Management system and method based on algorithm analysis technology

Through the smart bracelet, the physical sign data of the conversation object is collected and artificial intelligence analysis is carried out on the back end. The problem that traditional monitoring methods are susceptible to subjective factors is solved, real-time abnormal detection and early warning are realized, and the security of the conversation process is improved.

CN119033350AInactive Publication Date: 2024-11-29大连市纪检监察智慧云中心

Patent Information

Application Number
CN202411237124.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-05
Publication Date
2024-11-29
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The monitoring of physical abnormalities in traditional conversation subjects is susceptible to human-based subjective factors, making it difficult to detect hidden abnormalities, and increases the risk of conversation accidents.

Method used

Use smart bracelets to collect sign data of the conversation subjects in real time, such as heart rate and blood pressure values, and use artificial intelligence-based data processing and analysis algorithms to perform timing collaboration and correlation analysis on the backend, and detect it through the multimodal vital sign timing feature fusion module. Abnormal physical condition.

Benefits of technology

Real-time monitoring of the physical status of the conversation subject is achieved, abnormal warning is provided, and the safety and guarantee of the conversation process is improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119033350A_ABST
    Figure CN119033350A_ABST
Patent Text Reader

Abstract

The invention discloses a management system and method based on an algorithm analysis technology, and the method comprises the steps: monitoring and collecting the physical sign data, including the heart rate and the blood pressure value, of a conversation object in a conversation process in real time through a smart bracelet; and a data processing and analysis algorithm based on artificial intelligence is introduced at the rear end to carry out time sequence cooperation and correlation analysis on the sign data of the conversation objects, so that whether the physical state of the conversation objects is abnormal or not is detected, and abnormal conditions possibly occurring in the conversation process are recognized. Thus, the physical state of the conversation object can be monitored in real time, abnormal early warning can be provided, conversation accidents are avoided, and more safety and guarantee are brought to the conversation process.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of intelligent management, and more specifically, to a management system and method based on algorithm analysis technology. Background Art

[0002] With the advancement of science and technology, people are paying more and more attention to health and safety, especially in various social occasions. When having a conversation, the person being interviewed is usually only examined before the conversation begins, and during the conversation, it is impossible to understand the health status of the person being interviewed in real time. However, during the conversation, the individual's physical state may be affected by a variety of factors such as emotions, environment, and stress, resulting in abnormal fluctuations in physiological indicators. These abnormal fluctuations may not only affect the effectiveness of the conversation, but may also indicate potential health problems or emergencies, thereby increasing the risk of conversation accidents. Traditionally, abnormal monitoring of the person being interviewed's physical state is usually performed through manual identification, which is easily affected by human subjective factors and is difficult to detect hidden abnormalities in the person being interviewed's physical state.

[0003] Therefore, a management system based on algorithm analysis technology is desired, which can monitor the physical condition of the conversation partner in real time and provide abnormal warnings. Summary of the invention

[0004] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a management system and method based on algorithm analysis technology, which uses a smart bracelet to monitor and collect the vital sign data of the conversation partner during the conversation, including heart rate and blood pressure values, in real time, and introduces artificial intelligence-based data processing and analysis algorithms at the back end to perform time series coordination and correlation analysis of these conversation partner's vital sign data, so as to detect whether there is any abnormality in the physical state of the conversation partner, thereby identifying abnormal situations that may occur during the conversation. In this way, the physical state of the conversation partner can be monitored in real time and abnormal warnings can be provided to avoid the occurrence of conversation accidents, bringing more safety and security to the conversation process.

[0005] According to one aspect of the present application, a management system based on algorithm analysis technology is provided, which includes: The conversation partner's vital sign data collection module is used to collect the time series of the conversation partner's vital sign data during the conversation through a smart bracelet, wherein the vital sign data includes heart rate value and blood pressure value; A vital sign data transmission module, used to transmit the time series of the vital sign data to a conversation process monitoring server through the wireless communication module of the smart bracelet; A vital sign data time series pattern feature extraction module, used for performing time series autocorrelation coding and time series pattern feature extraction on the time series of the vital sign data in the conversation process monitoring server to obtain a heart rate time series correlation feature graph and a blood pressure time series correlation feature graph; A heart rate time series associated semantic representation module is used to input the heart rate time series associated feature map into an information excitation-compression module including a gated screening adaptive attention module and a feature multi-scale collapse module in the conversation process monitoring server to obtain a heart rate time series associated semantic representation vector; A multimodal vital sign time series feature fusion module is used to input the heart rate time series associated semantic representation vector and the blood pressure time series associated feature map into a semantic induced attention fusion network in the conversation process monitoring server to obtain a multimodal representation feature map of vital signs; The body state abnormality detection module is used to determine whether the body state of the conversation partner is abnormal based on the multimodal representation feature graph of the vital signs.

[0006] According to another aspect of the present application, a management method based on algorithm analysis technology is provided, which includes: Collecting a time series of vital sign data of the conversation partner during the conversation through a smart bracelet, wherein the vital sign data includes heart rate value and blood pressure value; Transmitting the time series of the vital sign data to a conversation process monitoring server through the wireless communication module of the smart bracelet; In the conversation process monitoring server, time series autocorrelation coding and time series pattern feature extraction are performed on the time series of the vital sign data to obtain a heart rate time series correlation feature graph and a blood pressure time series correlation feature graph; In the conversation process monitoring server, the heart rate time series associated feature map is input into an information excitation-compression module including a gated screening adaptive attention module and a feature multi-scale collapse module to obtain a heart rate time series associated semantic representation vector; In the conversation process monitoring server, the heart rate time series associated semantic representation vector and the blood pressure time series associated feature map are input into a semantically induced attention fusion network to obtain a vital sign multimodal representation feature map; Based on the multimodal representation feature graph of vital signs, determine whether the conversation partner's physical state is abnormal.

