A breathing detection method and system for improving data synchronization

Abstract

The present invention provides a breathing detection method and system for improving data synchronization, and relates to the field of data processing technology. The method comprises: performing human body modeling on a detection target user, obtaining a three-dimensional human body model, interacting with a breathing sound detection server, obtaining a plurality of breathing sound time domain information of a breathing sound source position and a plurality of breathing sound detection positions, generating a plurality of breathing sound conduction paths for time delay analysis, generating a plurality of time delay characteristic values ​​for performing time domain synchronization on a plurality of breathing sound time domain information, obtaining breathing sound time domain synchronization information, and sending the information to a breathing sound detection client. The present invention solves the technical problem in the prior art that the detection of breathing sounds at different positions has time delay, resulting in poor data synchronization of breathing sounds, and low accuracy of breathing detection, and realizes fitting the time delay of detecting breathing sounds at different positions, thereby improving data synchronization of breathing sounds and the accuracy of breathing detection.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a breathing detection method and system for improving data synchronization. Background Art

[0002] Breathing is one of the most basic life activities of human beings. Each person needs to take at least 20,000 breaths to complete a day's life activities. With the development of modern society, people's living standards have gradually improved, and the prevalence of respiratory diseases has become increasingly high. Timely and accurate detection of respiratory status has become one of the important issues in modern medicine. However, in the prior art, the detection of respiratory sounds at different positions has a time delay, resulting in poor synchronization of respiratory sound data, resulting in a technical problem of low accuracy of respiratory detection. Summary of the invention

[0003] The present application provides a breathing detection method and system for improving data synchronization, which is used to solve the technical problem in the prior art that the detection of breathing sounds at different positions has a time delay, resulting in poor data synchronization of breathing sounds and low accuracy of breathing detection.

[0004] In view of the above problems, the present application provides a breathing detection method and system for improving data synchronization.

[0005] In a first aspect, the present application provides a breathing detection method for improving data synchronization, the method comprising: performing human body modeling on a detection target user to obtain a three-dimensional human body model; interacting with a breathing sound detection server to obtain a number of breathing sound time domain information of a breathing sound source position and a number of breathing sound detection positions; marking the several breathing sound detection positions and the breathing sound source positions on the three-dimensional human body model to generate a number of breathing sound conduction paths; traversing the several breathing sound conduction paths to perform time delay analysis to generate a number of time delay feature values; performing time domain synchronization on the several breathing sound time domain information according to the several time delay feature values ​​to obtain breathing sound time domain synchronization information, and sending it to a breathing sound detection client.

[0006] In a second aspect, the present application provides a breathing detection system for improving data synchronization, the system comprising: a model building module, the model building module is used to perform human body modeling for the detection target user to obtain a three-dimensional human body model; a position acquisition module, the position acquisition module is used to interact with a breathing sound detection server to obtain a number of breathing sound time domain information of the breathing sound source position and a number of breathing sound detection positions; a first identification module, the first identification module is used to identify the several breathing sound detection positions and the breathing sound source positions in the three-dimensional human body model to generate a number of breathing sound conduction paths; a first analysis module, the first analysis module is used to traverse the several breathing sound conduction paths to perform time delay analysis and generate a number of delay feature values; a first information acquisition module, the first information acquisition module is used to perform time domain synchronization on the several breathing sound time domain information according to the several delay feature values, obtain breathing sound time domain synchronization information, and send it to the breathing sound detection client.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] The present application provides a breathing detection method and system for improving data synchronization, which relates to the field of data processing technology. It solves the technical problem in the prior art that the detection of breathing sounds at different positions has a time delay, resulting in poor data synchronization of breathing sounds and low accuracy of breathing detection. It achieves the time delay of fitting breathing sounds detected at different positions, improves the data synchronization of breathing sounds and the accuracy of breathing detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic flow chart of a breathing detection method for improving data synchronization is provided for this application;

[0010] Figure 2 A schematic diagram of the structure of a breathing detection system for improving data synchronization is provided for this application.

[0011] Explanation of the accompanying drawings: platform construction module 11, first connection module 12, sequence generation module 13, feature recognition module 14, first analysis module 15, scheduling module 16. DETAILED DESCRIPTION

[0012] The present application provides a breathing detection method and system for improving data synchronization, so as to solve the technical problem in the prior art that the detection of breathing sounds at different positions has a time delay, resulting in poor synchronization of breathing sound data and low accuracy of breathing detection.

