Remote fetal heart rate monitoring automatic analysis system and method based on smart watch

Through smart watches, collecting mixed signals and using artificial intelligence technology to perform signal decomposition and feature extraction on remote servers, the problem of difficulty in separating fetal heart rate signals by existing fetal heart monitors is solved, and more accurate fetal heart rate monitoring and more intelligent monitoring method.

CN118787329BActive Publication Date: 2024-11-19JILIN UNIVERSITY
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411266235.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-11
Publication Date
2024-11-19
Estimated Expiration
2044-09-11

AI Technical Summary

Technical Problem

Existing fetal heart rate monitors are difficult to accurately separate fetal heart rate signals from mixed signals, which are affected by ambient noise and maternal heart rate, resulting in inaccurate monitoring results.

Method used

Using a remote fetal heart monitoring system based on smart watches, the mixed signals are collected through the smart watch and transmitted to the remote server, and using artificial intelligence technology to perform signal decomposition and semantic feature extraction and enhancement of IMF component waveforms, clustering and mean calculations are performed to obtain fetal heart rate value. .

Benefits of technology

It realizes more accurate fetal heart rate data capture and monitoring, improves the accuracy and reliability of monitoring, and provides a more intelligent remote fetal heart monitoring method.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118787329B_ABST
    Figure CN118787329B_ABST
Patent Text Reader

Abstract

The present application provides a remote fetal heart monitoring automatic analysis system and method based on a smart watch, which relates to the field of intelligent signal analysis. It collects mixed signals through a smart watch and transmits the mixed signals to a remote server for fetal heart monitoring. In the remote server for fetal heart monitoring, artificial intelligence-based signal processing and analysis technology is used to decompose the mixed signals and extract and enhance the semantic features of the IMF component waveforms. Then, the fetal heart rate value is intelligently obtained based on the semantic features of the fetal signal waveform obtained after clustering and mean calculation of the semantic features of the waveforms of each IMF component, and sent to the smart watch. In this way, artificial intelligence technology can be used to decompose and extract the mixed signals of the heart rate of pregnant women and fetuses, which can more fully capture the key information in the signal, and then obtain more accurate fetal heart rate data, realizing more intelligent remote fetal heart monitoring.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of intelligent signal analysis, and more specifically, to a remote fetal heart monitoring automatic analysis system and method based on a smart watch. Background Art

[0002] A major challenge facing portable fetal heart rate monitors for home use on the current market is the difficulty in accurately separating the maternal and fetal heart rate signals from mixed signals, as well as eliminating other interfering factors, which brings considerable trouble to the health monitoring of pregnant women and fetuses. Specifically, existing fetal heart rate monitors usually use photoelectric sensors and microphones to collect heart rate signals. However, due to the limitations of signal processing technology and signal quality, the monitoring results are often affected by factors such as environmental noise and the mother's own heart rate. These problems make it difficult to accurately separate the fetal heart rate signal from the mixed signal, which in turn affects the accuracy and reliability of monitoring.

[0003] Based on this, the invention with publication number CN117918843A discloses a fetal ECG monitoring system based on data analysis, which first collects a mixed signal containing maternal and fetal signals through a wearable fetal heart monitor; then, through signal preprocessing, independent component analysis (ICA) is used to extract and offset the maternal signal, and then ICA is applied again to extract the fetal signal; finally, the time domain nonlinear transformation threshold method is used to determine the fetal ECG R wave peak, and the fetal heart rate is calculated through cluster detection and RR interval to achieve fetal heart rate monitoring.

[0004] However, although the above existing methods have achieved the separation of fetal signals and the calculation of fetal heart rate, they have many defects because they are based on traditional signal extraction and calculation methods for analysis. Specifically, ICA is an effective signal separation technology, but it may not be able to completely solve the problem of complex nonlinear signal mixing. That is, in actual situations, the heart rate signals of the mother and the fetus may not be completely independent, which may lead to inaccurate fetal signal separation. In addition, traditional signal feature extraction methods may not be able to fully capture important information in the signal. For example, the time domain nonlinear transformation threshold method may be limited when processing signals with high noise levels or nonlinear dynamics, making it difficult to accurately identify key information of the fetal electrocardiogram.

[0005] Therefore, an automatic analysis solution for remote fetal heart rate monitoring is needed. Summary of the invention

[0006] In order to solve the above technical problems, this application is proposed.

[0007] According to one aspect of the present application, a remote fetal heart monitoring automatic analysis system based on a smart watch is provided, which includes:

[0008] A mixed signal acquisition module, used to acquire a mixed signal collected by a smart watch, wherein the mixed signal includes a maternal signal and a fetal signal;

[0009] A mixed signal transmission module, used to transmit the mixed signal to a fetal heart monitoring remote server through the communication module of the smart watch;

[0010] A mixed signal transmission decomposition module, used for performing signal decomposition on the mixed signal at the fetal heart monitoring remote server to obtain a set of IMF components of the monitoring signal;

[0011] A monitoring signal IMF component waveform semantic feature extraction module is used to, at the fetal heart monitoring remote server, pass each monitoring signal IMF component in the set of monitoring signal IMF components through a signal waveform semantic feature extractor to obtain a set of monitoring signal IMF component waveform semantic feature vectors;

[0012] A monitoring signal IMF component waveform semantic enhancement module is used to input the set of monitoring signal IMF component waveform semantic feature vectors into a feature association optimization network based on sequence endogenous correlation semantic measurement in the fetal heart monitoring remote server to obtain a set of monitoring signal IMF component enhanced waveform semantic feature vectors;

[0013] A monitoring signal IMF component enhanced waveform clustering module is used to perform cluster analysis on the set of the monitoring signal IMF component enhanced waveform semantic feature vectors on the fetal heart monitoring remote server to obtain a first subset of the monitoring signal IMF component enhanced waveform semantic feature vectors as a maternal signal feature set and a second subset of the monitoring signal IMF component enhanced waveform semantic feature vectors as a fetal signal feature set;

[0014] A fetal signal waveform semantic calculation module, used for calculating, on the fetal heart monitoring remote server, a global mean vector of a second subset of the semantic feature vectors of the enhanced waveform of the IMF component of the monitoring signal as a fetal signal waveform semantic feature vector;

[0015] The fetal heart rate value generating module is used to obtain the fetal heart rate value based on the semantic feature vector of the fetal signal waveform, and send the fetal heart rate value to the smart watch.

[0016] In the above-mentioned remote fetal heart monitoring automatic analysis system based on a smart watch, the mixed signal transmission decomposition module is used to: perform CEEMDAN decomposition on the mixed signal at the fetal heart monitoring remote server to obtain a set of IMF components of the monitoring signal.

