CN116530996B - Low-measurement-load-oriented electrocardiographic data abnormality early warning method and system - Google Patents
Low-measurement-load-oriented electrocardiographic data abnormality early warning method and system Download PDFInfo
- Publication number
- CN116530996B CN116530996B CN202310754754.5A CN202310754754A CN116530996B CN 116530996 B CN116530996 B CN 116530996B CN 202310754754 A CN202310754754 A CN 202310754754A CN 116530996 B CN116530996 B CN 116530996B
- Authority
- CN
- China
- Prior art keywords
- abnormal
- track
- heart rhythm
- fingerprint
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 61
- 230000005856 abnormality Effects 0.000 title claims abstract description 31
- 230000002159 abnormal effect Effects 0.000 claims abstract description 180
- 230000033764 rhythmic process Effects 0.000 claims abstract description 144
- 238000013528 artificial neural network Methods 0.000 claims abstract description 69
- 238000012544 monitoring process Methods 0.000 claims abstract description 62
- 208000024891 symptom Diseases 0.000 claims abstract description 42
- 238000012549 training Methods 0.000 claims abstract description 35
- 238000005259 measurement Methods 0.000 claims abstract description 22
- 206010003119 arrhythmia Diseases 0.000 claims abstract description 19
- 230000006793 arrhythmia Effects 0.000 claims abstract description 19
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 16
- 238000007781 pre-processing Methods 0.000 claims abstract description 11
- 230000000747 cardiac effect Effects 0.000 claims description 22
- 210000002569 neuron Anatomy 0.000 claims description 21
- 230000000946 synaptic effect Effects 0.000 claims description 20
- 239000011159 matrix material Substances 0.000 claims description 16
- 230000006870 function Effects 0.000 claims description 13
- 238000003909 pattern recognition Methods 0.000 claims description 13
- 210000000225 synapse Anatomy 0.000 claims description 11
- 230000008859 change Effects 0.000 claims description 10
- 230000002401 inhibitory effect Effects 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 10
- 230000001629 suppression Effects 0.000 claims description 9
- 238000009826 distribution Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 4
- 210000002364 input neuron Anatomy 0.000 claims description 4
- 230000006386 memory function Effects 0.000 claims description 4
- 238000004088 simulation Methods 0.000 claims description 4
- 230000006641 stabilisation Effects 0.000 claims description 4
- 238000011105 stabilization Methods 0.000 claims description 4
- 230000036962 time dependent Effects 0.000 claims description 4
- 230000005764 inhibitory process Effects 0.000 claims description 3
- 238000004422 calculation algorithm Methods 0.000 abstract description 6
- 238000001514 detection method Methods 0.000 abstract description 5
- 238000007635 classification algorithm Methods 0.000 abstract description 2
- 238000005265 energy consumption Methods 0.000 abstract description 2
- 230000007774 longterm Effects 0.000 description 4
- 208000024172 Cardiovascular disease Diseases 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000001052 transient effect Effects 0.000 description 3
- 206010003658 Atrial Fibrillation Diseases 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 206010047281 Ventricular arrhythmia Diseases 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000012634 fragment Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 238000001208 nuclear magnetic resonance pulse sequence Methods 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 230000001242 postsynaptic effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 210000004243 sweat Anatomy 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/318—Heart-related electrical modalities, e.g. electrocardiography [ECG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
- A61B5/024—Detecting, measuring or recording pulse rate or heart rate
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4806—Sleep evaluation
- A61B5/4809—Sleep detection, i.e. determining whether a subject is asleep or not
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/74—Details of notification to user or communication with user or patient ; user input means
- A61B5/746—Alarms related to a physiological condition, e.g. details of setting alarm thresholds or avoiding false alarms
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B2503/00—Evaluating a particular growth phase or type of persons or animals
- A61B2503/04—Babies, e.g. for SIDS detection
Abstract
The application discloses an anomaly early warning method and system for low measurement load electrocardiograph data, comprising the following steps: acquiring low-load electrocardiographic monitoring data, preprocessing, and extracting abnormal sign features based on an empirical mode decomposition method; constructing a heart rhythm abnormal track fingerprint according to the abnormal symptom characteristics, coding the heart rhythm abnormal track fingerprint by using a pulse neural network, training the pulse neural network, establishing a memristor model, and importing weight parameters after the training of the pulse neural network into the memristor model to generate a heart rhythm abnormal early warning model; and (5) utilizing the arrhythmia early warning model to realize mode classification through output current and judging whether the rhythm is normal or not. The method for estimating the abnormal symptoms from the electrocardiosignal residual errors of the adjacent time windows reduces algorithm complexity, improves abnormal symptom detection capability, realizes rhythm abnormality early warning mode classification based on the pulse neural network and the memristor, and reduces energy consumption of a wearable rhythm abnormality classification algorithm.
Description
Technical Field
The application relates to the technical field of electrocardiograph monitoring, in particular to an anomaly early warning method and system for electrocardiograph data with low measurement load.