[0013] Embodiment 1

[0014] like Figure 1As shown, an embodiment of the present application provides a breathing detection method for improving data synchronization, the method comprising:

[0015] Step A100: Performing human body modeling on the target user to obtain a three-dimensional human body model;

[0016] Furthermore, step A100 of the present application also includes:

[0017] Step A110: Obtain the age characteristic value, gender characteristic value, height characteristic value, weight characteristic value, head-to-body ratio characteristic value, waist-to-leg ratio characteristic value, upper body thickness characteristic value and upper body width characteristic value of the detected target user, and construct an 8-dimensional constraint matrix;

[0018] Step A120: performing a correlation analysis based on the 8-dimensional constraint matrix to determine the organ size identifier and the skeleton size identifier;

[0019] Step A130: collecting the external dimension identification of the detection target user;

[0020] Step A140: activating the 3D modeling network, processing the organ size identifier, the skeleton size identifier and the outer shape size identifier, and generating the three-dimensional model of the human body.

[0021] Furthermore, step A120 of the present application includes:

[0022] Step A121: configuring a first organ size identification type set;

[0023] Step A122: using the 8-dimensional constraint matrix and the first organ size identification type set as constraints, collecting a first organ size identification matrix, wherein the row attribute of the first organ size identification matrix is ​​the size identification type, and the column attribute is the size identification record value;

[0024] Step A123: performing a correlation analysis based on the first organ size identification matrix to generate a first organ size identification, and adding the first organ size identification to the organ size identification, including:

[0025] Step A124: obtaining the k-th row size identification type and the k-th row size identification record value set according to the first organ size identification matrix, where k belongs to any row of the first organ size identification matrix;

[0026] Step A125: performing a central tendency analysis on the k-th row size identification record value set, generating a size feature identification of the k-th row size identification type, and adding the identification to the first organ size identification;

[0027] The analysis process of the skeleton size identification and the organ size identification is the same.

[0028] In the present application, a breathing detection method for improving data synchronization provided in an embodiment of the present application is applied to a breathing detection system for improving data synchronization. In order to more accurately detect the breathing of a target user, the breathing detection system for improving data synchronization first uses the target user to be detected as basic reference data to perform human body modeling of the target user. The method is to first determine the age characteristic value, gender characteristic value, height characteristic value, weight characteristic value, head-to-body ratio characteristic value, waist-to-leg ratio characteristic value, upper body thickness characteristic value and upper body width characteristic value of the target user, extract the age feature, height feature, weight feature, head-to-body ratio feature, waist-to-leg ratio feature, upper body thickness feature and upper body width feature of the target user, and respectively Correspondingly, the age, height, weight, head-to-body ratio, waist-to-leg ratio, upper body thickness and upper body width of the target user are limited to the highest or lowest extreme value range, and the user gender is limited to male or female. The constraints are adjusted and modified to construct an 8-dimensional constraint matrix. At the same time, the horizontal and vertical columns in the 8-dimensional constraint matrix have a one-to-one correspondence with the age characteristic value, gender characteristic value, height characteristic value, weight characteristic value, head-to-body ratio characteristic value, waist-to-leg ratio characteristic value, upper body thickness characteristic value and upper body width characteristic value of the target user. Furthermore, the 8-dimensional constraint matrix is ​​subjected to correlation analysis. Since different organs correspond to different sizes, such as the chest cavity for thickness and width, and the intestine for inner diameter, on this basis, multiple organ size identification types are identified. The set is configured, and each organ size identification type has a corresponding relationship with the organ size of an organ, so as to randomly extract an organ size identification type set and record it as the first organ size identification type set. Further, the 8-dimensional constraint matrix and the first organ size identification type set are used as constraints, the size identification type corresponding to the first organ is collected in the first organ size identification type set, and the size identification record value corresponding to the first organ is collected in the 8-dimensional constraint matrix. At the same time, the size identification type is used as a row attribute and the size identification record value is used as a column attribute to construct a first organ size identification matrix. Then, the correlation analysis is performed according to the first organ size identification matrix, which means that the first organ size identification matrix is ​​traversed as the basic reference data to obtain the first organ size identification matrix. The k-th row size identification type and the k-th row size identification record value set included in the organ size identification matrix, k belongs to any row of the first organ size identification matrix, wherein k is an integer greater than or equal to 1. Further, a central tendency analysis is performed on the k-th row size identification record value set, and the data dispersion degree in the k-th row size identification record value set is recorded by calculating the standard deviation of the k-th row size identification record value set. A larger standard deviation means that the data is more dispersed, and a smaller standard deviation means that the data is closer to the average value. On this basis, a size feature identification of the k-th row size identification type is generated for presenting the organ size feature of the corresponding k-th organ, and the first organ size identification is added to iterate thereby to obtain the organ size identification of the target user to be detected.The analysis process of the skeleton size identification and the organ size identification is the same, so the organ size identification and the skeleton size identification are determined respectively.