[0007] Compared with the prior art, the present application provides a management system and method based on algorithm analysis technology, which uses a smart bracelet to monitor and collect the vital sign data of the conversation partner in real time during the conversation, including heart rate and blood pressure values, and introduces artificial intelligence-based data processing and analysis algorithms at the back end to perform time series coordination and correlation analysis of these conversation partner's vital sign data, so as to detect whether there is any abnormality in the physical state of the conversation partner, thereby identifying abnormal situations that may occur during the conversation. In this way, the physical state of the conversation partner can be monitored in real time and abnormal warnings can be provided to avoid the occurrence of conversation accidents, bringing more safety and security to the conversation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0009] Figure 1 is a block diagram of a management system based on algorithm analysis technology according to an embodiment of the present application; Figure 2 A system architecture diagram of a management system based on algorithm analysis technology according to an embodiment of the present application; Figure 3 It is a block diagram of a vital sign data time series pattern feature extraction module in a management system based on algorithm analysis technology according to an embodiment of the present application; Figure 4 A block diagram of a heart rate timing association semantic representation module in a management system based on an algorithm analysis technology according to an embodiment of the present application; Figure 5 It is a flowchart of a management method based on algorithm analysis technology according to an embodiment of the present application. DETAILED DESCRIPTION

[0010] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0011] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0012] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0013] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0014] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0015] Traditionally, abnormal monitoring of the physical state of the conversation partner is usually performed through manual identification, which is easily affected by human subjective factors and is difficult to detect the hidden abnormal physical state of the conversation partner. Therefore, a management system based on algorithm analysis technology is desired, which can monitor the physical state of the conversation partner in real time and provide abnormal warnings.

[0016] In the technical solution of the present application, a management system based on algorithm analysis technology is proposed. Figure 1 It is a block diagram of a management system based on algorithm analysis technology according to an embodiment of the present application. Figure 2 : is a system architecture diagram of a management system based on algorithm analysis technology according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to the management system 300 based on algorithm analysis technology of the embodiment of the present application, it includes: a conversation object vital sign data collection module 310, which is used to collect the time series of the conversation object's vital sign data during the conversation process through a smart bracelet, wherein the vital sign data includes a heart rate value and a blood pressure value; a vital sign data transmission module 320, which is used to transmit the time series of the vital sign data to a conversation process monitoring server through the wireless communication module of the smart bracelet; a vital sign data time series pattern feature extraction module 330, which is used to perform time series autocorrelation coding and time series pattern feature extraction on the time series of the vital sign data on the conversation process monitoring server to obtain a heart rate time series correlation feature map and a blood pressure time series correlation feature map. Figure; a heart rate timing association semantic representation module 340, which is used to input the heart rate timing association feature map into an information excitation-compression module including a gated screening adaptive attention module and a feature multi-scale collapse module in the conversation process monitoring server to obtain a heart rate timing association semantic representation vector; a multimodal vital sign timing feature fusion module 350, which is used to input the heart rate timing association semantic representation vector and the blood pressure timing association feature map into a semantic induced attention fusion network in the conversation process monitoring server to obtain a vital sign multimodal representation feature map; a body state abnormality detection module 360, which is used to determine whether the body state of the conversation partner is abnormal based on the vital sign multimodal representation feature map.

[0017] In particular, the conversation partner's vital sign data collection module 310 and the vital sign data transmission module 320 are used to collect the time series of the conversation partner's vital sign data during the conversation through a smart bracelet, wherein the vital sign data includes heart rate value and blood pressure value; then, the time series of the vital sign data is transmitted to the conversation process monitoring server through the wireless communication module of the smart bracelet. It should be understood that by monitoring the heart rate value and blood pressure value of the conversation partner during the conversation, potential physical abnormalities or health problems can be discovered in time. These data can serve as important reference indicators of health status, which help to discover and deal with potential health risks in advance. Further, transmitting the time series of vital sign data to the monitoring server can realize real-time monitoring of the conversation partner's physical state. The monitoring server can receive and process these data in time to detect any abnormal situation in time.

[0018] In particular, the vital sign data time series pattern feature extraction module 330 is used to perform time series autocorrelation coding and time series pattern feature extraction on the time series of the vital sign data in the conversation process monitoring server to obtain a heart rate time series correlation feature graph and a blood pressure time series correlation feature graph. In particular, in a specific example of the present application, Figure 3As shown, the vital sign data time series pattern feature extraction module 330 includes: a vital sign time series data regularization unit 331, which is used to regularize the time series of the vital sign data according to the parameter sample dimension in the conversation process monitoring server to obtain the time series of heart rate values ​​and the time series of blood pressure values; a vital sign data time series autocorrelation information capture unit 332, which is used to calculate the spatial covariance matrix of the time series of the heart rate values ​​and the time series of the blood pressure values ​​in the conversation process monitoring server respectively to obtain the heart rate time series correlation matrix and the blood pressure time series correlation matrix; a vital sign time series pattern feature extraction unit 333, which is used to input the heart rate time series correlation matrix and the blood pressure time series correlation matrix into the time series pattern feature extractor based on the void convolutional neural network model in the conversation process monitoring server to obtain the heart rate time series correlation feature graph and the blood pressure time series correlation feature graph.

[0019] Specifically, the vital sign time series data regularization unit 331 is used to regularize the time series of the vital sign data according to the parameter sample dimension in the conversation process monitoring server to obtain the time series of the heart rate value and the time series of the blood pressure value. Considering that during the conversation, the vital sign data of the conversation partner, such as the heart rate value and the blood pressure value, will continue to change with the passage of time, that is, the heart rate and blood pressure have a dynamic change law of time series in the time dimension. In order to be able to more intuitively present the change information of the heart rate and blood pressure, and provide a data basis for the subsequent extraction of the heart rate time series features and the blood pressure time series features and the analysis of the correlation between the vital sign time series, in the technical solution of the present application, it is necessary to regularize the time series of the vital sign data according to the parameter sample dimension in the conversation process monitoring server to obtain the time series of the heart rate value and the time series of the blood pressure value.

