion analysis to obtain an association fusion multidimensional monitoring data sequence. The characteristic vector is measured, and the basic characteristics of the target user are obtained, and data mining is performed in big data in combination with the associated fusion multi-dimensional monitoring characteristic vector, and a matching monitoring abnormal record set is retrieved, and then the matching monitoring abnormal record set is clustered and analyzed with the abnormal type as the index, and the asynchronous monitoring step set is determined, wherein the asynchronous monitoring step set includes the first monitoring step and the second monitoring step, and the first monitoring step is less than or equal to the second monitoring step. The target monitoring device is controlled to perform asynchronous breathing monitoring on the target user in the feedback monitoring window according to the first monitoring step and the second monitoring step, respectively, to obtain the first breathing monitoring data set and the second breathing monitoring data set, and then the breathing abnormality analyzer is used to perform abnormal analysis on the first breathing monitoring data set and the second breathing monitoring data set to obtain the breathing abnormality analysis result. The technical effect of analyzing the user's breathing abnormality in accordance with the actual situation and improving the efficiency of monitoring and analysis on the basis of ensuring the monitoring accuracy is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a structural schematic diagram of a respiratory abnormality analysis system for monitoring and analyzing respiratory data provided by an embodiment of the present invention.

[0017] Figure 2 It is a flowchart diagram of the execution steps of an environmental monitoring data subsequence acquisition module in a respiratory abnormality analysis system for respiratory data monitoring and analysis provided by an embodiment of the present invention.

[0018] Explanation of the accompanying drawings: monitoring data sequence acquisition module 11, respiratory monitoring data subsequence acquisition module 12, environmental monitoring data subsequence acquisition module 13, feature vector acquisition module 14, abnormal record set acquisition module 15, asynchronous monitoring step set acquisition module 16, monitoring data set acquisition module 17, respiratory abnormality analysis result acquisition module 18. DETAILED DESCRIPTION

[0019] The present application provides a respiratory abnormality analysis system for monitoring and analyzing respiratory data, which is used to solve the technical problems of low accuracy of respiratory abnormality analysis and long data analysis cycle in the prior art.

[0020] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0021] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, system, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules not clearly listed or inherent to these processes, systems, products or devices.

[0022] Examples, such as Figure 1 As shown, the present application provides a respiratory abnormality analysis system for respiratory data monitoring and analysis, wherein the system comprises:

[0023] A monitoring data sequence acquisition module 11 is used to interact with the target monitoring device and the environmental monitoring module to extract the monitoring data of the preset monitoring window, and obtain a respiratory monitoring data sequence and an environmental monitoring data sequence, wherein the respiratory monitoring data corresponds to the environmental monitoring data one by one;

[0024] In a possible embodiment, the preset monitoring window is a monitoring time period preset by a person skilled in the art, for example, setting a window period of every 15 minutes so as to obtain representative data in a shorter time. This window should be optimized according to the usage scenario of the device and the actual situation of the user to ensure the real-time nature of the data.

[0025] Preferably, in each monitoring window, the target monitoring device (such as a respiratory waveform monitor and a blood oximeter, etc.) and the environmental monitoring module (such as a temperature sensor, a humidity sensor, and a light sensor) start collecting data at the same time. In order to improve the one-to-one correspondence between respiratory monitoring data and environmental monitoring data, the target monitoring device and the environmental monitoring module need to ensure that physiological data and environmental data are obtained at the same time point or frequency.

[0026] Preferably, the respiratory monitoring data (such as respiratory rate, respiratory waveform, minute ventilation, etc.) collected by the target monitoring device are integrated to form a "respiratory monitoring data sequence", and the environmental data (such as temperature, humidity, light, etc.) monitored by the environmental monitoring module are integrated to form an "environmental monitoring data sequence". In these two sequences, each piece of data corresponds to a time point one by one, so that the respiratory state and environmental changes can be simultaneously referenced in subsequent analysis.

[0027] By generating synchronous and corresponding respiratory monitoring data sequences and environmental data sequences, the technical effect of providing accurate data support for subsequent respiratory abnormality analysis and monitoring strategy adjustment is achieved.