[0029] Furthermore, the external dimension identification of the detection target user is collected, and the external dimension refers to the human body external dimension of the detection target user, and then the 3D modeling network is activated. The 3D modeling network refers to a network based on deep learning technology, which is used to generate and process three-dimensional models. The 3D modeling network can automatically learn and generate new three-dimensional models by learning a large amount of three-dimensional shape data. Because under normal circumstances, the distribution positions of the organs and skeletons of the detection target users are fixed, the size identification can be determined based on the organ size identification, skeleton size identification and external dimension identification, that is, the corresponding three-dimensional model can be constructed through intelligent model processing, and finally the organ size identification, skeleton size identification and external dimension identification are processed by the 3D modeling network to generate a three-dimensional model of the human body for output, which serves as an important reference for the later realization of breathing detection based on improving data synchronization.

[0030] Step A200: Interacting with the breathing sound detection server, obtaining a plurality of breathing sound time domain information of the breathing sound source position and a plurality of breathing sound detection positions;

[0031] In the present application, in order to better improve the data synchronization of breathing detection, firstly, by interacting with the breathing sound detection server, the breathing sound detection server is connected to a breathing detection system for improving data synchronization, and the server-side program for processing and analyzing breathing sound data is usually used to receive the breathing sound data uploaded from the client such as mobile devices, sensors, etc., and perform real-time or offline processing and analysis, and collect the breathing sound data of the target user by using a sound receiving device, and the collected breathing sound data can be a time domain signal, and the collected breathing sound data is removed Background noise, filtering, noise reduction and other operations are performed to improve signal quality. Furthermore, features are extracted from the preprocessed signal, which may include time domain features, such as amplitude, energy, duration, etc. By analyzing the time domain features through a machine learning classifier, the source location of respiratory sounds can be identified, such as bronchial respiratory sounds, bronchoalveolar respiratory sounds, alveolar respiratory sounds, tracheal respiratory sounds, etc., and by analyzing the entire respiratory sound signal, the start and end positions of the respiratory sounds are determined in the time domain, and several respiratory sound time domain information and several respiratory sound detection positions are obtained, thereby ensuring the realization of respiratory detection based on improving data synchronization.

[0032] Step A300: marking the plurality of respiratory sound detection positions and the respiratory sound source positions on the three-dimensional human body model to generate a plurality of respiratory sound conduction paths;

[0033] In the present application, the respiratory sound detection positions and the respiratory sound source positions are identified by synchronizing several respiratory sound detection positions and the respiratory sound source positions into the three-dimensional human body model constructed as above. Under normal circumstances, the respiratory sound source is located in the respiratory system, such as the throat, trachea, bronchi, etc., and then the corresponding area or point is selected on the three-dimensional model and marked as the respiratory sound source position. Sensors are placed on the chest, abdomen, etc., or a microphone is used for sound collection to determine the position where respiratory sound detection is required, and the corresponding area or point is selected on the three-dimensional model and marked as the respiratory sound detection position. According to the identified respiratory sound source position and respiratory sound detection position, the path between the source position and the detection position is visualized on the three-dimensional model by using relevant software or programming tools to draw a connecting line or path to generate a respiratory sound conduction path, thereby laying a solid foundation for the subsequent realization of respiratory detection based on improving data synchronization.