[0017] In the above-mentioned smart watch-based remote fetal heart monitoring automatic analysis system, the monitoring signal IMF component waveform semantic feature extraction module is used to: on the fetal heart monitoring remote server, each monitoring signal IMF component in the set of monitoring signal IMF components is respectively passed through a signal waveform semantic feature extractor based on a void convolutional neural network model to obtain a set of monitoring signal IMF component waveform semantic feature vectors.

[0018] In the above-mentioned remote fetal heart monitoring automatic analysis system based on a smart watch, the monitoring signal IMF component waveform semantic enhancement module includes: a monitoring signal IMF component waveform semantic association score vector calculation unit, which is used to calculate the semantic association score vector between any two monitoring signal IMF component waveform semantic feature vectors in the set of the monitoring signal IMF component waveform semantic feature vectors to obtain a set of monitoring signal IMF component waveform semantic association score vectors; a monitoring signal IMF component waveform sequence endogenous correlation global calculation unit, which is used to determine the endogenous correlation global representation vector of the monitoring signal IMF component waveform sequence based on the set of the monitoring signal IMF component waveform semantic association score vectors; a monitoring signal IMF component waveform semantic association optimization factor calculation unit, which is used to calculate the association optimization factor of each monitoring signal IMF component waveform semantic feature vector in the set of the monitoring signal IMF component waveform semantic feature vector based on the endogenous correlation global representation vector of the monitoring signal IMF component waveform sequence to obtain a sequence of monitoring signal IMF component waveform semantic association optimization factors; a monitoring signal IMF component waveform semantic association optimization weight factor calculation unit, which is used to input the sequence of the monitoring signal IMF component waveform semantic association optimization factors into the monitoring signal IMF component waveform semantic association optimization factor. An activation function is used to obtain a sequence of monitoring signal IMF component waveform semantic association optimization weight factors; a monitoring signal IMF component enhanced waveform semantic feature generation unit is used to use each monitoring signal IMF component waveform semantic association optimization weight factor in the sequence of monitoring signal IMF component waveform semantic association optimization weight factors as a weight, and weight each monitoring signal IMF component waveform semantic feature vector in the set of monitoring signal IMF component waveform semantic feature vectors to obtain the set of monitoring signal IMF component enhanced waveform semantic feature vectors.

[0019] In the above-mentioned smart watch-based remote fetal heart monitoring automatic analysis system, the monitoring signal IMF component waveform semantic association score vector calculation unit is used to: cascade any two monitoring signal IMF component waveform semantic feature vectors in the set of monitoring signal IMF component waveform semantic feature vectors to obtain a set of monitoring signal IMF component waveform semantic cascade feature vectors; multiply each monitoring signal IMF component waveform semantic cascade feature vector in the set of monitoring signal IMF component waveform semantic cascade feature vectors by the monitoring signal IMF component weight matrix, and then add the resultant vectors by position with the monitoring signal IMF component bias vector to obtain a set of monitoring signal IMF component waveform semantic association score vectors.

[0020] In the above-mentioned smart watch-based remote fetal heart monitoring automatic analysis system, the global calculation unit for the intrinsic correlation of the monitoring signal IMF component waveform sequence is used to: calculate the mean vector of the set of semantic association score vectors of the monitoring signal IMF component waveforms as the global representation vector of the intrinsic correlation of the monitoring signal IMF component waveform sequence.

[0021] In the above-mentioned remote fetal heart monitoring automatic analysis system based on a smart watch, the monitoring signal IMF component waveform semantic association optimization factor calculation unit is used to: multiply the global representation vector of the endogenous correlation of the monitoring signal IMF component waveform sequence with the global weight matrix of the monitoring signal IMF component waveform to obtain the global representation weight vector of the monitoring signal IMF component waveform; multiply each monitoring signal IMF component waveform semantic feature vector in the set of the monitoring signal IMF component waveform semantic feature vectors with the local weight matrix of the monitoring signal IMF component waveform to obtain a set of local representation weight vectors of the monitoring signal IMF component waveform; calculate each monitoring signal IMF component waveform local representation weight vector in the set of the monitoring signal IMF component waveform local representation weight vector, add the monitoring signal IMF component waveform global representation weight vector and the monitoring signal IMF component waveform bias vector according to position to obtain a set of monitoring signal IMF component waveform semantic association vectors; input the set of the monitoring signal IMF component waveform semantic association vectors into the Sigmoid function to obtain a sequence of the monitoring signal IMF component waveform semantic association optimization factors.

[0022] In the above-mentioned remote fetal heart rate monitoring automatic analysis system based on a smart watch, the fetal heart rate value generation module includes: a fetal heart rate decoding unit, which is used to input the fetal signal waveform semantic feature vector into a decoder-based fetal heart rate monitor in the fetal heart rate monitoring remote server to obtain a fetal heart rate decoding value; a fetal heart rate data transmission unit, which is used for the fetal heart rate monitoring remote server to send the fetal heart rate decoding value to the smart watch.

[0023] According to another aspect of the present application, a remote fetal heart rate monitoring automatic analysis method based on a smart watch is provided, which includes:

[0024] Acquire a mixed signal collected by a smart watch, where the mixed signal includes a maternal signal and a fetal signal;

[0025] Transmitting the mixed signal to a fetal heart monitoring remote server through a communication module of the smart watch;

[0026] At the fetal heart monitoring remote server, performing signal decomposition on the mixed signal to obtain a set of IMF components of the monitoring signal;

[0027] At the fetal heart monitoring remote server, each monitoring signal IMF component in the set of monitoring signal IMF components is respectively passed through a signal waveform semantic feature extractor to obtain a set of monitoring signal IMF component waveform semantic feature vectors;

[0028] In the fetal heart monitoring remote server, the set of semantic feature vectors of the IMF component waveform of the monitoring signal is input into a feature association optimization network based on the semantic measurement of sequence endogenous correlation to obtain a set of semantic feature vectors of the enhanced waveform of the IMF component of the monitoring signal;

[0029] In the fetal heart monitoring remote server, cluster analysis is performed on the set of semantic feature vectors of the IMF component enhanced waveform of the monitoring signal to obtain a first subset of the semantic feature vectors of the IMF component enhanced waveform of the monitoring signal as a maternal signal feature set and a second subset of the semantic feature vectors of the IMF component enhanced waveform of the monitoring signal as a fetal signal feature set;

[0030] At the fetal heart monitoring remote server, calculating a global mean vector of a second subset of the semantic feature vectors of the enhanced waveform of the IMF component of the monitoring signal as a fetal signal waveform semantic feature vector;

[0031] A fetal heart rate value is obtained based on the fetal signal waveform semantic feature vector, and the fetal heart rate value is sent to the smart watch.

[0032] In the above-mentioned automatic analysis method of remote fetal heart monitoring based on a smart watch, at the fetal heart monitoring remote server, the mixed signal is decomposed to obtain a set of IMF components of the monitoring signal, including: at the fetal heart monitoring remote server, the mixed signal is decomposed by CEEMDAN to obtain a set of IMF components of the monitoring signal.