[0028] A respiratory monitoring data subsequence obtaining module 12 is used to conditionally select K-1 respiratory monitoring data from the respiratory monitoring data sequence as K-1 starting division points, divide the respiratory monitoring data sequence with the K-1 starting division points, and obtain K respiratory monitoring data subsequences, wherein K is an integer greater than or equal to 3;

[0029] Furthermore, the respiratory monitoring data subsequence obtaining module 12 is used to perform the following steps:

[0030] Randomly select K-1 respiratory monitoring data from the respiratory monitoring data sequence as K-1 first division points;

[0031] Performing left and right neighbor similarity analysis on the K-1 first division points to determine the K-1 neighbor similarities of the K-1 first division points;

[0032] Determine whether the K-1 nearest neighbor similarities are all less than or equal to a preset similarity difference threshold, if not, reselect K-1 first division points;

[0033] If so, the K-1 first division points are used as the K-1 starting division points.

[0034] In a possible embodiment, K-1 data points are selected from the entire respiratory monitoring data sequence according to preset conditions and used as dividing points. The selection of each dividing point should meet specific conditions so as to accurately divide the respiratory monitoring data sequence and obtain K respiratory monitoring data subsequences. Optionally, the preset condition is that the data similarity on the left and right sides of the K-1 starting dividing points should be less than or equal to a preset similarity threshold, indicating that it can be used as a dividing point. These dividing points divide the sequence into K subsequences, where K is an integer greater than or equal to 3.

[0035] Optionally, K-1 respiratory monitoring data are randomly selected from the respiratory monitoring data sequence as K-1 first dividing points. The respiratory monitoring data sequence between the two first dividing points is respectively used as the left data sequence of a first dividing point and the right data sequence of another first dividing point. Then, the cosine similarity formula is used to identify the approximation of the left data sequence and the right data sequence of each first dividing point to obtain K-1 near-neighbor similarities. Among them, each near-neighbor similarity reflects the similarity of the data sequences on the left and right sides of each first dividing point. The larger the near-neighbor similarity, the higher the similarity of the data sequences on the left and right sides of the first dividing point, and the first dividing point cannot accurately divide different data in the respiratory monitoring data sequence.

[0036] After obtaining the K-1 nearest neighbor similarities, determine whether the K-1 nearest neighbor similarities are all less than or equal to the preset similarity difference threshold, if not, reselect the K-1 first division points. If yes, use the K-1 first division points as the K-1 starting division points.

[0037] The respiratory monitoring data sequence is divided by these final K-1 starting division points to form K respiratory monitoring data subsequences. By performing reasonable subsequence division on the respiratory monitoring data sequence, a technical effect of providing more accurate respiratory monitoring data for subsequent analysis is achieved.

[0038] An environment monitoring data subsequence obtaining module 13, configured to divide the environment monitoring data sequence into subsequences based on the K respiratory monitoring data subsequences to obtain K environment monitoring data subsequences;

[0039] In one embodiment, since there is a one-to-one correspondence between the respiratory monitoring data and the environmental monitoring data, the environmental monitoring data sequence is divided according to the start and end time of each subsequence in the K respiratory monitoring data subsequences, and the corresponding K environmental monitoring data subsequences can be obtained. Among them, the K environmental monitoring data subsequences are the environmental changes corresponding to the K respiratory monitoring data subsequences. The technical effect of providing data support for subsequent multi-dimensional data association fusion analysis is achieved.

[0040] A feature vector obtaining module 14 is used to divide the environment monitoring data sequence into subsequences based on the K respiratory monitoring data subsequences to obtain K environment monitoring data subsequences;

[0041] Furthermore, the feature vector obtaining module 14 is used to perform the following steps:

[0042] Traversing the K respiratory monitoring data subsequences and the K environmental monitoring data subsequences to perform data concentration trend analysis respectively, and determining K concentrated respiratory monitoring data and K concentrated environmental monitoring data;

[0043] Respectively extracting the first centralized respiratory monitoring data and the first centralized environmental monitoring data that are at the top of the K centralized respiratory monitoring data and the K centralized environmental monitoring data, and using the first centralized respiratory monitoring data and the first centralized environmental monitoring data as the first memory space;

[0044] Extracting the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data located at the second position among the K concentrated respiratory monitoring data and the K concentrated environmental monitoring data again, capturing associated keywords for the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data with the first memory space as an index, and adding the capture result and the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data into the second memory space;

[0045] Extract the K-th concentrated respiratory monitoring data and the K-th concentrated environmental monitoring data, respectively use the first memory space, the second memory space to the K-1-th memory space as indexes, capture the associated keywords of the K-th concentrated respiratory monitoring data and the K-th concentrated environmental monitoring data, and add the capture results and the K-th concentrated respiratory monitoring data and the K-th concentrated environmental monitoring data into the K-th memory space;

[0046] The semantic recognition network layer is used to perform fusion feature vector recognition on the data in the Kth memory space to obtain the associated fusion multi-dimensional monitoring feature vector.