[0034] Step A400: traversing the plurality of respiratory sound conduction paths to perform delay analysis and generate a plurality of delay characteristic values;

[0035] Furthermore, step A400 of the present application also includes:

[0036] Step A410: extracting a first respiratory sound conduction path according to the plurality of respiratory sound conduction paths, wherein the first respiratory sound conduction path has a conduction medium attribute label sequence and a conduction distance label sequence;

[0037] Step A420: performing frequency mining according to the conduction medium attribute label sequence and the conduction distance label sequence to obtain a frequent transmission duration label sequence;

[0038] Step A430: Generate a first time delay characteristic value of the first respiratory sound conduction path according to the frequent transmission duration label sequence, and add it to the plurality of time delay characteristic values.

[0039] Furthermore, step A420 of the present application includes:

[0040] Step A421: obtaining a first conductive medium attribute label according to the conductive medium attribute label sequence, and obtaining a first conductive distance label according to the conductive distance label sequence, wherein the first conductive medium attribute label is solid or liquid or space;

[0041] Step A422: collecting a sound transmission time feature set with the preset human body characteristic temperature, the first conduction medium attribute label and the first conduction distance label as constraints;

[0042] Step A423: Perform outlier factor analysis on the sound transfer duration feature set, generate a first frequent transfer duration label, and add it to the frequent transfer duration label sequence.

[0043] In the present application, by sequentially traversing each of the several respiratory sound conduction paths obtained by the above construction to perform delay analysis, firstly, by randomly selecting from several respiratory sound conduction paths, a first respiratory sound conduction path is extracted, and there is a conduction medium attribute label sequence and a conduction distance label sequence in the first respiratory sound conduction path, the conduction medium attribute label sequence refers to the type of medium for conducting the first respiratory sound in the human body, which is obtained by sorting according to the medium attribute, and the conduction distance label sequence refers to the distance of the first respiratory sound from the initial point of the first respiratory sound to the end point in the human body through the conduction medium, which is obtained by arranging from small to large. Furthermore, frequent mining is performed according to the conduction medium attribute label sequence and the conduction distance label sequence, which refers to randomly extracting the first conduction medium attribute label sequence to generate a first conduction medium attribute label sequence. A first conduction distance label is generated by randomly extracting a conduction distance label sequence, wherein the first conduction medium attribute label is solid or liquid or space, and the human body characteristic temperature is preset by detecting the average body temperature of the target user. The first conduction medium attribute label and the first conduction distance label are used as constraints, and the sound transmission time of the first breath sound in the body of the target user is recorded. On this basis, a sound transmission time feature set is obtained. Further, an outlier factor analysis is performed on the sound transmission time feature set. The outlier factor score of each sound transmission time feature sample in the sound transmission time feature set can be calculated by an outlier factor algorithm, such as LOF, CBLOF, etc. The outlier factor is a numerical value indicating the degree of sample outlier. A larger value indicates a more abnormal or outlier. A suitable outlier factor threshold is set based on the historical outlier boundary value. Samples exceeding the threshold will be considered as outliers. According to the set threshold, samples with outlier factor scores exceeding the threshold will be marked as outliers. According to the distribution of outliers, that is, the number and density of outliers and other indicators, it is judged whether there are transmission durations with the first frequent transmission, and then the first frequent transmission duration label is generated and added to the frequent transmission duration label sequence. At the same time, the frequent transmission duration label sequence is obtained.

[0044] The first respiratory sound is traversed and matched according to the frequently transmitted time-length label sequence, and the first time delay characteristic value corresponding to the first respiratory sound conduction path is generated according to the matching result. The first time delay characteristic value is then added to a plurality of time delay characteristic values, thereby iteratively obtaining a plurality of time delay characteristic values ​​for output, which has a limited effect on realizing respiratory detection based on improving data synchronization.