[0033] Compared with the prior art, the remote fetal heart monitoring automatic analysis system and method based on a smart watch provided by the present application collects mixed signals through a smart watch, and transmits the mixed signals to a remote server for fetal heart monitoring, in which artificial intelligence-based signal processing and analysis technology is used to decompose the mixed signals and extract and enhance the semantic features of the IMF component waveforms, and then the fetal heart rate value is intelligently obtained based on the semantic features of the fetal signal waveform obtained after clustering and mean calculation of the semantic features of the waveforms of each IMF component, and sent to the smart watch. In this way, artificial intelligence technology can be used to decompose and extract the mixed signals of the heart rate of pregnant women and fetuses, which can more fully capture the key information in the signal, and then obtain more accurate fetal heart rate data, realizing more intelligent remote fetal heart monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0035] Figure 1 4 is a system block diagram of a remote fetal heart monitoring automatic analysis system based on a smart watch according to an embodiment of the present application.

[0036] Figure 2 Schematic diagram of data flow of a remote fetal heart rate monitoring automatic analysis system based on a smart watch according to an embodiment of the present application.

[0037] Figure 3 It is a block diagram of a waveform semantic enhancement module of an IMF component of a monitoring signal in a remote fetal heart monitoring automatic analysis system based on a smart watch according to an embodiment of the present application.

[0038] Figure 4 The present invention is a block diagram of a fetal heart rate value generation module in a remote fetal heart rate monitoring automatic analysis system based on a smart watch according to an embodiment of the present application.

[0039] Figure 5 Flow chart of a remote fetal heart rate monitoring automatic analysis method based on a smart watch according to an embodiment of the present application. DETAILED DESCRIPTION

[0040] 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.

[0041] In the current home portable fetal heart rate monitor market, a core challenge is how to accurately distinguish the maternal and fetal heart rate signals from the mixed signals and eliminate various interferences. This challenge has brought considerable trouble to the health monitoring of pregnant women and fetuses. Existing fetal heart rate monitoring equipment usually uses photoelectric sensors and microphones to collect heart rate data. However, due to the insufficiency of signal processing technology and the limitation of signal quality, the monitoring results may be affected by environmental noise and maternal heart rate, which makes it difficult to accurately extract the fetal heart rate signal from the mixed signal, thus affecting the accuracy and reliability of monitoring.

[0042] To address this problem, the invention with publication number CN117918843A proposes a fetal ECG monitoring system based on data analysis. The system first collects mixed signals containing maternal and fetal signals through wearable devices, then uses independent component analysis (ICA) to extract and eliminate maternal signals through signal preprocessing, and then uses ICA again to extract fetal signals. Finally, the system uses the time domain nonlinear transformation threshold method to determine the peak of the fetal ECG R wave, and monitors the fetal heart rate through cluster detection and calculation of the RR interval.

[0043] Although this method can achieve the separation of fetal signals and calculation of heart rate to a certain extent, it still has some limitations. Although ICA is an effective signal separation technology, it may not be perfect when dealing with complex nonlinear signal mixtures, because the maternal and fetal heart rate signals may not be completely independent, which may lead to inaccurate separation of fetal signals. In addition, traditional signal feature extraction methods may not be able to fully capture the key information in the signal. For example, when dealing with high-noise or nonlinear dynamic signals, the time domain nonlinear transformation threshold method may be limited, making it difficult to accurately identify the key features of the fetal ECG.

[0044] Based on this, the present application proposes an automatic analysis solution for remote fetal heart monitoring based on a smart watch, which collects mixed signals (maternal signals and fetal signals) through a smart watch, and the mixed signals are transmitted to a remote server for fetal heart monitoring, in which artificial intelligence-based signal processing and analysis technology is used to decompose the mixed signals and extract and enhance the semantic features of IMF (Intrinsic Mode Functions) component waveforms, and then the fetal heart rate value is intelligently obtained based on the semantic features of the fetal signal waveform obtained after clustering and mean calculation of the semantic features of each IMF component waveform, and sent to the smart watch. In this way, artificial intelligence technology can be used to decompose and extract the mixed signals of the pregnant woman and fetal heart rate, so as to more fully capture the key information in the signal, so that even if the maternal and fetal heart rate signals are not completely independent, the fetal signal can be more accurately separated to obtain more accurate fetal heart rate data, providing a more intelligent remote fetal heart rate monitoring method.

[0045] Figure 1 4 is a system block diagram of a remote fetal heart monitoring automatic analysis system based on a smart watch according to an embodiment of the present application. Figure 2 FIG. 1 is a data flow diagram of a remote fetal heart rate monitoring automatic analysis system based on a smart watch according to an embodiment of the present application. Figure 1 and Figure 2As shown, in the remote fetal heart monitoring automatic analysis system 100 based on the smart watch, it includes: a mixed signal acquisition module 110, which is used to obtain a mixed signal collected by the smart watch, and the mixed signal includes a maternal signal and a fetal signal; a mixed signal transmission module 120, which is used to transmit the mixed signal to the fetal heart monitoring remote server through the communication module of the smart watch; a mixed signal transmission decomposition module 130, which is used to perform signal decomposition on the mixed signal at the fetal heart monitoring remote server to obtain a set of IMF components of the monitoring signal; a monitoring signal IMF component waveform semantic feature extraction module 140, which is used to extract the waveform semantic feature vectors of the monitoring signal IMF component in the set of the monitoring signal IMF components at the fetal heart monitoring remote server respectively through a signal waveform semantic feature extractor to obtain a set of waveform semantic feature vectors of the monitoring signal IMF component; a monitoring signal IMF component waveform semantic enhancement module 150, which is used to extract the waveform semantic feature vectors of the monitoring signal IMF component at the fetal heart monitoring remote server. The set of semantic feature vectors is input into a feature association optimization network based on the semantic metric of sequence endogenous correlation to obtain a set of semantic feature vectors of the IMF component enhanced waveform of the monitoring signal; a monitoring signal IMF component enhanced waveform clustering module 160 is used to perform cluster analysis on the set of semantic feature vectors of the IMF component enhanced waveform of the monitoring signal at the fetal heart monitoring remote server to obtain a first subset of the semantic feature vectors of the IMF component enhanced waveform of the monitoring signal as a maternal signal feature set and a second subset of the semantic feature vectors of the IMF component enhanced waveform of the monitoring signal as a fetal signal feature set; a fetal signal waveform semantic calculation module 170 is used to calculate the global mean vector of the second subset of the semantic feature vectors of the IMF component enhanced waveform of the monitoring signal at the fetal heart monitoring remote server as a fetal signal waveform semantic feature vector; a fetal heart rate value generation module 180 is used to obtain a fetal heart rate value based on the fetal signal waveform semantic feature vector, and send the fetal heart rate value to the smart watch.