[0020] Specifically, the time series autocorrelation information capture unit 332 of the vital sign data is used to calculate the spatial covariance matrix of the time series of the heart rate value and the time series of the blood pressure value in the conversation process monitoring server to obtain the heart rate time series correlation matrix and the blood pressure time series correlation matrix. Considering that the time series change pattern of heart rate and blood pressure can reflect the health status of an individual, in order to better analyze the time series change characteristics and patterns of heart rate and blood pressure in the time dimension, in the technical solution of the present application, the time series of the heart rate value and the time series of the blood pressure value are further calculated in the conversation process monitoring server to obtain the heart rate time series correlation matrix and the blood pressure time series correlation matrix. By calculating the spatial covariance matrix of the time series of the heart rate value and the time series of the blood pressure value, the autocorrelation relationship of the heart rate time series information of the conversation object and the autocorrelation relationship of the blood pressure time series information can be established respectively. In this way, it is conducive to analyzing the implicit correlation characteristics and patterns between the heart rates at different time points, which helps to understand the changing trend, periodicity and possible abnormal conditions of the heart rate. Similarly, it is also helpful to analyze the degree of correlation between blood pressure values ​​at different time points, which helps to discover the changing trend, periodicity and possible abnormal conditions of blood pressure. In a specific example of the present application, in the conversation process monitoring server, the spatial covariance matrix of the time series of the heart rate value and the time series of the blood pressure value are calculated respectively to obtain the heart rate time series correlation matrix and the blood pressure time series correlation matrix, including: arranging the time series of the heart rate value according to the time dimension to obtain the heart rate time series input vector; calculating the product between the heart rate time series input vector and the transposed vector of the heart rate time series input vector to obtain the heart rate time series autocorrelation input matrix; calculating the sum of the elements on the diagonal of the heart rate time series autocorrelation input matrix to obtain the heart rate time series autocorrelation value; dividing the eigenvalues ​​of each position of the heart rate time series autocorrelation input matrix by the heart rate time series autocorrelation value to obtain the heart rate time series correlation matrix composed of multiple heart rate time series autocorrelation trace coefficients.

[0021] The spatial covariance matrix of the heart rate time series input vector is calculated using the following heart rate time series spatial covariance formula to obtain a heart rate time series association matrix; wherein the heart rate time series spatial covariance formula is:

[0022] in, represents the heart rate timing input vector, represents the transposed vector of the heart rate timing input vector, It means to calculate the sum of each element on the diagonal of the matrix. The heart rate time series association matrix and the blood pressure time series association matrix are obtained; and the spatial covariance matrix of the blood pressure time series input vector is calculated using the following blood pressure time series spatial covariance formula to obtain the blood pressure time series association matrix; wherein the blood pressure time series spatial covariance formula is:

[0023] in, represents the blood pressure time series input vector, represents the transposed vector of the blood pressure time series input vector, It means to calculate the sum of each element on the diagonal of the matrix. is the blood pressure time series association matrix. Specifically, the vital sign time series pattern feature extraction unit 333 is used to input the heart rate time series association matrix and the blood pressure time series association matrix into the time series pattern feature extractor based on the void convolutional neural network model in the conversation process monitoring server to obtain the heart rate time series association feature graph and the blood pressure time series association feature graph. That is, in order to extract important vital sign time series features and change patterns from the heart rate time series association matrix and the blood pressure time series association matrix, and further help analyze and identify abnormal conditions of the conversation partner's physical state, in the conversation process monitoring server, the heart rate time series association matrix and the blood pressure time series association matrix are input into the time series pattern feature extractor based on the void convolutional neural network model to obtain the heart rate time series association feature graph and the blood pressure time series association feature graph. Through the processing of the temporal pattern feature extractor based on the void convolutional neural network model, the temporal autocorrelation feature information and long-distance temporal dependency associations of the heart rate and blood pressure values ​​of the conversation partner in the time dimension can be captured respectively, which can help the system identify important temporal patterns in the heart rate and blood pressure data, such as periodic changes, mutation points, etc., which helps to discover potential abnormal patterns or trends and improve the accuracy of abnormal body condition detection.

[0024] It is worth noting that the Dilated Convolutional Neural Network is a convolutional neural network structure that expands the receptive field by introducing holes (or dilation rates) in the convolution kernel. Dilated convolution can increase the receptive field of the network without increasing the number of parameters, thereby better capturing the long-distance dependencies in the input data. In traditional convolution operations, each element of the convolution kernel is convolved with the adjacent elements of the input data. In dilated convolution, there is a certain interval (hole) between the elements in the convolution kernel, which allows the convolution kernel to skip some positions in the receptive field for convolution operations, thereby expanding the receptive field. A key parameter of dilated convolution is the dilation rate, which defines the interval between elements in the convolution kernel. By adjusting the dilation rate, the size of the receptive field of dilated convolution can be controlled. A larger dilation rate allows the convolution kernel to skip more positions, thereby capturing a wider range of information. Dilated convolutional neural networks have been widely used in image processing, speech recognition, natural language processing and other fields, especially when processing data with long-distance dependencies. Atrous convolution can help the network better capture global information and improve the network's perception and generalization capabilities.