[0045] Furthermore, step A423 of the present application includes:

[0046] Step A4231: traverse the sound transmission duration feature set to construct an outlier analysis neighborhood set;

[0047] Step A4232: Construct outlier analysis function:

[0048] ;

[0049] in, The outlier coefficient that characterizes the duration of the i-th sound transmission, Characterizes the characteristic value of the i-th sound transmission duration, The kth neighborhood feature value representing the neighborhood of the i-th sound transmission duration feature, Q represents the neighborhood size, and M represents the size of the sound transmission duration feature set;

[0050] Step A4233: performing outlier analysis on the sound transmission duration feature set according to the outlier analysis neighborhood set, the outlier coefficient threshold and the outlier analysis function to generate an outlier transmission duration feature set;

[0051] Step A4234: Delete the outlier transfer duration feature set from the sound transfer duration feature set, average the remaining transfer duration feature sets, and obtain the first frequent transfer duration label.

[0052] Further, step A4231 of the present application includes:

[0053] Step A42311: construct a neighborhood size Q value, where M>Q≥5, and Q is an integer;

[0054] Step A42312: Based on the neighborhood size Q value, for the first sound transmission duration feature value of the sound transmission duration feature set, filter the first outlier analysis neighborhood from near to far from the sound transmission duration feature set, and add it into the outlier analysis neighborhood set.

[0055] In the present application, by performing outlier factor analysis on the sound transmission time feature set, it means first traversing the sound transmission time features contained in the sound transmission time feature set in turn, and selecting appropriate neighborhood parameters according to the data distribution and application scenarios of the sound transmission time feature set. The neighborhood parameters determine the size of the neighborhood around each sample, which can affect the calculation result of the outlier factor, and then calculating the distance matrix between each sample through the distance measurement method for each sample, selecting the z samples closest to its neighborhood to form a neighborhood set of the sample, calculating the distance from all points in its neighborhood set to the sample, and then sorting these distances, and selecting the zth distance pair to be outlier. A group analysis neighborhood set is constructed, and further, a neighborhood scale Q value is constructed, wherein M>Q≥5, Q is an integer, and M represents the scale of the sound transmission time feature set. Based on the neighborhood scale Q value, the first sound transmission time feature value in the sound transmission time feature set is screened from near to far from the sound transmission time feature set as the first outlier analysis neighborhood corresponding to the first sound transmission time feature value, and it is added to the outlier analysis neighborhood set, thereby iteratively generating an outlier analysis neighborhood set, and then using the outlier analysis neighborhood set as the basic data, it is combined with the outlier coefficient threshold and the outlier analysis function to perform outlier analysis on the sound transmission time feature set. The outlier analysis function is shown as follows:

[0056] ;

[0057] in, The outlier coefficient that characterizes the duration of the i-th sound transmission, Characterizes the characteristic value of the i-th sound transmission duration, The kth neighborhood feature value representing the neighborhood of the i-th sound transmission duration feature, Q represents the neighborhood size, and M represents the size of the sound transmission duration feature set;

[0058] The outlier coefficient of each sound transfer time feature in the sound transfer time feature set is obtained through the outlier analysis function, and feature extraction is performed based on the outlier coefficient of each sound transfer time feature to determine the outlier transfer time feature set. Furthermore, the outlier transfer time feature set is identified and deleted from the sound transfer time feature set, and the remaining transfer time feature sets are mean calculated. The first frequent transfer duration label is determined according to the mean result of the remaining transfer time feature sets, so as to be used as reference data for later breathing detection based on improving data synchronization.

[0059] Step A500: performing time domain synchronization on the plurality of respiratory sound time domain information according to the plurality of time delay characteristic values, obtaining respiratory sound time domain synchronization information, and sending the information to the respiratory sound detection client.

[0060] In the present application, in order to improve the accuracy of the synchronization of respiratory detection data, it is necessary to synchronize several delay characteristic values ​​generated by delay analysis with several respiratory sound time domain information to determine the delay characteristic value. The delay characteristic value is an indicator used to describe the time difference or relative position between different respiratory sound signals. Furthermore, the transmission duration of each respiratory sound can be determined based on the delay characteristic value, and then the source of the delay characteristic value can be fitted. The detection data corresponding to the respiratory sound of the same time emitted by each source can be determined, and the corresponding detection data can be synchronously associated to obtain the respiratory sound time domain synchronization information, which is sent to the respiratory sound detection client to facilitate the subsequent analysis of the target user's respiratory sound and improve the accuracy of respiratory detection based on improving data synchronization in the later stage.

[0061] In summary, the breathing detection method for improving data synchronization provided by the embodiment of the present application includes at least the following technical effects, which realizes the time delay of fitting the breathing sounds detected at different positions, improves the data synchronization of the breathing sounds and the accuracy of breathing detection.