[0046] In an embodiment of the present application, the mixed signal acquisition module 110 and the mixed signal transmission module 120 are respectively used to obtain a mixed signal including a maternal signal and a fetal signal collected by a smart watch and to transmit the mixed signal to a fetal heart monitoring remote server through the communication module of the smart watch. It should be understood that the mixed signal contains an electrocardiogram signal from the pregnant woman herself (i.e., a maternal signal) and an electrocardiogram signal that can reflect the heart activity of the fetus (i.e., a fetal signal). In addition, the mixed signal also contains some environmental noise. The interference of environmental noise and maternal signals may lead to inaccurate signal feature extraction, thereby affecting the calculation accuracy of the fetal heart rate. The present application can extract the fetal heart rate information from the mixed signal through an effective signal separation and feature extraction process, thereby ensuring the accuracy and reliability of fetal heart monitoring. Considering that the processing of mixed signals involves complex signal separation and feature extraction algorithms, these algorithms usually require higher computing resources, and the remote server usually has stronger processing capabilities and can perform these calculations more efficiently. Therefore, after the smart watch collects the mixed signal, the collected signal data needs to be transmitted to the fetal heart monitoring remote server through the communication module of the smart watch.

[0047] In an embodiment of the present application, the mixed signal transmission decomposition module 130 is used to perform signal decomposition on the mixed signal at the fetal heart monitoring remote server to obtain a set of IMF components of the monitoring signal. Specifically, in an embodiment of the present application, the mixed signal transmission decomposition module 130 is used to: perform CEEMDAN decomposition on the mixed signal at the fetal heart monitoring remote server to obtain a set of IMF components of the monitoring signal. It should be understood that different time scales in the mixed signal have different frequency oscillation modes. Therefore, in order to more carefully and accurately analyze and process the signal frequency information in different time segments, in the technical solution of the present application, the mixed signal is decomposed by CEEMDAN to obtain a set of IMF components of the monitoring signal. CEEMDAN is a signal processing technology based on empirical mode decomposition (EMD). It enhances the stability and effect of EMD by introducing auxiliary noise, and is suitable for processing nonlinear and non-stationary signals. The basic idea of ​​CEEMDAN is to add Gaussian white noise to the original signal, and then perform EMD decomposition on the new signal, and obtain a more stable and accurate decomposition result by averaging the generated multiple modal components. This process will be repeated many times, and the amplitude of the white noise added each time will gradually decrease to ensure the accuracy of the final result. By performing CEEMDAN decomposition on the mixed signal, the essential characteristics of the signal can be better reflected, which is helpful for the subsequent extraction of key information such as fetal heart rate.

[0048] In an embodiment of the present application, the monitoring signal IMF component waveform semantic feature extraction module 140 is used to, at the fetal heart monitoring remote server, pass each monitoring signal IMF component in the set of monitoring signal IMF components through a signal waveform semantic feature extractor to obtain a set of monitoring signal IMF component waveform semantic feature vectors. Specifically, in an embodiment of the present application, the monitoring signal IMF component waveform semantic feature extraction module 140 is used to: at the fetal heart monitoring remote server, pass each monitoring signal IMF component in the set of monitoring signal IMF components through a signal waveform semantic feature extractor based on a hole convolutional neural network model to obtain a set of monitoring signal IMF component waveform semantic feature vectors. It should be understood that the key features in the fetal heart rate signal may span a longer period of time, and the hole convolution can expand the receptive field without increasing the network depth and capture longer-distance semantic dependencies. Therefore, in order to further capture and mine the key waveform semantic feature information in each monitoring signal IMF component, in the technical solution of the present application, each monitoring signal IMF component in the set of monitoring signal IMF components is respectively passed through a signal waveform semantic feature extractor based on a hole convolutional neural network model to obtain a set of monitoring signal IMF component waveform semantic feature vectors. In particular, the signal waveform semantic feature extraction based on the hole convolutional neural network model here is essentially a hole convolutional neural network model.

[0049] In an embodiment of the present application, the monitoring signal IMF component waveform semantic enhancement module 150 is used to input the set of the monitoring signal IMF component waveform semantic feature vectors into a feature association optimization network based on the semantic measurement of the sequence endogenous correlation in the fetal heart monitoring remote server to obtain a set of monitoring signal IMF component enhanced waveform semantic feature vectors. Accordingly, considering that there are different degrees of semantic association or difference between the semantic features of the individual monitoring signal IMF component waveforms in the set of the monitoring signal IMF component waveform semantic feature vectors, these subtle semantic association features have an important influence and effect on the subsequent calculation of the fetal heart rate. In order to semantically optimize each monitoring signal IMF component waveform semantic feature vector based on the semantic association features between the features, in the technical solution of the present application, the set of the monitoring signal IMF component waveform semantic feature vectors is input into a feature association optimization network based on the semantic measurement of the sequence endogenous correlation to obtain a set of monitoring signal IMF component enhanced waveform semantic feature vectors. In other words, the feature association optimization network based on the semantic measurement of the sequence endogenous correlation is a method that focuses on improving the quality of feature representation by quantifying and optimizing the semantic associations between feature vectors. In detail, first, the intrinsic semantic connection between different features is quantified by calculating the semantic correlation between any two vectors in the set of semantic feature vectors of the IMF component waveform of the monitoring signal, and a set of semantic association score vectors of the IMF component waveform of the monitoring signal is generated. Then, the overall structural monitoring signal IMF component waveform semantic information of the entire sequence is aggregated and captured by calculating the mean vector of the set of semantic association score vectors of the IMF component waveform of the monitoring signal to obtain the global representation vector of the intrinsic correlation of the IMF component waveform sequence of the monitoring signal. Next, based on the global representation vector of the intrinsic correlation of the IMF component waveform sequence of the monitoring signal, each monitoring signal IMF component waveform semantic feature vector is optimized for association, so as to highlight the semantic similarity or difference between each feature and the overall global feature. Furthermore, each feature after association optimization is input into the Softmax activation function for weight normalization and nonlinear mapping, so as to automatically adjust the parameters to achieve optimal monitoring signal IMF component waveform feature extraction and association. Finally, the waveform semantic association optimization factor of each monitoring signal IMF component obtained after mapping is used as a weight to perform weighted optimization on the waveform semantic feature vector of each monitoring signal IMF component to obtain the enhanced waveform semantic feature of the monitoring signal IMF component containing association information and original feature information.