[0047] In a possible embodiment, the K respiratory monitoring data subsequences are set as "foreground data" to describe the user's respiratory state. The K environmental monitoring data subsequences are set as "background data" to analyze environmental conditions that affect the respiratory state. In this way, synchronous analysis of the user's breathing and the environment can be achieved.

[0048] Optionally, by performing a central tendency analysis on the data in each of the K respiratory monitoring data subsequences and the K environmental monitoring data subsequences, the central value of the data in each subsequence is determined, thereby reducing the amount of data analysis for subsequent association fusion analysis. Moreover, since the most representative and more distributed central data in each subsequence are selected, the accuracy of the analysis will not be reduced.

[0049] Then, analyze each centralized respiratory monitoring data and each centralized environmental data in turn, and the user's respiratory condition and environmental status reflected in them, and their continuation in the next centralized respiratory monitoring data and the next centralized environmental data. The first centralized respiratory monitoring data and the first centralized environmental monitoring data that are at the top of the K centralized respiratory monitoring data and the K centralized environmental monitoring data are extracted respectively, and the first centralized respiratory monitoring data and the first centralized environmental monitoring data are used as the first memory space. The first memory space is used to store the foreground data and background data that are at the top.

[0050] Optionally, extract the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data that are located in the second position among the K concentrated respiratory monitoring data and the K concentrated environmental monitoring data again. Take the first memory space as an index, and perform associated keyword capture on the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data, that is, retrieve the data associated with the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data and the data stored in the first memory space, and then identify the associated keywords based on the retrieval results to obtain the capture results. Among them, the capture result reflects the association between the first memory space and the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data.

[0051] Exemplarily, when the respiratory rate of the first concentrated respiratory monitoring data stored in the first memory space is 25 times / minute, the temperature of the first concentrated environmental monitoring data is 22.8°. At this time, the heart rate of the second concentrated respiratory monitoring data is 20 times / minute, and the temperature of the second concentrated environmental monitoring data is 23°. When the first memory space is used as an index to capture associated keywords, keywords such as "respiratory rate change" and "temperature rise" can be obtained, thereby obtaining the current respiratory state and the trend of environmental changes. Then, the capture results, the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data are added to the second memory space. Therefore, the second memory space stores the data in the first memory space, as well as the capture results of the associated fusion analysis of the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data, and the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data.

[0052] Based on the same acquisition principle as that of the second memory space, the first memory space, the second memory space and the K-1th memory space are used as indexes, and the associated keywords of the K-th concentrated respiratory monitoring data and the K-th concentrated environmental monitoring data are captured, and then the capture results and the K-th concentrated respiratory monitoring data and the K-th concentrated environmental monitoring data are added to the K-th memory space. Among them, the K-th memory space reflects the user's respiratory state and environmental state at the last time point of the preset monitoring window, as well as the user's respiratory state change associated data and environmental state change associated data at different time points within the preset monitoring window. By obtaining the K-th memory space, the changes in the user's respiratory state and environmental state within the preset monitoring window can be precisely grasped.

[0053] In one possible embodiment, multiple sample data sets and multiple sample associated fused multidimensional monitoring feature vectors of multiple sample memory spaces are obtained as training data, and supervised training is performed on the framework constructed based on the feedforward neural network. During the training, the mapping relationship between the data in the memory space and the associated fused multidimensional monitoring feature vector is learned until the training converges, and the trained semantic recognition network layer is obtained. Furthermore, the data in the Kth memory space is input into the semantic recognition network layer for fused feature vector identification to obtain the associated fused multidimensional monitoring feature vector. Exemplarily, the associated fused multidimensional monitoring feature vector can be "stable breathing stage, moderate temperature". The associated fused multidimensional monitoring feature vector reflects the user's breathing state and environmental state within the preset monitoring window.

[0054] Through hierarchical analysis and keyword capture of the user's breathing status and environmental data, the gradually accumulated and integrated monitoring data are fused into a feature vector that reflects the user's actual breathing condition, providing accurate multi-dimensional feature support for subsequent intelligent monitoring strategy adjustments.

[0055] Furthermore, the feature vector obtaining module 14 is used to perform the following steps:

[0056] Extracting a first respiratory monitoring data subsequence from the K respiratory monitoring data subsequences, wherein the first respiratory monitoring data subsequence includes a plurality of first respiratory monitoring data;

[0057] Performing mean processing on the plurality of first respiratory monitoring data to obtain a mean of the first respiratory monitoring data;

[0058] Taking the mean of the first respiratory monitoring data as a starting point, iterating through the plurality of first respiratory monitoring data according to a preset iteration step length to determine first concentrated respiratory monitoring data;

[0059] Performing data concentration trend analysis on the K-1 respiratory monitoring data subsequences to obtain K-1 concentrated respiratory monitoring data;

[0060] Performing data concentration trend analysis on the K environmental monitoring data subsequences to obtain the K centralized environmental monitoring data.