[0025] It is worth mentioning that in other specific examples of the present application, the time series of the vital sign data can be subjected to temporal autocorrelation coding and temporal pattern feature extraction in the conversation process monitoring server in other ways to obtain a heart rate temporal correlation feature graph and a blood pressure temporal correlation feature graph, for example: inputting the time series of the vital sign data; performing temporal autocorrelation coding on the time series data of the heart rate and blood pressure to capture the temporal correlation in the data; performing temporal pattern feature extraction on the data after temporal autocorrelation coding to capture important features and patterns in the data; combining the heart rate features and blood pressure features obtained from the temporal pattern feature extraction into a heart rate temporal correlation feature graph and a blood pressure temporal correlation feature graph, respectively, to obtain the heart rate temporal correlation feature graph and the blood pressure temporal correlation feature graph. In particular, the heart rate temporal correlation semantic representation module 340 is used in the conversation process monitoring server to input the heart rate temporal correlation feature graph into an information excitation-compression module including a gated screening adaptive attention module and a feature multi-scale collapse module to obtain a heart rate temporal correlation semantic representation vector. Considering that the heart rate timing association feature graph contains correlation semantics and autocorrelation features related to the heart rate timing, this heart rate timing association semantics contains not only semantic information related to the task of monitoring the abnormal physical state of the conversation partner, but also some feature information with low correlation with the task. Therefore, in order to further extract and highlight important heart rate timing semantic features, and capture timing features of different scales for the important heart rate timing semantics to further improve the monitoring capability and accuracy of abnormal physical states of the conversation partners, in the conversation process monitoring server, the heart rate timing association feature graph is further input into an information excitation-compression module including a gated screening adaptive attention module and a feature multi-scale collapse module to obtain a heart rate timing association semantic representation vector. In particular, in a specific example of the present application, such as Figure 4As shown, the heart rate timing association semantic representation module 340 includes: a heart rate timing semantic feature gated screening and enhancement unit 341, which is used to process the heart rate timing association feature map through the gated screening adaptive attention module of the information excitation-compression module to obtain a heart rate timing semantic association enhanced feature map; a heart rate timing multi-scale association semantic capture unit 342, which is used to process the heart rate timing semantic association enhanced feature map through the feature multi-scale collapse module of the information excitation-compression module to obtain the heart rate timing association semantic representation vector. Specifically, the heart rate timing semantic feature gated screening and enhancement unit 341 is used to process the heart rate timing association feature map through the gated screening adaptive attention module of the information excitation-compression module to obtain a heart rate timing semantic association enhanced feature map. It should be understood that the processing by the gated screening adaptive attention module can focus on the most important part of the heart rate timing association feature map, which is related to the timing pattern and abnormal fluctuations of the heart rate, and can improve the model's ability to recognize key information of the heart rate timing, thereby assisting in more accurate detection of abnormal physical conditions of the conversation partner. In a specific example of the present application, the heart rate timing association feature map is processed in the gated screening adaptive attention module of the information excitation-compression module to obtain a heart rate timing semantic association enhanced feature map, including: performing global pooling processing based on the maximum value, global pooling processing based on random values, and global pooling processing based on average values ​​on the heart rate timing association feature map to obtain a global maximum pooling feature vector of the heart rate timing semantic association feature, a global mean pooling feature vector of the heart rate timing semantic association feature, and a global random value pooling feature vector of the heart rate timing semantic association feature; calculating the The method comprises the following steps: performing a position-weighted sum of the global maximum pooling feature vector of the heart rate time series semantic association feature, the global mean pooling feature vector of the heart rate time series semantic association feature, and the global random value pooling feature vector of the heart rate time series semantic association feature to obtain a multi-scale pooling representation vector of the heart rate time series semantic association feature; calculating the matrix multiplication of the multi-scale pooling representation vector of the heart rate time series semantic association feature and the weight matrix, and then performing vector addition with the bias vector to obtain a bias-adjusted multi-scale pooling representation vector of the heart rate time series semantic association feature; and adding the bias-adjusted multi-scale pooling representation vector of the heart rate time series semantic association feature to the multi-scale pooling representation vector of the heart rate time series semantic association feature. An activation process is performed in the activation function to obtain a heart rate timing channel semantic association feature vector; based on a gating mechanism, each position feature value in the heart rate timing channel semantic association feature vector is masked to obtain a masked heart rate timing semantic enhancement representation vector, wherein, in response to each position feature value in the heart rate timing channel semantic association feature vector being greater than or equal to a predetermined threshold, each position feature value in the heart rate timing channel semantic association feature vector takes an original value, otherwise it is set to zero; each position feature value in the masked heart rate timing semantic enhancement representation vector is normalized to obtain a normalized heart rate timing semantic weight vector; using the feature value of each position in the normalized heart rate timing semantic weight vector as a weighting coefficient, each feature matrix along the channel dimension in the heart rate timing association feature map is weighted optimized to obtain the heart rate timing semantic association enhanced feature map. In summary, in the above embodiment, the heart rate timing association feature map is processed in the gated screening adaptive attention module of the information excitation-compression module to obtain the heart rate timing semantic association enhanced feature map, including: the heart rate timing association feature map is processed in the gated screening adaptive attention module of the information excitation-compression module with the following gated screening adaptive enhancement formula to obtain the heart rate timing semantic association enhanced feature map; wherein the gated screening adaptive enhancement formula is:

[0026] in, is the heart rate time series correlation characteristic diagram, , and They are global mean pooling, maximum value pooling and random value pooling. , and are preset trainable weight values, is the multi-scale pooling representation vector of the semantic association feature of heart rate time series, and are the weight matrix and bias vector of the convolutional layer, respectively. for Activation function, is the semantic enhancement representation vector of heart rate time series, is the masked heart rate time series semantic enhancement representation vector, is the first The eigenvalues ​​at the positions, is the normalized heart rate time series semantic weight vector, It means point multiplication by position. It is the heart rate time series semantic association enhanced feature map. Specifically, the heart rate time series multi-scale association semantic capture unit 342 is used to process the heart rate time series semantic association enhanced feature map through the feature multi-scale collapse module of the information excitation-compression module to obtain the heart rate time series association semantic representation vector. It should be understood that the feature multi-scale collapse module can analyze the heart rate time series data on local and global time scales, which helps to capture the short-term and long-term patterns of heart rate changes. Multi-scale analysis of the heart rate time series can provide a more comprehensive perspective to help identify heart rate anomalies on different time scales, such as short-term fluctuations and long-term change trends of the heart rate. Therefore, the important feature excitation of the heart rate time series association feature map and the compression fusion of multi-scale features through the information excitation-compression module will help to understand the overall health status of the conversation partner more comprehensively later. In a specific example of the present application, first, the heart rate timing semantic association enhancement feature map is upsampled to obtain an upsampled heart rate timing semantic association enhancement feature map; then, each feature matrix along the channel dimension in the upsampled heart rate timing semantic association enhancement feature map is globally mean pooled to obtain an upsampled heart rate timing semantic channel feature vector; the upsampled heart rate timing semantic channel feature vector is point convolution encoded to obtain a heart rate timing global semantic feature vector; then, the upsampled heart rate timing semantic association enhancement feature map is point convolution encoded to obtain a channel-modulated upsampled heart rate timing semantic association enhancement feature map; the channel-modulated upsampled heart rate timing semantic association enhancement feature map is two-dimensionally convolution encoded to obtain a heart rate timing local semantic feature vector; and then, the heart rate timing global semantic feature vector and the heart rate timing local semantic feature vector are fused to obtain the heart rate timing association semantic representation vector. It is worth mentioning that in other specific examples of the present application, the heart rate timing-related feature map can also be input into the information excitation-compression module including the gated screening adaptive attention module and the feature multi-scale collapse module in the conversation process monitoring server in other ways to obtain the heart rate timing-related semantic representation vector, for example: input the heart rate timing-related feature map; input the heart rate timing-related feature map into the gated screening adaptive attention module; determine the importance of each time step by calculating the attention weight; weight the feature map according to the calculated attention weight to emphasize or weaken the feature representation of different time steps; extract multi-scale feature representation through convolution or pooling operations of different scales to capture information at different levels; fuse features of different scales to integrate multi-scale information; convert the multi-scale feature representation into a more compact form through compression operation to reduce the dimension while retaining key information; integrate the features processed by the gated screening adaptive attention module and the feature multi-scale collapse module; apply the excitation function to activate the feature representation and introduce nonlinearity to obtain the heart rate timing-related semantic representation vector.In particular, the multimodal vital sign time series feature fusion module 350 is used to input the heart rate time series associated semantic representation vector and the blood pressure time series associated feature map into the semantic induced attention fusion network in the conversation process monitoring server to obtain the multimodal representation feature map of vital signs. It should be understood that heart rate and blood pressure are two different types of physiological signals, which reflect different physiological states of the human body. In addition, since there is a hidden correlation relationship and mutual influence between the heart rate time series associated semantic features and the blood pressure time series features, for example, under normal circumstances, the changes in heart rate and blood pressure should be coordinated. If the relationship between them is abnormal, it may mean that there is a potential health problem. Therefore, the time series fusion semantics of the two can more accurately reflect the changes in the physical state of the conversation object, and is conducive to detecting the atypical relationship between heart rate and blood pressure and the abnormality of the physical state. Based on this, in order to more comprehensively understand the health status of the conversation object so as to detect abnormal conditions, the heart rate time series associated semantic representation vector and the blood pressure time series associated feature map are further input into the semantic induced attention fusion network to obtain the multimodal representation feature map of vital signs in the conversation process monitoring server. Through the processing of the semantically induced attention fusion network, the semantic features of the conversation partner's heart rate timing association can be used to optimize the expression ability of the blood pressure timing features, so that the timing feature information of these two physiological parameters can be comprehensively utilized to more comprehensively characterize the vital signs of the conversation partner. In other words, the semantically induced attention fusion network can learn and capture the hidden association between the conversation partner's heart rate timing association semantics and blood pressure timing features, so as to dynamically adjust the weights using the important semantic features of the heart rate's timing autocorrelation to assist in the feature expression of the blood pressure timing association feature graph. In this way, the model can better adapt to different conversation scenarios and individual differences, improve the generalization ability of the model, and provide a basis for monitoring abnormal physical conditions of the conversation partner, thereby improving the accuracy and reliability of abnormality detection.

[0027] In a specific example of the present application, in the conversation process monitoring server, the heart rate time series associated semantic representation vector and the blood pressure time series associated feature map are input into a semantically induced attention fusion network to obtain a multimodal representation feature map of vital signs, including: after matrix multiplication of the heart rate time series associated semantic representation vector and a predetermined weight coefficient matrix, the obtained feature vector is positionally added to a predetermined offset vector to obtain a modulated heart rate time series associated semantic representation vector; using The activation function activates the modulated heart rate timing association semantic representation vector to obtain the heart rate timing semantic fully connected feature vector; and the blood pressure timing association feature map is weighted along the channel dimension using the respective position feature values ​​in the heart rate timing semantic fully connected feature vector to obtain the vital sign multimodal representation feature map.