[0050] Specifically, Figure 3 FIG. 1 is a block diagram of a waveform semantic enhancement module for the IMF component of a monitoring signal in a remote fetal heart rate monitoring automatic analysis system based on a smart watch according to an embodiment of the present application. Figure 3As shown, the monitoring signal IMF component waveform semantic enhancement module 150 includes: a monitoring signal IMF component waveform semantic association score vector calculation unit 151, which is used to calculate the semantic association score vector between any two monitoring signal IMF component waveform semantic feature vectors in the set of the monitoring signal IMF component waveform semantic feature vectors to obtain a set of monitoring signal IMF component waveform semantic association score vectors; a monitoring signal IMF component waveform sequence endogenous correlation global calculation unit 152, which is used to determine the endogenous correlation global representation vector of the monitoring signal IMF component waveform sequence based on the set of the monitoring signal IMF component waveform semantic association score vectors; a monitoring signal IMF component waveform semantic association optimization factor calculation unit 153, which is used to calculate each endogenous correlation global representation vector of the monitoring signal IMF component waveform sequence in the set of the monitoring signal IMF component waveform semantic feature vectors based on the endogenous correlation global representation vector of the monitoring signal IMF component waveform sequence. The monitoring signal IMF component waveform semantic association optimization factor is used to obtain a sequence of monitoring signal IMF component waveform semantic association optimization factors; the monitoring signal IMF component waveform semantic association optimization weight factor calculation unit 154 is used to input the sequence of monitoring signal IMF component waveform semantic association optimization factors into the Softmax activation function to obtain a sequence of monitoring signal IMF component waveform semantic association optimization weight factors; the monitoring signal IMF component enhanced waveform semantic feature generation unit 155 is used to use each monitoring signal IMF component waveform semantic association optimization weight factor in the sequence of monitoring signal IMF component waveform semantic association optimization weight factors as a weight, and respectively weight each monitoring signal IMF component waveform semantic feature vector in the set of monitoring signal IMF component waveform semantic feature vectors to obtain a set of monitoring signal IMF component enhanced waveform semantic feature vectors.

[0051] More specifically, in an embodiment of the present application, the monitoring signal IMF component waveform semantic association score vector calculation unit 151 is used to: cascade any two monitoring signal IMF component waveform semantic feature vectors in the set of monitoring signal IMF component waveform semantic feature vectors to obtain a set of monitoring signal IMF component waveform semantic cascade feature vectors; multiply each monitoring signal IMF component waveform semantic cascade feature vector in the set of monitoring signal IMF component waveform semantic cascade feature vectors by the monitoring signal IMF component weight matrix, and then add the resultant vectors by position with the monitoring signal IMF component bias vector to obtain the set of monitoring signal IMF component waveform semantic association score vectors.

[0052] More specifically, in an embodiment of the present application, the global calculation unit 152 of the intrinsic correlation of the monitoring signal IMF component waveform sequence is used to: calculate the mean vector of the set of semantic association score vectors of the monitoring signal IMF component waveform as the global representation vector of the intrinsic correlation of the monitoring signal IMF component waveform sequence.

[0053] More specifically, in an embodiment of the present application, the monitoring signal IMF component waveform semantic association optimization factor calculation unit 153 is used to: multiply the global representation vector of the endogenous correlation of the monitoring signal IMF component waveform sequence with the global weight matrix of the monitoring signal IMF component waveform to obtain the global representation weight vector of the monitoring signal IMF component waveform; multiply each monitoring signal IMF component waveform semantic feature vector in the set of the monitoring signal IMF component waveform semantic feature vectors with the local weight matrix of the monitoring signal IMF component waveform to obtain a set of local representation weight vectors of the monitoring signal IMF component waveform; calculate each monitoring signal IMF component waveform local representation weight vector in the set of the monitoring signal IMF component waveform local representation weight vectors, add the monitoring signal IMF component waveform global representation weight vector and the monitoring signal IMF component waveform bias vector by position to obtain a set of monitoring signal IMF component waveform semantic association vectors; input the set of monitoring signal IMF component waveform semantic association vectors into the Sigmoid function to obtain a sequence of monitoring signal IMF component waveform semantic association optimization factors.

[0054] In the embodiment of the present application, specifically, the monitoring signal IMF component waveform semantic enhancement module 150 is used to: input the set of the monitoring signal IMF component waveform semantic feature vectors into a feature association optimization network based on the sequence endogenous correlation semantic measurement, and process it with the following feature association optimization formula to obtain the set of the monitoring signal IMF component enhanced waveform semantic feature vectors; wherein the feature association optimization formula is:

[0055] ;

[0056] ;

[0057] ;

[0058] ;

[0059] ;

[0060] in, is the set of semantic feature vectors of the IMF component waveform of the monitoring signal, represents a set of feature vectors, represents the number of feature vectors in the set of semantic feature vectors of the IMF component waveform of the monitoring signal, , , , and are respectively the first, second, and third in the set of semantic feature vectors of the IMF component waveform of the monitoring signal , and The IMF component waveform semantic feature vector of the monitoring signal, For cascade processing operations, is the monitoring signal IMF component weight matrix, is the monitoring signal IMF component bias vector, is the first in the set of semantic association score vectors of the IMF component waveform of the monitoring signal The semantic association score vector of the IMF component waveform of the monitoring signal, is the number of vectors in the set of semantic association score vectors of the IMF component waveform of the monitoring signal, is the global representation vector of the endogenous correlation of the IMF component waveform sequence of the monitoring signal, and are respectively the monitoring signal IMF component waveform global weight matrix and the monitoring signal IMF component waveform local weight matrix, is the monitoring signal IMF component waveform bias vector, yes function, yes function, The first in the sequence of weight factors for semantic association optimization of the IMF component waveform of the monitoring signal The weight factor of the IMF component waveform semantic association optimization of the monitoring signal is is the first in the set of semantic feature vectors of the IMF component enhanced waveform of the monitoring signal The IMF components of the monitoring signal enhance the waveform semantic feature vector.

[0061] In an embodiment of the present application, the monitoring signal IMF component enhanced waveform clustering module 160 is used to perform cluster analysis on the set of the monitoring signal IMF component enhanced waveform semantic feature vectors at the fetal heart monitoring remote server to obtain a first subset of the monitoring signal IMF component enhanced waveform semantic feature vectors as a maternal signal feature set and a second subset of the monitoring signal IMF component enhanced waveform semantic feature vectors as a fetal signal feature set. Accordingly, considering that the set of the monitoring signal IMF component enhanced waveform semantic feature vectors contains maternal and fetal signals, and maternal signals and fetal signals usually overlap with each other, and maternal signals (such as maternal heartbeat, breathing, etc.) may interfere with the detection of fetal signals. Therefore, in order to improve the accuracy of subsequent fetal heart rate detection and reduce the interference of maternal signals on fetal signal detection, in the technical solution of the present application, a cluster analysis is performed on the set of semantic feature vectors of the IMF component enhanced waveform of the monitoring signal to obtain a first subset of the semantic feature vectors of the IMF component enhanced waveform of the monitoring signal as the maternal signal feature set and a second subset of the semantic feature vectors of the IMF component enhanced waveform of the monitoring signal as the fetal signal feature set. In this way, the maternal signal and the fetal signal can be distinguished more clearly, thereby improving the accuracy of fetal heart rate detection.