[0062] Embodiment 2

[0063] Based on the same inventive concept as the breathing detection method for improving data synchronization in the aforementioned embodiment, Figure 2 As shown, the present application provides a breathing detection system for improving data synchronization, the system comprising:

[0064] A model building module 1, wherein the model building module 1 is used to perform human body modeling on the detection target user to obtain a three-dimensional human body model;

[0065] A position acquisition module 2, wherein the position acquisition module 2 is used to interact with the respiratory sound detection server to obtain a plurality of respiratory sound time domain information of the respiratory sound source position and a plurality of respiratory sound detection positions;

[0066] A first identification module 3, wherein the first identification module 3 is used to identify the plurality of respiratory sound detection positions and the respiratory sound source positions in the three-dimensional human body model to generate a plurality of respiratory sound conduction paths;

[0067] A first analysis module 4, the first analysis module 4 is used to traverse the plurality of respiratory sound conduction paths to perform delay analysis and generate a plurality of delay characteristic values;

[0068] The first information acquisition module 5 is used to perform time domain synchronization on the plurality of respiratory sound time domain information according to the plurality of time delay characteristic values, obtain respiratory sound time domain synchronization information, and send the information to the respiratory sound detection client.

[0069] Furthermore, the system also includes:

[0070] A matrix construction module, wherein the matrix construction module is used to obtain the age characteristic value, gender characteristic value, height characteristic value, weight characteristic value, head-to-body ratio characteristic value, waist-to-leg ratio characteristic value, upper body thickness characteristic value and upper body width characteristic value of the detected target user, and construct an 8-dimensional constraint matrix;

[0071] A second analysis module, the second analysis module is used to perform a correlation analysis based on the 8-dimensional constraint matrix to determine an organ size identifier and a skeleton size identifier;

[0072] A second identification module, the second identification module is used to collect the external dimension identification of the detection target user;

[0073] The first activation module is used to activate the 3D modeling network, process the organ size identifier, the skeleton size identifier and the shape size identifier, and generate the three-dimensional model of the human body.

[0074] Furthermore, the system also includes:

[0075] A configuration module, the configuration module is used to configure a first organ size identification type set;

[0076] A matrix acquisition module, the matrix acquisition module is used to acquire a first organ size identification matrix with the 8-dimensional constraint matrix and the first organ size identification type set as constraints, wherein the row attribute of the first organ size identification matrix is ​​the size identification type, and the column attribute is the size identification record value;

[0077] A third identification module, the third identification module is used to perform a correlation analysis based on the first organ size identification matrix, generate a first organ size identification, and add it to the organ size identification, including:

[0078] A record value acquisition module, the record value acquisition module is used to obtain a k-th row size identification type and a k-th row size identification record value set according to the first organ size identification matrix, where k belongs to any row of the first organ size identification matrix;

[0079] a fourth identification module, the fourth identification module being used to perform a central tendency analysis on the k-th row size identification record value set, generate a size feature identification of the k-th row size identification type, and add the size feature identification to the first organ size identification;

[0080] The analysis process of the skeleton size identification and the organ size identification is the same.

[0081] Furthermore, the system also includes:

[0082] a first extraction module, the first extraction module being used to extract a first respiratory sound conduction path according to the plurality of respiratory sound conduction paths, wherein the first respiratory sound conduction path has a conduction medium attribute label sequence and a conduction distance label sequence;

[0083] A data mining module, the data mining module is used to perform frequency mining according to the conduction medium attribute label sequence and the conduction distance label sequence to obtain a frequent transmission duration label sequence;

[0084] A first adding module is used to generate a first delay characteristic value of the first respiratory sound conduction path according to the frequent transmission duration label sequence, and add it into the plurality of delay characteristic values.

[0085] Furthermore, the system also includes:

[0086] a label acquisition module, the label acquisition module being used to obtain a first conductive medium attribute label according to the conductive medium attribute label sequence, and to obtain a first conductive distance label according to the conductive distance label sequence, wherein the first conductive medium attribute label is solid or liquid or space;

[0087] A constraint module, the constraint module is used to collect a sound transmission time feature set with a preset human body characteristic temperature, the first conduction medium attribute label and the first conduction distance label as constraints;

[0088] The second adding module is used to perform outlier factor analysis on the sound transfer duration feature set, generate a first frequent transfer duration label, and add the label to the frequent transfer duration label sequence.