[0061] Further, such as Figure 2 As shown, the feature vector obtaining module 14 is used to perform the following steps:

[0062] Taking the mean of the first respiratory monitoring data as a starting point, iterating through the plurality of first respiratory monitoring data according to the preset iteration step length to obtain iterated first respiratory monitoring data;

[0063] Determine whether the neighborhood concentration density of the iterative first respiratory monitoring data is greater than or equal to the neighborhood concentration density of the mean of the first respiratory monitoring data, and if so, update the iterative first respiratory monitoring data as the starting point;

[0064] If not, calling the rand function to randomly generate a first value and a second value, and when the first value is greater than or equal to the second value, updating the iterative first respiratory monitoring data to the starting point;

[0065] After multiple iterations and updates until the preset number of iterations is met, the starting point corresponding to the maximum value of the neighborhood concentration density during the iteration process is used as the target starting point, and the first respiratory monitoring data corresponding to the target starting point is used as the first concentrated respiratory monitoring data.

[0066] Furthermore, the feature vector obtaining module 14 is used to perform the following steps:

[0067] Constructing a mean neighborhood of the mean of the first respiratory monitoring data based on the preset iteration step, wherein the mean neighborhood includes the first respiratory monitoring data whose difference with the mean of the first respiratory monitoring data is within the preset iteration step among the plurality of first respiratory monitoring data;

[0068] The total number of the first respiratory monitoring data in the mean neighborhood is counted, and the statistical result is compared with the difference between the maximum value of the first respiratory monitoring data and the minimum value of the first respiratory monitoring data in the mean neighborhood to obtain the neighborhood concentration density of the first respiratory monitoring data mean.

[0069] In one embodiment, a first subsequence, i.e., a first respiratory monitoring data subsequence, is selected from the K respiratory monitoring data subsequences. A plurality of data points in the subsequence are averaged, and the average of the first respiratory monitoring data of the subsequence is calculated as the starting point of the analysis. Exemplarily, assuming that the first respiratory monitoring data subsequence is [18, 18, 17, 15, 19], the average calculation result is 17.4.

[0070] Taking the mean of the first respiratory monitoring data as the starting point, iterating in the plurality of first respiratory monitoring data according to the preset iteration step length to obtain iterated first respiratory monitoring data. The preset iteration step length is a difference value of the first respiratory monitoring data for a single iteration pre-set by a person skilled in the art.

[0071] Preferably, a mean neighborhood of the mean of the first respiratory monitoring data is constructed based on the preset iteration step, wherein the mean neighborhood includes the first respiratory monitoring data whose difference with the mean of the first respiratory monitoring data among the multiple first respiratory monitoring data is within the preset iteration step. Further, the total number of the first respiratory monitoring data within the mean neighborhood is counted, and the statistical result is compared with the difference between the maximum value of the first respiratory monitoring data and the minimum value of the first respiratory monitoring data in the mean neighborhood to obtain the neighborhood concentration density of the mean of the first respiratory monitoring data. The neighborhood concentration density of the mean of the first respiratory monitoring data reflects the distribution density of the means of the first respiratory monitoring data gathered around the mean of the first respiratory monitoring data. The larger the neighborhood concentration density, the more representative the data is. Based on the same principle, the neighborhood concentration density of the iterative first respiratory monitoring data is obtained.

[0072] Furthermore, determine whether the neighborhood concentration density of the iterative first respiratory monitoring data is greater than or equal to the neighborhood concentration density of the mean of the first respiratory monitoring data. If so, it indicates that the iteratively obtained iterative first respiratory monitoring data is more representative. At this time, the iterative first respiratory monitoring data is updated to the starting point.

[0073] If not, it indicates that the mean of the first respiratory monitoring data is more representative. However, in order to avoid falling into a local optimal solution, two random numbers are generated by the rand function, and the sizes of the two random numbers are compared to determine whether it is necessary to accept an inferior solution during the iteration process.