[0062] In the embodiment of the present application, the fetal signal waveform semantic calculation module 170 is used to calculate the global mean vector of the second subset of the semantic feature vector of the enhanced waveform of the IMF component of the monitoring signal as the fetal signal waveform semantic feature vector at the fetal heart monitoring remote server. It should be understood that by calculating the global mean vector of the second subset of the semantic feature vector of the enhanced waveform of the IMF component of the monitoring signal as the fetal signal waveform semantic feature vector, the key overall semantic features of the fetal signal can be further refined and summarized, while reducing the signal fluctuations caused by noise or other random variations, so as to provide a clearer and more stable representation of the fetal signal features, so as to accurately identify and track the fetal heart rate.

[0063] In the embodiment of the present application, the fetal heart rate value generating module 180 is used to obtain the fetal heart rate value based on the fetal signal waveform semantic feature vector, and send the fetal heart rate value to the smart watch. Specifically, Figure 4 FIG. 1 is a block diagram of a fetal heart rate value generation module in a remote fetal heart rate monitoring automatic analysis system based on a smart watch according to an embodiment of the present application. Figure 4As shown, the fetal heart rate value generation module 180 includes: a fetal heart rate decoding unit 181, which is used to input the fetal signal waveform semantic feature vector into the decoder-based fetal heart rate monitor in the fetal heart monitoring remote server to obtain the fetal heart rate decoding value; a fetal heart rate data transmission unit 182, which is used for the fetal heart monitoring remote server to send the fetal heart rate decoding value to the smart watch. That is, the fetal signal waveform semantic feature obtained by global mean calculation using the second subset of the waveform semantic feature vector of the IMF component of the monitoring signal is decoded and processed to intelligently obtain the fetal heart rate value and send it to the smart watch. In this way, the real-time monitoring of the heart rate of pregnant women and fetuses can be achieved through smart watches, and continuous health data can be provided. At the same time, the mixed signal is decomposed and extracted by artificial intelligence technology, which can more fully capture the key information in the signal, so that even if the maternal and fetal heart rate signals are not completely independent, the fetal signal can be more accurately separated to obtain more accurate fetal heart rate data, thereby providing more intelligent remote fetal heart monitoring automatic analysis.

[0064] In particular, considering that the set of enhanced waveform semantic feature vectors of the monitoring signal IMF component respectively represents the image semantic correlation measurement association characteristics within the single sample domain of the waveform image semantic features of each monitoring signal IMF component, when further clustering analysis and global mean calculation are performed, the mapping difference between the single sample domain image semantic feature representation in the full sample domain under clustering and mean aggregation will cause aggregation alignment conflicts between sample domains, thereby causing the loss of aggregation key information and affecting the expression effect of the fetal signal waveform semantic feature vector.

[0065] Based on this, in a preferred example, the fetal signal waveform semantic feature vector is input into the decoder-based fetal heart rate monitor to obtain a fetal heart rate decoding value, including the steps of: clustering all feature values ​​of the fetal signal waveform semantic feature vector based on the distance between feature values, such as the difference absolute value distance, and determining the ratio of the number of features of the cluster features to the number of features of the fetal signal waveform semantic feature vector; arranging the cluster features into a fetal signal waveform semantic cluster vector, and dividing the second norm of the fetal signal waveform semantic cluster vector by the second norm of the fetal signal waveform semantic feature vector to obtain a first fetal signal waveform semantic clustering related weight value; dividing a first power value of the one norm of the fetal signal waveform semantic cluster vector exponentially by the ratio value by a second power value of the one norm of the fetal signal waveform semantic feature vector exponentially by the ratio value to obtain a second fetal signal waveform semantic clustering related weight value. Weight value; for each eigenvalue of the fetal signal waveform semantic clustering vector, multiply it by the inverse of the difference between the weight value associated with the first fetal signal waveform semantic clustering and the weight value associated with the second fetal signal waveform semantic clustering to obtain the optimized eigenvalue of the fetal signal waveform semantic clustering vector; for each eigenvalue outside the cluster in the fetal signal waveform semantic feature vector, multiply it by the inverse of the sum of the weight value associated with the first fetal signal waveform semantic clustering and the weight value associated with the second fetal signal waveform semantic clustering to obtain the optimized out-of-class eigenvalue of the fetal signal waveform semantic feature vector; the optimized eigenvalue of the fetal signal waveform semantic clustering vector and the optimized out-of-class eigenvalue of the fetal signal waveform semantic feature vector are combined to form an optimized fetal signal waveform semantic feature vector; the optimized fetal signal waveform semantic feature vector is input into the decoder-based fetal heart rate monitor to obtain a fetal heart rate decoding value.

[0066] The optimized representation of the semantic feature vector of the fetal signal waveform is:

[0067] ;

[0068] ;

[0069] in, is each eigenvalue of the fetal signal waveform semantic clustering vector, is the fetal signal waveform semantic feature vector, is the fetal signal waveform semantic clustering vector, is the number of features of the fetal signal waveform semantic feature vector, is the number of features of the fetal signal waveform semantic clustering vector, is the ratio of the number of features of the fetal signal waveform semantic clustering vector to the number of features of the fetal signal waveform semantic feature vector, represents the clustering feature set corresponding to the semantic clustering vector of the fetal signal waveform, and Represents the binary norm and the uninorm of the vector respectively. Power, is each eigenvalue of the optimized fetal signal waveform semantic feature vector.

[0070] In order to avoid the loss of key suffix semantic information of the fetal signal waveform semantic feature vector relative to the original feature set as a whole due to aggregation conflict, the cluster ratio of the number of eigenvalues ​​of the fetal signal waveform semantic cluster vector relative to the number of features of the fetal signal waveform semantic feature vector is used as a decision function to perform an adversarial decision of the set absolute representation of the one-norm of the fetal signal waveform semantic cluster vector and the fetal signal waveform semantic feature vector, and positive and negative interactions are performed with the cluster intrinsic conflict representation of the two-norm of the fetal signal waveform semantic cluster vector and the fetal signal waveform semantic feature vector, respectively, to construct a solid alignment guardrail of the fetal signal waveform semantic feature vector based on the aggregated features and the original feature set as a whole, thereby realizing the mitigation of the intention of the harmful information loss of the fetal signal waveform semantic feature vector based on the transferability of the aggregation risk, improving the expression effect of the fetal signal waveform semantic feature vector, and thus improving the accuracy of the fetal heart rate decoding value obtained by the fetal signal waveform semantic feature vector input into the decoder-based fetal heart rate monitor. In this way, real-time monitoring of the heart rate of pregnant women and fetuses can be achieved through smart watches, and continuous health data can be provided. At the same time, the use of artificial intelligence technology to decompose and extract mixed signals can more fully capture the key information in the signal, so that even if the maternal and fetal heart rate signals are not completely independent, the fetal signal can be more accurately separated to obtain more accurate fetal heart rate data, thereby providing more intelligent remote fetal heart monitoring automatic analysis.