[0089] Furthermore, the system also includes:

[0090] A neighborhood construction module, the neighborhood construction module is used to traverse the sound transmission duration feature set and construct an outlier analysis neighborhood set;

[0091] Function building module, which is used to build outlier analysis functions:

[0092] ;

[0093] in, The outlier coefficient that characterizes the duration of the i-th sound transmission, Characterizes the characteristic value of the i-th sound transmission duration, The kth neighborhood feature value representing the neighborhood of the i-th sound transmission duration feature, Q represents the neighborhood size, and M represents the size of the sound transmission duration feature set;

[0094] A third analysis module, the third analysis module is used to perform outlier analysis on the sound transmission duration feature set according to the outlier analysis neighborhood set, the outlier coefficient threshold and the outlier analysis function, and generate an outlier transmission duration feature set;

[0095] A deletion module is used to delete the outlier transfer duration feature set from the sound transfer duration feature set, average the remaining transfer duration feature sets, and obtain the first frequent transfer duration label.

[0096] Furthermore, the system also includes:

[0097] A construction value module, wherein the construction value module is used to construct a neighborhood scale Q value, wherein M>Q≥5, and Q is an integer;

[0098] A third adding module is used to select a first outlier analysis neighborhood from near to far from the sound transmission duration feature set for the first sound transmission duration feature value of the sound transmission duration feature set based on the neighborhood scale Q value, and add the first outlier analysis neighborhood to the outlier analysis neighborhood set.

[0099] Through the above detailed description of a breathing detection method for improving data synchronization in this specification, those skilled in the art can clearly understand a breathing detection system for improving data synchronization in this embodiment. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0100] 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.

Claims

1. A breathing detection method for improving data synchronization, characterized in that: include: Perform human body modeling on the target user to obtain a three-dimensional human body model; The interactive breathing sound detection server obtains a number of breathing sound time domain information of the breathing sound source position and a number of breathing sound detection positions; Marking the plurality of respiratory sound detection positions and the respiratory sound source positions on the three-dimensional human body model to generate a plurality of respiratory sound conduction paths; Traversing the plurality of respiratory sound conduction paths to perform time delay analysis and generate a plurality of time delay characteristic values; Performing time domain synchronization on the plurality of respiratory sound time domain information according to the plurality of time delay characteristic values, obtaining respiratory sound time domain synchronization information, and sending the information to the respiratory sound detection client; Traversing the plurality of respiratory sound conduction paths to perform delay analysis, and generating a plurality of delay characteristic values, including: Extracting a first respiratory sound conduction path according to the plurality of respiratory sound conduction paths, wherein the first respiratory sound conduction path has a conduction medium attribute label sequence and a conduction distance label sequence; Perform frequency mining according to the conduction medium attribute label sequence and the conduction distance label sequence to obtain a frequent transmission duration label sequence; According to the frequent transmission duration label sequence, a first delay characteristic value of the first respiratory sound conduction path is generated and added to the plurality of delay characteristic values.

2. The method according to claim 1, characterized in that Perform human body modeling on the target user to obtain a 3D human body model, including: Obtaining the age characteristic value, gender characteristic value, height characteristic value, weight characteristic value, head-to-body ratio characteristic value, waist-to-leg ratio characteristic value, upper body thickness characteristic value and upper body width characteristic value of the detected target user, and constructing an 8-dimensional constraint matrix; Performing a correlation analysis based on the 8-dimensional constraint matrix to determine the organ size identifier and the skeleton size identifier; Collecting the external dimensions of the target user; The 3D modeling network is activated to process the organ size identifier, the skeleton size identifier and the outer shape size identifier to generate the human body three-dimensional model.