Background
Long-term electrocardiographic monitoring is highly appreciated as an important means for timely early detection of cardiovascular disease, particularly in situations where transient changes in the relevant signals may indicate severe pathological conditions, such as Holter (dynamic electrocardiogram) monitoring. Clinical experience shows that the longer the electrocardio monitoring time is, the higher the detection rate and the accuracy rate of the electrocardio abnormal events such as arrhythmia, atrial fibrillation and the like are, but for transient and sporadic arrhythmia, the abnormal electrocardiographic fragments are difficult to capture by Holter in 24/48 hours because of short and infrequent attack time. However, under the monitoring load, most patients clinically have difficulty in matching with the completion of electrocardiographic monitoring for more than 48 hours, which presents challenges for early detection of cardiovascular disease. Wearable devices are considered to play an important role in monitoring and early warning because of their low measurement load with long-term monitoring capabilities.
In recent years, low-load electrocardiographic monitoring equipment represented by a wearable watch is highly concerned by long-term monitoring of patients in natural life scenes, and current researches indicate that the low-load electrocardiographic monitoring equipment can detect atrial fibrillation events. Although wearable health monitoring devices can achieve results related to clinical "gold standard" indicators, the accuracy of wearable health monitoring data for long-term continuous life monitoring, particularly in non-laboratory environments, still faces a number of challenges such as factors that interfere with the quality of the data, such as motion noise, sweat, temperature, skin impedance, etc. Therefore, how to fully dig out invisible and potential abnormal heart rhythm characteristics from low-load long-time electrocardiograph monitoring data such as a wearable watch and build an early abnormal heart rhythm early warning model is one of the problems which need to be solved effectively for serious cardiovascular diseases.
Disclosure of Invention
In order to solve the technical problems, the application provides an anomaly early warning method and system for low-measurement-load electrocardiographic data.
The first aspect of the application provides an anomaly early warning method for low measurement load electrocardiographic data, which comprises the following steps:
acquiring low-load electrocardio monitoring data, preprocessing the low-load electrocardio monitoring data, and extracting abnormal sign features based on an empirical mode decomposition method;
constructing a heart rhythm abnormal track fingerprint according to the abnormal symptom characteristics, and representing the mode of the heart rhythm abnormal overall dynamic characteristics through the heart rhythm abnormal track fingerprint;
encoding the fingerprint of the abnormal cardiac rhythm track by using a pulse neural network, training the pulse neural network to construct a mode identification strategy, establishing a memristor model, and importing weight parameters after training the pulse neural network into the memristor model to generate an abnormal cardiac rhythm early warning model;
and acquiring current low-load electrocardio monitoring data, importing a heart rhythm abnormality early warning model, realizing mode classification through output current, judging whether the heart rhythm is normal or not, and generating early warning information according to a judging result.
In the scheme, the method based on empirical mode decomposition extracts abnormal sign features, specifically:
acquiring low-load electrocardio monitoring data, preprocessing the low-load electrocardio monitoring data, windowing the preprocessed low-load electrocardio monitoring data, acquiring historical experience data according to a big data method, and performing simulation evaluation through the historical experience data to acquire window length;
carrying out stabilization treatment on nonstationary electrocardiograph monitoring data in a window, and carrying out experimental mode decomposition to obtain a preset number of eigenmode functions, so as to ensure that each eigenmode function component decomposed contains local characteristic information on different time scales of an original signal;
performing time-frequency conversion on each different mode to obtain a corresponding spectrogram;
based on the characteristic that the electrocardiosignal has rich frequency information in the normal-to-abnormal change process, a plurality of frequency points with the largest amplitude in a spectrogram corresponding to each eigenmode are selected to be used as suspected abnormal symptom characteristics, and an abnormal symptom characteristic plane matrix is formed.
In the scheme, an abnormal heart rhythm track fingerprint is constructed according to abnormal sign characteristics, and specifically comprises the following steps:
acquiring an abnormal symptom characteristic plane matrix, performing backward difference on the abnormal symptom characteristic plane matrix to obtain a heart rhythm abnormal symptom characteristic residual error characteristic graph, and acquiring suspected abnormal symptom characteristics under a window;
presetting a large-scale time window, combining all suspected abnormal symptom residual feature images into a 'track runway' body containing rhythm abnormal track features in a time direction under the large-scale time window, wherein each track in the 'track runway' body represents a frequency change track corresponding to a certain intrinsic mode in the large-scale time window, namely a rhythm abnormal residual track vector;
accumulating the frequency points corresponding to each eigenmode in the direction of a time axis to form a heart rhythm abnormal residual error track accumulated vector in the scale time window, and normalizing each heart rhythm abnormal residual error track accumulated vector to a gray level image value range;
presetting a gray threshold, performing binarization processing according to the gray threshold, and constructing a fingerprint containing arrhythmia residual error track information in a large-scale time window.