[0028] In summary, in the above embodiment, in the conversation process monitoring server, the heart rate time series associated semantic representation vector and the blood pressure time series associated feature map are input into the semantic induced attention fusion network to obtain the multimodal representation feature map of vital signs, including: inputting the heart rate time series associated semantic representation vector and the blood pressure time series associated feature map into the semantic induced attention fusion network to process them with the following semantic induced fusion formula to obtain the multimodal representation feature map of vital signs; wherein the semantic induced fusion formula is:

[0029] in, is the heart rate time series associated semantic representation vector, represents the weight coefficient matrix, is the offset vector, for Activation function, is the semantic fully connected feature vector of heart rate time series, Indicates that the feature map is weighted using the weight vector. is the blood pressure time series correlation characteristic diagram, It is the multimodal representation feature graph of the vital signs. In particular, the body state abnormality detection module 360 ​​is used to determine whether there is an abnormality in the body state of the conversation partner based on the multimodal representation feature graph of the vital signs. In the technical solution of the present application, in the conversation process monitoring server, the multimodal representation feature graph of the vital signs is passed through a classifier-based monitor to obtain a monitoring result, and the monitoring result is used to indicate whether there is an abnormality in the body state of the conversation partner. In other words, the multimodal temporal correlation feature representation information of the body signs of the conversation partner is used for classification processing to detect whether there is an abnormality in the body state of the conversation partner, thereby identifying abnormal situations that may occur during the conversation. In this way, the body state of the conversation partner can be monitored in real time and abnormal warnings can be provided to avoid the occurrence of conversation accidents and bring more safety and security to the conversation process.

[0030] Here, it is considered that the blood pressure time series association feature map and the heart rate time series association feature map represent the local time domain semantic association feature of blood pressure and the local time domain semantic association feature of heart rate respectively, and each local feature of the blood pressure time series association feature map and the heart rate time series association feature map is distributed along the channel dimension of the time series pattern feature extractor of the hollow convolutional neural network model. When the heart rate time series association feature map is input into the information excitation-compression module including the gated screening adaptive attention module and the feature multi-scale collapse module, the information excitation-compression module including the gated screening adaptive attention module and the feature multi-scale collapse module uses the attention mechanism to strengthen the significance of each feature matrix of the heart rate time series association feature map and uses the feature multi-scale collapse mechanism to retain the effective feature components so that the obtained heart rate time series association semantic representation vector has stronger feature representation ability. When the heart rate time series association semantic representation vector and the blood pressure time series association feature map are input into the semantic induced attention fusion network, the heart rate time series semantic association information is essentially used to guide the multimodal feature fusion to obtain the multimodal representation feature map of vital signs. However, considering that when the heart rate time-series association feature map is subjected to information excitation and compression, feature interactions occur among the feature matrices of the heart rate time-series association feature map, which causes a shift in the feature correspondence between the heart rate time-series association semantic representation vector and the blood pressure time-series association feature map in the channel dimension, thereby causing an overflow of the monitoring results obtained by inputting the multimodal representation feature map of vital signs obtained by semantically induced attention fusion into the classifier-based monitor, affecting its accuracy.

[0031] In a preferred example, based on this, when the multimodal representation feature graph of vital signs is input into a classifier-based monitor to obtain monitoring results, the multimodal representation feature graph of vital signs is optimized, and the optimization process includes the steps of: Determine an abnormal probability value corresponding to the presence of an abnormality in the body state of the conversation partner obtained by inputting the multimodal representation feature graph of the vital signs based on the monitor of the classifier; Passing the vital sign multimodal representation feature map through a probabilistic activation function to obtain a probabilistic vital sign multimodal representation feature map; Subtracting the probabilistic multimodal representation feature graph of vital signs from the unit feature graph, and using the feature graph resulting from the subtraction as a base to calculate a power function with the abnormal probability value as an exponent; Performing a point multiplication of the power function feature map and the probabilistic vital sign multimodal representation feature map to obtain a vital sign multimodal representation intermediate feature map; Taking the absolute value of each eigenvalue of the probabilistic multimodal representation feature graph of vital signs as a base, calculating a power function with the reciprocal of the abnormal probability value as an exponent, and summing all eigenvalues ​​to obtain a bias eigenvalue of the multimodal representation of vital signs; After multiplying the bias eigenvalue of the multimodal representation of vital signs by a weight serving as a hyperparameter, the bias eigenvalue is interpolated with the intermediate feature graph of the multimodal representation of vital signs to obtain an optimized multimodal representation feature graph of vital signs.

[0032] Wherein, the probabilistic multimodal representation feature graph of vital signs is recorded as , then the optimized multimodal representation feature map of vital signs It is expressed as:

[0033] in, is the probabilistic multimodal representation feature graph of vital signs, is a unit eigenmap with all eigenvalues ​​being one, and , is the probabilistic multimodal representation feature map of vital signs The eigenvalues ​​at each position of It is the point multiplication by position. It is subtracted by position point. is the abnormal probability value, To calculate the power function of the feature value of each position in the feature map with the abnormal probability value as an exponent, It is added by location point. is the weight as a hyperparameter, The optimized multimodal representation feature graph of vital signs. Finally, the optimized multimodal representation feature graph of vital signs is input into a monitor based on a classifier to obtain a monitoring result. That is, for the multimodal representation feature graph of vital signs, based on the convergence limit of the probability distribution of the Cauchy-Hadamard form power series, the feature set of the multimodal representation feature graph of vital signs is represented by a power series based on the convergence probability, and a set convergence bounded bias based on each eigenvalue of the multimodal representation feature graph of vital signs is added as a convergence radius representation, so as to avoid the invalid overflow of the classification result caused by the deviation of the distribution mode of the multimodal representation feature graph of vital signs in the regression process of the monitor based on the classifier due to the gradient propagation of the model parameters based on the convergence probability, thereby improving the accuracy of the monitoring result obtained by the monitor based on the classifier. In this way, it is possible to more accurately detect whether there is an abnormality in the physical state of the conversation object, thereby identifying abnormal situations that may occur during the conversation, and bringing more safety and security to the conversation process.

[0034] As described above, the management system 300 based on the algorithm analysis technology according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with a management algorithm based on the algorithm analysis technology. In a possible implementation, the management system 300 based on the algorithm analysis technology according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the management system 300 based on the algorithm analysis technology can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the management system 300 based on the algorithm analysis technology can also be one of the many hardware modules of the wireless terminal.