[0074] Optionally, call the rand function to randomly generate a first value and a second value. When the first value is greater than or equal to the second value, accept the inferior solution and update the iterative first respiratory monitoring data to the starting point. After multiple iterations, until the preset number of iterations (the maximum number of iterations pre-set by a person skilled in the art) is met, the starting point corresponding to the maximum value of the neighborhood concentration density during the iteration is used as the target starting point, and the first respiratory monitoring data corresponding to the target starting point is used as the first concentrated respiratory monitoring data. When the first value is less than the second value, the inferior solution is not accepted, and the iteration is stopped at this time, and the mean of the first respiratory monitoring data is used as the first concentrated respiratory monitoring data.

[0075] Based on the same principle as that of obtaining the first centralized respiratory monitoring data, data concentration trend analysis is performed on K-1 respiratory monitoring data subsequences and K environmental monitoring data subsequences to obtain the K-1 centralized respiratory monitoring data and K centralized environmental monitoring data. The technical effect of performing centralized trend analysis on the data and reducing the amount of data analysis is achieved.

[0076] The abnormal record set acquisition module 15 is used to obtain the user basic features of the target user, perform data mining in the big data in combination with the associated fusion multi-dimensional monitoring feature vector, and retrieve and obtain the matching monitoring abnormal record set;

[0077] In a possible embodiment, the target user is any user who needs to undergo abnormal breathing analysis. Collect and record the basic information of the target user, including but not limited to age, gender, health status (such as whether there are chronic diseases), daily living habits, etc. Exemplary: The target user is a 65-year-old male with mild hypertension. This information is recorded as the basic features for subsequent analysis and is used to screen and match abnormal records. Retrieve the obtained associated fusion multidimensional monitoring feature vector, which contains multidimensional information about the current user's breathing state and environmental influences. Example: The feature vector may include information such as breathing frequency, breathing rate, temperature influence, etc., to form a comprehensive breathing feature of the current user.

[0078] Preferably, a big data analysis platform is used to perform data mining in combination with the user's basic features and the monitoring feature vectors to retrieve historical abnormal records with similar features in the big database. For example, if the feature vector shows frequent respiratory frequency fluctuations in breathing, the system can filter out historical records with similar respiratory frequency fluctuations in the database.

[0079] According to the basic characteristics of the user and the monitoring feature vector, set the matching conditions, such as similar respiratory rate fluctuations, minute ventilation volume, ambient temperature range, etc. The matching conditions should also take into account the personalized characteristics of the target user.

[0080] Based on the set matching conditions, the qualified abnormal monitoring records are retrieved from the database. Each abnormal record should have similar characteristics to the current user monitoring data. In the initially screened abnormal record set, the incompletely matched records are further removed, and the most relevant abnormal data are retained to ensure the accuracy of the matching set, and the matching monitoring abnormal record set is obtained. For example, similar abnormal records of users are sorted by frequency of occurrence or importance, and more common or more influential abnormal records are retained first, so that subsequent abnormal analysis is more accurate.

[0081] By effectively identifying abnormal situations that are similar to the user's current monitoring data, it provides a basis for subsequent monitoring, early warning, and dynamic regulation, and provides reliable reference data for personalized respiratory data monitoring.

[0082] An asynchronous monitoring step set acquisition module 16 is used to perform cluster analysis on the matching monitoring anomaly record set with the anomaly type as an index to determine an asynchronous monitoring step set, wherein the asynchronous monitoring step set includes a first monitoring step and a second monitoring step, and the first monitoring step is less than or equal to the second monitoring step;

[0083] The monitoring data set acquisition module 17 is used to control the target monitoring device to perform asynchronous breathing monitoring on the target user within the feedback monitoring window according to the first monitoring step size and the second monitoring step size, so as to obtain a first breathing monitoring data set and a second breathing monitoring data set.

[0084] Furthermore, the asynchronous monitoring step set acquisition module 16 is used to perform the following steps:

[0085] Taking the anomaly type as an index, cluster analysis is performed on the matching monitoring anomaly record set to determine a plurality of cluster matching monitoring anomaly record sets;

[0086] Traversing the plurality of cluster matching monitoring abnormal record sets to extract the average value of abnormal interval duration, and obtaining a plurality of cluster monitoring abnormal interval duration average value sets;

[0087] Respectively extracting the maximum and minimum values ​​of the plurality of cluster monitoring abnormal interval time mean value sets, and determining the first monitoring step length and the second monitoring step length according to the extraction results;

[0088] The first monitoring step length and the second monitoring step length are used as the asynchronous monitoring step length set.