[0071] In summary, the remote fetal heart monitoring automatic analysis system 100 based on the smart watch according to the embodiment of the present application is explained, which collects mixed signals through the smart watch and transmits the mixed signals to the fetal heart monitoring remote server, and uses artificial intelligence-based signal processing and analysis technology in the fetal heart monitoring remote server to decompose the mixed signal and extract and enhance the semantic features of the IMF component waveforms, and then obtain the fetal heart rate value intelligently based on the semantic features of the fetal signal waveform obtained after clustering and mean calculation of the semantic features of the waveforms of each IMF component, and send it to the smart watch. In this way, the mixed signal of the heart rate of pregnant women and fetuses can be decomposed and extracted by artificial intelligence technology, and the key information in the signal can be more fully captured, thereby obtaining more accurate fetal heart rate data, and realizing more intelligent remote fetal heart monitoring.

[0072] As described above, the remote fetal heart monitoring automatic analysis system 100 based on a smart watch according to the embodiment of the present application can be implemented in various terminal devices, such as a server for remote fetal heart monitoring automatic analysis based on a smart watch. In one example, the remote fetal heart monitoring automatic analysis system 100 based on a smart watch according to the embodiment of the present application can be integrated into the terminal device as a software module and / or a hardware module. For example, the remote fetal heart monitoring automatic analysis system 100 based on a smart watch can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the remote fetal heart monitoring automatic analysis system 100 based on a smart watch can also be one of the many hardware modules of the terminal device.

[0073] Alternatively, in another example, the smart watch-based remote fetal heart monitoring automatic analysis system 100 and the terminal device may also be separate devices, and the smart watch-based remote fetal heart monitoring automatic analysis system 100 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0074] Figure 5 FIG. 1 is a flow chart of a remote fetal heart rate monitoring automatic analysis method based on a smart watch according to an embodiment of the present application. Figure 5As shown, in the automatic analysis method of remote fetal heart monitoring based on a smart watch, it includes: S110, obtaining a mixed signal collected by the smart watch, wherein the mixed signal includes a maternal signal and a fetal signal; S120, transmitting the mixed signal to a fetal heart monitoring remote server through the communication module of the smart watch; S130, at the fetal heart monitoring remote server, performing signal decomposition on the mixed signal to obtain a set of IMF components of the monitoring signal; S140, at the fetal heart monitoring remote server, passing each monitoring signal IMF component in the set of IMF components of the monitoring signal through a signal waveform semantic feature extractor to obtain a set of waveform semantic feature vectors of the IMF components of the monitoring signal; S150, at the fetal heart monitoring remote server, inputting the set of waveform semantic feature vectors of the IMF components of the monitoring signal into a sequence-based A feature association optimization network of endogenous correlation semantic metric is used to obtain a set of semantic feature vectors of enhanced waveforms of IMF components of monitoring signals; S160, in the fetal heart monitoring remote server, a cluster analysis is performed on the set of semantic feature vectors of enhanced waveforms of IMF components of monitoring signals to obtain a first subset of semantic feature vectors of enhanced waveforms of IMF components of monitoring signals as a maternal signal feature set and a second subset of semantic feature vectors of enhanced waveforms of IMF components of monitoring signals as a fetal signal feature set; S170, in the fetal heart monitoring remote server, a global mean vector of the second subset of semantic feature vectors of enhanced waveforms of IMF components of monitoring signals is calculated as a fetal signal waveform semantic feature vector; S180, based on the semantic feature vector of the fetal signal waveform, a fetal heart rate value is obtained, and the fetal heart rate value is sent to the smart watch.

[0075] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned remote fetal heart rate monitoring automatic analysis method based on a smart watch have been referred to above. Figures 1 to 4 The invention has been introduced in detail in the description of the remote fetal heart monitoring automatic analysis system based on a smart watch, and therefore, its repeated description will be omitted.

[0076] In summary, the automatic analysis method of remote fetal heart monitoring based on a smart watch according to the embodiment of the present application is explained, which collects mixed signals through a smart watch and transmits the mixed signals to a remote server for fetal heart monitoring, and uses artificial intelligence-based signal processing and analysis technology in the remote server for fetal heart monitoring to decompose the mixed signals and extract and enhance the semantic features of the IMF component waveforms, thereby intelligently obtaining the fetal heart rate value based on the semantic features of the fetal signal waveform obtained after clustering and mean calculation of the semantic features of the waveforms of each IMF component, and sending it to the smart watch. In this way, artificial intelligence technology can be used to decompose and extract the mixed signals of the heart rate of pregnant women and fetuses, so that the key information in the signal can be more fully captured, thereby obtaining more accurate fetal heart rate data, and realizing more intelligent remote fetal heart monitoring.

[0077] It should also be noted that in the system and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0078] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features of the present application.