[0035] Alternatively, in another example, the management system 300 based on algorithm analysis technology and the wireless terminal may also be separate devices, and the management system 300 based on algorithm analysis technology may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0036] Furthermore, a management method based on algorithm analysis technology is also provided.

[0037] Figure 5 FIG. 1 is a flow chart of a management method based on algorithm analysis technology according to an embodiment of the present application. Figure 5 As shown, according to the management method based on algorithm analysis technology of the embodiment of the present application, the steps include: S1, collecting the time series of vital sign data of the conversation partner during the conversation process through a smart bracelet, wherein the vital sign data includes heart rate value and blood pressure value; S2, transmitting the time series of the vital sign data to the conversation process monitoring server through the wireless communication module of the smart bracelet; S3, at the conversation process monitoring server, performing temporal autocorrelation coding and temporal pattern feature extraction on the time series of the vital sign data to obtain a heart rate time series correlation feature map and a blood pressure time series correlation feature map; S4, at the conversation process monitoring server, inputting the heart rate time series correlation feature map into an information excitation-compression module including a gated screening adaptive attention module and a feature multi-scale collapse module to obtain a heart rate time series correlation semantic representation vector; S5, at the conversation process monitoring server, inputting the heart rate time series correlation semantic representation vector and the blood pressure time series correlation feature map into a semantic induced attention fusion network to obtain a multimodal representation feature map of vital signs; S6, based on the multimodal representation feature map of vital signs, determining whether the physical state of the conversation partner is abnormal.

[0038] In summary, the management method based on algorithm analysis technology according to the embodiment of the present application is explained, which uses a smart bracelet to monitor and collect the vital sign data of the conversation partner during the conversation, including heart rate and blood pressure values, in real time, and introduces an artificial intelligence-based data processing and analysis algorithm at the back end to perform time series coordination and correlation analysis of these conversation partner's vital sign data, so as to detect whether there is any abnormality in the physical state of the conversation partner, thereby identifying abnormal situations that may occur during the conversation. In this way, the physical state of the conversation partner can be monitored in real time and abnormal warnings can be provided to avoid the occurrence of conversation accidents, bringing more safety and security to the conversation process.

[0039] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A management system based on algorithm analysis technology, characterized in that: include: The conversation partner's vital sign data collection module is used to collect the time series of the conversation partner's vital sign data during the conversation through a smart bracelet, wherein the vital sign data includes heart rate value and blood pressure value; A vital sign data transmission module, used to transmit the time series of the vital sign data to a conversation process monitoring server through the wireless communication module of the smart bracelet; A vital sign data time series pattern feature extraction module, used for performing time series autocorrelation coding and time series pattern feature extraction on the time series of the vital sign data in the conversation process monitoring server to obtain a heart rate time series correlation feature graph and a blood pressure time series correlation feature graph; A heart rate time series associated semantic representation module is used to input the heart rate time series associated feature map into an information excitation-compression module including a gated screening adaptive attention module and a feature multi-scale collapse module in the conversation process monitoring server to obtain a heart rate time series associated semantic representation vector; A multimodal vital sign time series feature fusion module is used to input the heart rate time series associated semantic representation vector and the blood pressure time series associated feature map into a semantic induced attention fusion network in the conversation process monitoring server to obtain a multimodal representation feature map of vital signs; The body state abnormality detection module is used to determine whether the body state of the conversation partner is abnormal based on the multimodal representation feature graph of the vital signs.

2. The management system based on algorithm analysis technology according to claim 1 is characterized in that: The vital sign data temporal pattern feature extraction module comprises: A vital sign time series data regularization unit, used for regularizing the time series of the vital sign data according to the parameter sample dimension in the conversation process monitoring server to obtain a time series of heart rate values ​​and a time series of blood pressure values; A vital sign data time series autocorrelation information capturing unit is used to calculate, on the conversation process monitoring server, the spatial covariance matrices of the time series of the heart rate value and the time series of the blood pressure value to obtain a heart rate time series correlation matrix and a blood pressure time series correlation matrix; The vital sign timing pattern feature extraction unit is used to input the heart rate timing association matrix and the blood pressure timing association matrix into the timing pattern feature extractor based on the void convolutional neural network model in the conversation process monitoring server to obtain the heart rate timing association feature graph and the blood pressure timing association feature graph.

3. The management system based on algorithm analysis technology according to claim 2 is characterized in that: The vital sign data temporal autocorrelation information capturing unit is further used to: Arranging the time series of the heart rate values ​​according to the time dimension to obtain a heart rate time series input vector; Calculating the product between the heart rate time series input vector and the transposed vector of the heart rate time series input vector to obtain a heart rate time series autocorrelation input matrix; Calculating the sum of the elements on the diagonal of the heart rate time series autocorrelation input matrix to obtain the heart rate time series autocorrelation value; The eigenvalues ​​of each position of the heart rate time series autocorrelation input matrix are divided by the heart rate time series autocorrelation value to obtain the heart rate time series association matrix composed of a plurality of heart rate time series autocorrelation trace coefficients.

4. The management system based on algorithm analysis technology according to claim 3 is characterized in that: The heart rate time series associated semantic representation module includes: A heart rate time series semantic feature gated screening and enhancement unit, used for processing the heart rate time series associated feature map through the gated screening adaptive attention module of the information excitation-compression module to obtain a heart rate time series semantic associated enhanced feature map; The heart rate time series multi-scale association semantics capture unit is used to process the heart rate time series semantic association enhanced feature map through the feature multi-scale collapse module of the information excitation-compression module to obtain the heart rate time series association semantics representation vector.