[0089] Furthermore, the asynchronous monitoring step set acquisition module 16 is used to perform the following steps:

[0090] Respectively extracting the minimum values ​​of the plurality of cluster monitoring abnormality interval duration mean value sets to obtain the plurality of cluster monitoring abnormality interval duration mean minimum values, and calculating the mean of the plurality of cluster monitoring abnormality interval duration mean minimum values ​​to obtain the first monitoring step length;

[0091] The maximum values ​​of the plurality of cluster monitoring abnormality interval duration mean value sets are extracted respectively to obtain the plurality of cluster monitoring abnormality interval duration mean value maximum values, and the mean of the plurality of cluster monitoring abnormality interval duration mean value maximum values ​​is calculated to obtain the second monitoring step.

[0092] In one possible embodiment, the abnormal data in the matching monitoring abnormal record set are indexed by abnormal type (such as abnormal breathing rate, abnormal breathing waveform, frequent body movement, etc.), so that abnormal records of the same type can be more accurately classified during cluster analysis, and then after comprehensive analysis, the asynchronous monitoring step set is determined, wherein the asynchronous monitoring step set includes a first monitoring step and a second monitoring step. And the first monitoring step is less than or equal to the second monitoring step, ensuring simultaneous monitoring with a shorter monitoring interval and a longer monitoring interval, ensuring monitoring accuracy and response timeliness.

[0093] Perform cluster analysis on the abnormal record set under each abnormal type, cluster similar abnormal events together to determine the distribution of abnormal data of different categories. For example, under the type of "abnormal fluctuation of respiratory frequency", the abnormalities of high-frequency respiratory frequency fluctuation and low-frequency respiratory frequency fluctuation can be separated by cluster analysis to form multiple cluster sets. Traverse the abnormal records in each cluster, calculate the mean of the abnormal occurrence interval of each cluster, and obtain the mean set of cluster monitoring abnormal interval duration.

[0094] For example, for the “abnormal respiratory rate fluctuation” cluster set, if the intervals between multiple abnormal respiratory rate fluctuation events are 10 minutes, 12 minutes, and 15 minutes, respectively, the average interval of the cluster is 12.3 minutes.

[0095] The maximum and minimum values ​​of the mean value sets of each cluster monitoring abnormal interval duration are extracted respectively to determine the step range of asynchronous monitoring. The minimum value of the plurality of cluster monitoring abnormal interval duration mean values ​​is extracted to obtain the minimum values ​​of the plurality of cluster monitoring abnormal interval duration mean values, and the mean value of the minimum values ​​of the plurality of cluster monitoring abnormal interval duration mean values ​​is calculated and used as the first monitoring step.

[0096] Preferably, the maximum values ​​of the sets of the mean values ​​of the multiple cluster monitoring abnormality interval durations are extracted respectively to obtain the maximum values ​​of the mean values ​​of the multiple cluster monitoring abnormality interval durations, and the mean value of the maximum values ​​of the mean values ​​of the multiple cluster monitoring abnormality interval durations is calculated and used as the second monitoring step.

[0097] In one embodiment, the asynchronous breathing monitoring frequency of the target monitoring device is set according to the first monitoring step length and the second monitoring step length, and the target monitoring device is controlled to perform asynchronous breathing monitoring on the target user. The technical effect of adjusting the monitoring frequency of the user's breathing condition and improving the intelligence and pertinence of monitoring is achieved on the basis of balancing the monitoring accuracy and the energy consumption of the device.

[0098] The breathing abnormality analysis result obtaining module 18 is used to perform abnormality analysis on the first breathing monitoring data set and the second breathing monitoring data set using a breathing abnormality analyzer to obtain a breathing abnormality analysis result.

[0099] In one possible embodiment, the respiratory abnormality analyzer is used to perform an overall intelligent analysis of the respiratory monitoring data of different monitoring frequencies to obtain the respiratory abnormality analysis results. Preferably, multiple sample first respiratory monitoring data sets and multiple sample second respiratory monitoring data sets, as well as corresponding multiple sample respiratory abnormality analysis results are obtained as training sample data. The training sample data is used to perform supervised training on a framework built based on a feedforward neural network to learn the two-to-one mapping relationship between the first respiratory monitoring data set and the second respiratory monitoring data set and the respiratory abnormality analysis results until the training converges, thereby obtaining the trained respiratory abnormality analyzer.

[0100] The first respiratory monitoring data set and the second respiratory monitoring data set are input into the respiratory abnormality analyzer for abnormality analysis to obtain the respiratory abnormality analysis result, wherein the respiratory abnormality analysis result reflects the user's respiratory abnormality within the feedback monitoring window.