[0079] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A remote fetal heart rate monitoring automatic analysis system based on a smart watch, characterized in that: include: A mixed signal acquisition module, used to acquire a mixed signal collected by a smart watch, wherein the mixed signal includes a maternal signal and a fetal signal; A mixed signal transmission module, used to transmit the mixed signal to a fetal heart monitoring remote server through the communication module of the smart watch; A mixed signal transmission decomposition module, used for performing signal decomposition on the mixed signal at the fetal heart monitoring remote server to obtain a set of IMF components of the monitoring signal; A monitoring signal IMF component waveform semantic feature extraction module is used to, at the fetal heart monitoring remote server, pass each monitoring signal IMF component in the set of monitoring signal IMF components through a signal waveform semantic feature extractor to obtain a set of monitoring signal IMF component waveform semantic feature vectors; A monitoring signal IMF component waveform semantic enhancement module is used to input the set of monitoring signal IMF component waveform semantic feature vectors into a feature association optimization network based on sequence endogenous correlation semantic measurement in the fetal heart monitoring remote server to obtain a set of monitoring signal IMF component enhanced waveform semantic feature vectors; A monitoring signal IMF component enhanced waveform clustering module is used to perform cluster analysis on the set of the monitoring signal IMF component enhanced waveform semantic feature vectors on the fetal heart monitoring remote server to obtain a first subset of the monitoring signal IMF component enhanced waveform semantic feature vectors as a maternal signal feature set and a second subset of the monitoring signal IMF component enhanced waveform semantic feature vectors as a fetal signal feature set; A fetal signal waveform semantic calculation module, used for calculating, on the fetal heart monitoring remote server, a global mean vector of a second subset of the semantic feature vectors of the enhanced waveform of the IMF component of the monitoring signal as a fetal signal waveform semantic feature vector; A fetal heart rate value generating module, used to obtain a fetal heart rate value based on the fetal signal waveform semantic feature vector, and send the fetal heart rate value to the smart watch; The mixed signal transmission decomposition module is used to: perform CEEMDAN decomposition on the mixed signal at the fetal heart monitoring remote server to obtain a set of IMF components of the monitoring signal; The monitoring signal IMF component waveform semantic feature extraction module is used to: on the fetal heart monitoring remote server, pass each monitoring signal IMF component in the set of monitoring signal IMF components through a signal waveform semantic feature extractor based on a hole convolutional neural network model to obtain a set of monitoring signal IMF component waveform semantic feature vectors; The monitoring signal IMF component waveform semantic enhancement module includes: A monitoring signal IMF component waveform semantic association score vector calculation unit, used to calculate the semantic association score vector between any two monitoring signal IMF component waveform semantic feature vectors in the set of monitoring signal IMF component waveform semantic feature vectors to obtain a set of monitoring signal IMF component waveform semantic association score vectors; A monitoring signal IMF component waveform sequence endogenous correlation global calculation unit, used to determine a monitoring signal IMF component waveform sequence endogenous correlation global representation vector based on a set of monitoring signal IMF component waveform semantic association score vectors; A monitoring signal IMF component waveform semantic association optimization factor calculation unit is used to calculate the association optimization factor of each monitoring signal IMF component waveform semantic feature vector in the set of the monitoring signal IMF component waveform semantic feature vectors based on the global representation vector of the endogenous correlation of the monitoring signal IMF component waveform sequence to obtain a sequence of monitoring signal IMF component waveform semantic association optimization factors; A monitoring signal IMF component waveform semantic association optimization weight factor calculation unit, used for inputting the sequence of monitoring signal IMF component waveform semantic association optimization factors into a Softmax activation function to obtain a sequence of monitoring signal IMF component waveform semantic association optimization weight factors; The monitoring signal IMF component enhanced waveform semantic feature generating unit is used to use each monitoring signal IMF component waveform semantic association optimization weight factor in the sequence of the monitoring signal IMF component waveform semantic association optimization weight factors as a weight, and weight each monitoring signal IMF component waveform semantic feature vector in the set of the monitoring signal IMF component waveform semantic feature vectors to obtain the set of the monitoring signal IMF component enhanced waveform semantic feature vectors.

2. The remote fetal heart rate monitoring automatic analysis system based on a smart watch according to claim 1 is characterized in that: The monitoring signal IMF component waveform semantic association score vector calculation unit is used to: Cascading any two monitoring signal IMF component waveform semantic feature vectors in the set of monitoring signal IMF component waveform semantic feature vectors to obtain a set of monitoring signal IMF component waveform semantic cascade feature vectors; Each monitoring signal IMF component waveform semantic cascade feature vector in the set of the monitoring signal IMF component waveform semantic cascade feature vectors is multiplied by the monitoring signal IMF component weight matrix and then added positionally with the monitoring signal IMF component bias vector to obtain the set of monitoring signal IMF component waveform semantic association score vectors.

3. The remote fetal heart rate monitoring automatic analysis system based on a smart watch according to claim 2 is characterized in that: The monitoring signal IMF component waveform sequence endogenous correlation global calculation unit is used to calculate the mean vector of the set of monitoring signal IMF component waveform semantic association score vectors as the monitoring signal IMF component waveform sequence endogenous correlation global representation vector.

4. The remote fetal heart rate monitoring automatic analysis system based on a smart watch according to claim 3 is characterized in that: The monitoring signal IMF component waveform semantic association optimization factor calculation unit is used to: Multiplying the global representation vector of the endogenous correlation of the monitoring signal IMF component waveform sequence with the global weight matrix of the monitoring signal IMF component waveform to obtain a global representation weight vector of the monitoring signal IMF component waveform; Multiplying each monitoring signal IMF component waveform semantic feature vector in the set of monitoring signal IMF component waveform semantic feature vectors by a monitoring signal IMF component waveform local weight matrix to obtain a set of monitoring signal IMF component waveform local representation weight vectors; Calculate each monitoring signal IMF component waveform local representation weight vector in the set of the monitoring signal IMF component waveform local representation weight vectors, add the monitoring signal IMF component waveform global representation weight vector and the monitoring signal IMF component waveform bias vector according to position to obtain a set of monitoring signal IMF component waveform semantic association vectors; The set of the semantic association vectors of the IMF component waveform of the monitoring signal is input into a Sigmoid function to obtain a sequence of semantic association optimization factors of the IMF component waveform of the monitoring signal.

5. The remote fetal heart rate monitoring automatic analysis system based on a smart watch according to claim 4 is characterized in that: The fetal heart rate value generating module comprises: A fetal heart rate decoding unit, configured to input the fetal signal waveform semantic feature vector into a decoder-based fetal heart rate monitor at the fetal heart rate monitoring remote server to obtain a fetal heart rate decoding value; The fetal heart rate data transmission unit is used for the fetal heart rate monitoring remote server to send the fetal heart rate decoding value to the smart watch.

6. A remote fetal heart rate monitoring automatic analysis method based on a smart watch, using the remote fetal heart rate monitoring automatic analysis system based on a smart watch according to claim 1, characterized in that: include: Acquire a mixed signal collected by a smart watch, where the mixed signal includes a maternal signal and a fetal signal; Transmitting the mixed signal to a fetal heart monitoring remote server through a communication module of the smart watch; At the fetal heart monitoring remote server, performing signal decomposition on the mixed signal to obtain a set of IMF components of the monitoring signal; At the fetal heart monitoring remote server, each monitoring signal IMF component in the set of monitoring signal IMF components is respectively passed through a signal waveform semantic feature extractor to obtain a set of monitoring signal IMF component waveform semantic feature vectors; In the fetal heart monitoring remote server, the set of semantic feature vectors of the IMF component waveform of the monitoring signal is input into a feature association optimization network based on the semantic measurement of sequence endogenous correlation to obtain a set of semantic feature vectors of the enhanced waveform of the IMF component of the monitoring signal; In the fetal heart monitoring remote server, cluster analysis is performed on the set of semantic feature vectors of the IMF component enhanced waveform of the monitoring signal to obtain a first subset of the semantic feature vectors of the IMF component enhanced waveform of the monitoring signal as a maternal signal feature set and a second subset of the semantic feature vectors of the IMF component enhanced waveform of the monitoring signal as a fetal signal feature set; At the fetal heart monitoring remote server, calculating a global mean vector of a second subset of the semantic feature vectors of the enhanced waveform of the IMF component of the monitoring signal as a fetal signal waveform semantic feature vector; A fetal heart rate value is obtained based on the fetal signal waveform semantic feature vector, and the fetal heart rate value is sent to the smart watch.

Citation Information

Cited By

  • Flexible intelligent maternal and fetal monitoring belly supporting belt

    CN120324170A