3. The method according to claim 2, characterized in that Performing a correlation analysis based on the 8-dimensional constraint matrix to determine the organ size identifier and the skeleton size identifier includes: Configuring a first organ size identification type set; Using the 8-dimensional constraint matrix and the first organ size identification type set as constraints, collecting a first organ size identification matrix, wherein the row attribute of the first organ size identification matrix is ​​the size identification type, and the column attribute is the size identification record value; Performing a correlation analysis based on the first organ size identification matrix to generate a first organ size identification, and adding the first organ size identification to the organ size identification, comprises: According to the first organ size identification matrix, obtaining a k-th row size identification type and a k-th row size identification record value set, where k belongs to any row of the first organ size identification matrix; Performing a central tendency analysis on the k-th row size identification record value set to generate a size feature identification of the k-th row size identification type, and adding the identification to the first organ size identification; The analysis process of the skeleton size identification and the organ size identification is the same.

4. The method according to claim 1, characterized in that Frequent mining is performed according to the conduction medium attribute label sequence and the conduction distance label sequence to obtain a frequent transmission duration label sequence, including: According to the conductive medium attribute tag sequence, a first conductive medium attribute tag is obtained, and according to the conductive distance tag sequence, a first conductive distance tag is obtained, wherein the first conductive medium attribute tag is solid, liquid or space; Collecting a sound transmission time feature set based on the preset human body characteristic temperature, the first conduction medium attribute label and the first conduction distance label as constraints; An outlier factor analysis is performed on the sound transfer duration feature set to generate a first frequent transfer duration label, which is added to the frequent transfer duration label sequence.

5. The method according to claim 4, characterized in that Performing outlier factor analysis on the sound transfer duration feature set to generate a first frequent transfer duration label, and adding the first frequent transfer duration label sequence to the first frequent transfer duration label sequence, including: Traversing the sound transmission duration feature set to construct an outlier analysis neighborhood set; Construct an outlier analysis function: ; in, The outlier coefficient that characterizes the duration of the i-th sound transmission, Characterizes the characteristic value of the i-th sound transmission duration, The kth neighborhood feature value representing the neighborhood of the i-th sound transmission duration feature, Q represents the neighborhood size, and M represents the size of the sound transmission duration feature set; Performing outlier analysis on the sound transmission duration feature set according to the outlier analysis neighborhood set, the outlier coefficient threshold and the outlier analysis function to generate an outlier transmission duration feature set; The outlier transfer duration feature set is deleted from the sound transfer duration feature set, and the remaining transfer duration feature sets are averaged to obtain the first frequent transfer duration label.

6. The method according to claim 5, characterized in that Traversing the sound transmission duration feature set, constructing an outlier analysis neighborhood set, including: Construct the neighborhood size Q value, where M>Q≥5 and Q is an integer; Based on the neighborhood size Q value, for the first sound transmission duration feature value of the sound transmission duration feature set, a first outlier analysis neighborhood is screened from near to far from the sound transmission duration feature set and added into the outlier analysis neighborhood set.

7. A breathing detection system for improving data synchronization, characterized in that: include: A model building module, wherein the model building module is used to perform human body modeling on the detection target user to obtain a three-dimensional human body model; A position acquisition module, wherein the position acquisition module is used to interact with the respiratory sound detection server to obtain a plurality of respiratory sound time domain information of the respiratory sound source position and a plurality of respiratory sound detection positions; A first identification module, the first identification module is used to identify the plurality of respiratory sound detection positions and the respiratory sound source positions in the three-dimensional human body model to generate a plurality of respiratory sound conduction paths; A first analysis module, the first analysis module is used to traverse the plurality of respiratory sound conduction paths to perform delay analysis and generate a plurality of delay characteristic values; A first information acquisition module, the first information acquisition module is used to perform time domain synchronization on the plurality of respiratory sound time domain information according to the plurality of time delay characteristic values, obtain respiratory sound time domain synchronization information, and send the information to the respiratory sound detection client; The system further comprises: a first extraction module, the first extraction module being used to extract a first respiratory sound conduction path according to the plurality of respiratory sound conduction paths, wherein the first respiratory sound conduction path has a conduction medium attribute label sequence and a conduction distance label sequence; A data mining module, the data mining module is used to perform frequency mining according to the conduction medium attribute label sequence and the conduction distance label sequence to obtain a frequent transmission duration label sequence; A first adding module is used to generate a first delay characteristic value of the first respiratory sound conduction path according to the frequent transmission duration label sequence, and add it into the plurality of delay characteristic values.

Images