5. The management system based on algorithm analysis technology according to claim 4 is characterized in that: The heart rate time series semantic feature gated screening enhancement unit is further used to: Performing maximum-based global pooling processing, random-value-based global pooling processing, and average-based global pooling processing on the heart rate timing association feature graph to obtain a global maximum value pooling feature vector of a heart rate timing semantic association feature, a global average value pooling feature vector of a heart rate timing semantic association feature, and a global random value pooling feature vector of a heart rate timing semantic association feature; Calculate the position-weighted sum of the global maximum pooling feature vector of the heart rate time series semantic association feature, the global mean pooling feature vector of the heart rate time series semantic association feature, and the global random value pooling feature vector of the heart rate time series semantic association feature to obtain a multi-scale pooling representation vector of the heart rate time series semantic association feature; Calculate the matrix multiplication of the multi-scale pooling representation vector of the heart rate time series semantic association feature and the weight matrix, and then add the vector to the bias vector to obtain the bias-adjusted multi-scale pooling representation vector of the heart rate time series semantic association feature; The bias-adjusted heart rate temporal semantic association feature multi-scale pooling representation vector is passed through Activation processing is performed in the activation function to obtain the semantic association feature vector of the heart rate time series channel; Based on the gating mechanism, masking is performed on each position feature value in the semantic association feature vector of the heart rate timing channel to obtain a masked heart rate timing semantic enhancement representation vector, wherein, in response to each position feature value in the semantic association feature vector of the heart rate timing channel being greater than or equal to a predetermined threshold, each position feature value in the semantic association feature vector of the heart rate timing channel is taken to an original value, otherwise it is set to zero; Normalizing each position feature value in the masked heart rate time series semantic enhancement representation vector to obtain a normalized heart rate time series semantic weight vector; Taking the eigenvalues ​​of each position in the normalized heart rate time series semantic weight vector as weighting coefficients, weighted optimization is performed on each feature matrix along the channel dimension in the heart rate time series association feature map to obtain the heart rate time series semantic association enhanced feature map.

6. The management system based on algorithm analysis technology according to claim 5 is characterized in that: The heart rate time series multi-scale association semantic capture unit is further used to: Upsampling the heart rate time series semantic association enhanced feature map to obtain an upsampled heart rate time series semantic association enhanced feature map; Performing global mean pooling processing on each feature matrix along the channel dimension in the upsampled heart rate time series semantic association enhanced feature map to obtain an upsampled heart rate time series semantic channel feature vector; Performing point convolution encoding on the upsampled heart rate time series semantic channel feature vector to obtain a heart rate time series global semantic feature vector; Performing point convolution encoding on the upsampled heart rate temporal semantic association enhanced feature map to obtain a channel modulated upsampled heart rate temporal semantic association enhanced feature map; Performing two-dimensional convolution coding on the channel-modulated upsampled heart rate time series semantic association enhanced feature map to obtain a heart rate time series local semantic feature vector; The heart rate time series global semantic feature vector and the heart rate time series local semantic feature vector are fused to obtain the heart rate time series associated semantic representation vector.

7. The management system based on algorithm analysis technology according to claim 6 is characterized in that: The multimodal vital sign time series feature fusion module is further used to: After matrix multiplication of the heart rate time series associated semantic representation vector and a predetermined weight coefficient matrix, the obtained feature vector and a predetermined offset vector are added according to position to obtain a modulated heart rate time series associated semantic representation vector; use The activation function activates the modulated heart rate time series associated semantic representation vector to obtain a heart rate time series semantic fully connected feature vector; The blood pressure time series association feature map is weighted along the channel dimension using each position feature value in the heart rate time series semantic fully connected feature vector to obtain the vital sign multimodal representation feature map.

8. The management system based on algorithm analysis technology according to claim 7 is characterized in that: The abnormal body state detection module is further used to: in the conversation process monitoring server, pass the multimodal representation feature map of the vital signs through a classifier-based monitor to obtain a monitoring result, and the monitoring result is used to indicate whether there is an abnormality in the body state of the conversation partner.

9. A management method based on algorithm analysis technology, characterized in that: include: Collecting a time series of vital sign data of the conversation partner during the conversation through a smart bracelet, wherein the vital sign data includes heart rate value and blood pressure value; Transmitting the time series of the vital sign data to a conversation process monitoring server through the wireless communication module of the smart bracelet; In the conversation process monitoring server, time series autocorrelation coding and time series pattern feature extraction are performed on the time series of the vital sign data to obtain a heart rate time series correlation feature graph and a blood pressure time series correlation feature graph; In the conversation process monitoring server, the heart rate time series associated feature map is input into an information excitation-compression module including a gated screening adaptive attention module and a feature multi-scale collapse module to obtain a heart rate time series associated semantic representation vector; In the conversation process monitoring server, the heart rate time series associated semantic representation vector and the blood pressure time series associated feature map are input into a semantically induced attention fusion network to obtain a vital sign multimodal representation feature map; Based on the multimodal representation feature graph of vital signs, determine whether the conversation partner's physical state is abnormal.

Citation Information

Patent Citations

  • Online defect detection system and method for carbon fiber sucker rod

    CN118429335A

  • Combustible gas intelligent detection system and method based on big data

    CN118482874A

  • Method for efficiently and statically fermenting cattle bed padding

    CN118805693A

  • Production method and equipment of energy-saving building material dilute stone concrete wallboard

    CN118809767A

  • Early warning analysis method and system for financial data and readable storage medium

    CN118822722A

Cited By

  • Facial paralysis recognition and grade evaluation system and method based on expression analysis technology

    CN118840771A

  • Data processing method and system based on virtual simulation practical training simulation system

    CN118942691A

  • Optical fiber winding detection method and device

    CN119276347A

  • Fault positioning method, device and system for optical fiber line

    CN119276354A

  • Photovoltaic module control system and method based on photovoltaic module voltage prediction

    CN120123669A