[0101] In summary, the embodiments of the present application have at least the following technical effects:

[0102] The present application realizes the correlation analysis between environmental changes and respiratory status by synchronously collecting respiratory monitoring data and environmental monitoring data, so that the device can more accurately reflect the user's actual respiratory status. Then, K respiratory monitoring data subsequences are used as foreground data and K environmental monitoring subsequences are used as background data for correlation fusion analysis. The generated multi-dimensional monitoring feature vector can accurately characterize the respiratory state characteristics, which is helpful to identify anomalies. Two monitoring steps are used for simultaneous monitoring to ensure that the abnormal conditions reflected by the important respiratory monitoring data are captured in time, thereby achieving the technical effect of improving the reliability of the analysis of user respiratory abnormalities and effectively avoiding the false alarms and omissions caused by standardized monitoring.

[0103] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0104] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0105] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A respiratory abnormality analysis system for respiratory data monitoring and analysis, characterized in that: The system comprises: A monitoring data sequence acquisition module is used to interact with the target monitoring device and the environmental monitoring module to extract the monitoring data of the preset monitoring window, and obtain a respiratory monitoring data sequence and an environmental monitoring data sequence, wherein the respiratory monitoring data corresponds to the environmental monitoring data one by one; A respiratory monitoring data subsequence obtaining module, used to conditionally select K-1 respiratory monitoring data from the respiratory monitoring data sequence as K-1 starting division points, divide the respiratory monitoring data sequence with the K-1 starting division points, and obtain K respiratory monitoring data subsequences, wherein K is an integer greater than or equal to 3; An environment monitoring data subsequence obtaining module, used for dividing the environment monitoring data sequence into subsequences based on the K respiratory monitoring data subsequences to obtain K environment monitoring data subsequences; A feature vector acquisition module is used to use the K respiratory monitoring data subsequences as foreground data and the K environmental monitoring data subsequences as background data, perform association fusion analysis in sequence, and obtain an association fusion multi-dimensional monitoring feature vector; The abnormal record set acquisition module is used to obtain the basic user characteristics of the target user, perform data mining in the big data in combination with the associated fusion multi-dimensional monitoring feature vector, and retrieve and obtain the matching monitoring abnormal record set; An asynchronous monitoring step set acquisition module is used to perform cluster analysis on the matching monitoring anomaly record set with the anomaly type as an index to determine an asynchronous monitoring step set, wherein the asynchronous monitoring step set includes a first monitoring step and a second monitoring step, and the first monitoring step is less than or equal to the second monitoring step; a monitoring data set acquisition module, configured to control the target monitoring device to perform asynchronous breathing monitoring on the target user within a feedback monitoring window according to the first monitoring step length and the second monitoring step length, respectively, to obtain a first breathing monitoring data set and a second breathing monitoring data set; The breathing abnormality analysis result obtaining module is used to use a breathing abnormality analyzer to perform abnormality analysis on the first breathing monitoring data set and the second breathing monitoring data set to obtain a breathing abnormality analysis result.

2. A respiratory abnormality analysis system for respiratory data monitoring and analysis as claimed in claim 1, characterized in that: The feature vector acquisition module is used to perform the following steps: Traversing the K respiratory monitoring data subsequences and the K environmental monitoring data subsequences to perform data concentration trend analysis respectively, and determining K concentrated respiratory monitoring data and K concentrated environmental monitoring data; Respectively extracting the first centralized respiratory monitoring data and the first centralized environmental monitoring data that are at the top of the K centralized respiratory monitoring data and the K centralized environmental monitoring data, and using the first centralized respiratory monitoring data and the first centralized environmental monitoring data as the first memory space; Extracting the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data located at the second position among the K concentrated respiratory monitoring data and the K concentrated environmental monitoring data again, capturing associated keywords for the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data with the first memory space as an index, and adding the capture result and the second concentrated respiratory monitoring data and the second concentrated environmental monitoring data into the second memory space; Extract the K-th concentrated respiratory monitoring data and the K-th concentrated environmental monitoring data, respectively use the first memory space, the second memory space to the K-1-th memory space as indexes, capture the associated keywords of the K-th concentrated respiratory monitoring data and the K-th concentrated environmental monitoring data, and add the capture results and the K-th concentrated respiratory monitoring data and the K-th concentrated environmental monitoring data into the K-th memory space; The semantic recognition network layer is used to perform fusion feature vector recognition on the data in the Kth memory space to obtain the associated fusion multi-dimensional monitoring feature vector.

3. A respiratory abnormality analysis system for respiratory data monitoring and analysis as claimed in claim 2, characterized in that: The feature vector acquisition module is used to perform the following steps: Extracting a first respiratory monitoring data subsequence from the K respiratory monitoring data subsequences, wherein the first respiratory monitoring data subsequence includes a plurality of first respiratory monitoring data; Performing mean processing on the plurality of first respiratory monitoring data to obtain a mean of the first respiratory monitoring data; Taking the mean of the first respiratory monitoring data as a starting point, iterating through the plurality of first respiratory monitoring data according to a preset iteration step length to determine first concentrated respiratory monitoring data; Performing data concentration trend analysis on the K-1 respiratory monitoring data subsequences to obtain K-1 concentrated respiratory monitoring data; Performing data concentration trend analysis on the K environmental monitoring data subsequences to obtain the K centralized environmental monitoring data.

4. A respiratory abnormality analysis system for monitoring and analyzing respiratory data as claimed in claim 3, characterized in that: The feature vector acquisition module is used to perform the following steps: Taking the mean of the first respiratory monitoring data as a starting point, iterating through the plurality of first respiratory monitoring data according to the preset iteration step length to obtain iterated first respiratory monitoring data; Determine whether the neighborhood concentration density of the iterative first respiratory monitoring data is greater than or equal to the neighborhood concentration density of the mean of the first respiratory monitoring data, and if so, update the iterative first respiratory monitoring data as the starting point; If not, calling the rand function to randomly generate a first value and a second value, and when the first value is greater than or equal to the second value, updating the iterative first respiratory monitoring data to the starting point; After multiple iterations and updates until the preset number of iterations is met, the starting point corresponding to the maximum value of the neighborhood concentration density during the iteration process is used as the target starting point, and the first respiratory monitoring data corresponding to the target starting point is used as the first concentrated respiratory monitoring data.

5. A respiratory abnormality analysis system for monitoring and analyzing respiratory data according to claim 4, characterized in that: The feature vector acquisition module is used to perform the following steps: Constructing a mean neighborhood of the mean of the first respiratory monitoring data based on the preset iteration step, wherein the mean neighborhood includes the first respiratory monitoring data whose difference with the mean of the first respiratory monitoring data is within the preset iteration step among the plurality of first respiratory monitoring data; The total number of the first respiratory monitoring data in the mean neighborhood is counted, and the statistical result is compared with the difference between the maximum value of the first respiratory monitoring data and the minimum value of the first respiratory monitoring data in the mean neighborhood to obtain the neighborhood concentration density of the first respiratory monitoring data mean.

6. A respiratory abnormality analysis system for monitoring and analyzing respiratory data as claimed in claim 1, characterized in that: The asynchronous monitoring step set acquisition module is used to perform the following steps: Taking the anomaly type as an index, cluster analysis is performed on the matching monitoring anomaly record set to determine a plurality of cluster matching monitoring anomaly record sets; Traversing the plurality of cluster matching monitoring anomaly record sets to extract the mean of the anomaly interval duration, and obtaining a plurality of cluster monitoring anomaly interval duration mean sets; Respectively extracting the maximum and minimum values ​​of the plurality of cluster monitoring abnormal interval time mean value sets, and determining the first monitoring step length and the second monitoring step length according to the extraction results; The first monitoring step length and the second monitoring step length are used as the asynchronous monitoring step length set.

7. A respiratory abnormality analysis system for monitoring and analyzing respiratory data according to claim 6, characterized in that: The asynchronous monitoring step set acquisition module is used to perform the following steps: Respectively extracting the minimum values ​​of the plurality of cluster monitoring abnormality interval duration mean value sets to obtain the plurality of cluster monitoring abnormality interval duration mean minimum values, and calculating the mean of the plurality of cluster monitoring abnormality interval duration mean minimum values ​​to obtain the first monitoring step length; The maximum values ​​of the plurality of cluster monitoring abnormality interval duration mean value sets are extracted respectively to obtain the plurality of cluster monitoring abnormality interval duration mean value maximum values, and the mean of the plurality of cluster monitoring abnormality interval duration mean value maximum values ​​is calculated to obtain the second monitoring step.

8. A respiratory abnormality analysis system for monitoring and analyzing respiratory data as claimed in claim 1, characterized in that: The respiratory monitoring data subsequence acquisition module is used to perform the following steps: Randomly select K-1 respiratory monitoring data from the respiratory monitoring data sequence as K-1 first division points; Performing left and right neighbor similarity analysis on the K-1 first division points to determine the K-1 neighbor similarities of the K-1 first division points; Determine whether the K-1 nearest neighbor similarities are all less than or equal to a preset similarity difference threshold, if not, reselect K-1 first division points; If so, the K-1 first division points are used as the K-1 starting division points.

Images