In the scheme, the pulse neural network is utilized to encode the fingerprint of the abnormal cardiac rhythm track, and the pulse neural network is trained to construct a pattern recognition strategy, which is specifically as follows:
acquiring a heart rhythm abnormal track fingerprint, wherein the heart rhythm abnormal track fingerprint is a set of different eigenvectors of abnormal sign characteristics in a large time window, each pixel value in the fingerprint represents the sum of the changes of the eigenvector abnormal tracks of the eigenvector in the large-scale time window, and different pixel distributions form different anomaly modes;
encoding the fingerprint of the abnormal cardiac rhythm track through a pulse neural network, wherein the pulse neural network consists of an input layer, a suppression layer and an output layer;
the number of neurons contained in the input layer is the matrix point number of the fingerprint of the abnormal cardiac rhythm track, the input neurons are used for coding pixels to a time window, and a coding strategy adopts pulse rate coding;
the output layer consists of a variable number of excited neurons, the mode classification task is executed, and the number of the inhibitory neurons of the inhibition layer is the same as that of the output layer;
the synaptic connection between the input and output layers in the model has a full connection with positive weight, and two unidirectional connections of forward synapses and backward synapses exist between the output layer and the inhibiting layer;
training the impulse neural network, wherein the learning rule adopts impulse time-dependent plasticity, and the weight of synapse changes correspondingly along with the time difference between the front impulse transmitted by the front neuron and the rear impulse transmitted by the rear neuron, so as to perform pattern recognition.
In the scheme, a memristor model is established, weight parameters after pulse neural network training is completed are imported into the memristor model, and a heart rhythm abnormality early warning model is generated, specifically:
establishing a memristor model with a memory function and a programmable function according to track fingerprint patterns under normal and abnormal evolution rules of the heart rhythm;
setting positive and negative weights in the memristor model, and generating a synaptic array by using the memristor synaptic module to construct a pulse neural network;
each pixel of the track fingerprint is converted into voltage input and is used as input into a pulse neural network, and the weight parameters of the output layer of the pulse neural network are trained offline based on the normal track fingerprint of the heart rhythm, so that the current output by the output layer can distinguish normal track fingerprint from abnormal track fingerprint;
in the training process, the word line and the bit line of the synaptic array are controlled through the switch control time sequence of the switch tube, the memristor model is cleared or weight is written, the weight parameters after the pulse neural network training is completed are programmed into the memristor model, and the arrhythmia early warning model is generated.
In the scheme, current low-load electrocardiograph monitoring data is acquired, a heart rhythm abnormality early warning model is imported, mode classification is realized through output current, whether the heart rhythm is normal is judged, early warning information is generated according to a judging result, and the method specifically comprises the following steps:
acquiring current low-load electrocardiograph monitoring data, constructing a heart rhythm abnormal track fingerprint, converting each pixel into a voltage signal with the same range, inputting the voltage signal into an input layer of a heart rhythm early-warning model, and carrying out mode discrimination through the magnitude of output current after passing through an output layer and a suppression layer to realize mode classification;
judging whether the heart rhythm corresponding to the current low-load electrocardio monitoring data is normal or not according to the mode classification result, generating early warning information if the heart rhythm is abnormal, and sending and displaying the early warning information according to a preset method.
The second aspect of the present application also provides an anomaly early warning system for low measurement load electrocardiographic data, the system comprising: the system comprises a memory and a processor, wherein the memory comprises an abnormality early warning method program for low-measurement-load electrocardiograph data, and the abnormality early warning method program for the low-measurement-load electrocardiograph data realizes the following steps when being executed by the processor:
acquiring low-load electrocardio monitoring data, preprocessing the low-load electrocardio monitoring data, and extracting abnormal sign features based on an empirical mode decomposition method;
constructing a heart rhythm abnormal track fingerprint according to the abnormal symptom characteristics, and representing the mode of the heart rhythm abnormal overall dynamic characteristics through the heart rhythm abnormal track fingerprint;
encoding the fingerprint of the abnormal cardiac rhythm track by using a pulse neural network, training the pulse neural network to construct a mode identification strategy, establishing a memristor model, and importing weight parameters after training the pulse neural network into the memristor model to generate an abnormal cardiac rhythm early warning model;
and acquiring current low-load electrocardio monitoring data, importing a heart rhythm abnormality early warning model, realizing mode classification through output current, judging whether the heart rhythm is normal or not, and generating early warning information according to a judging result.
The application solves the technical problems in the background technology and has the following beneficial effects:
because the acquisition load, the data characteristic and the algorithm model of the abnormal heart rhythm scene are changed under the low measurement load, the problem of high false alarm rate of the traditional signal processing method under the low measurement load scene is caused;
according to the mechanism that early electrocardio abnormal signals have abundant time-varying information to distinguish the early electrocardio abnormal signals from normal heart rhythm and the characteristic that abnormal symptoms are hidden in time and are not obvious in amplitude, the method is different from the traditional method for extracting abnormal features from original signals, and the method is provided for estimating the abnormal symptoms from adjacent time window electrocardio signal residual errors by considering that the abnormal time-varying features are easy to extract from the adjacent time window residual errors, so that algorithm complexity is reduced, and abnormal symptom detection capability is improved;
the application converts the arrhythmia pre-warning problem into the classification problem based on the arrhythmia track fingerprint, and provides a new means for realizing arrhythmia pre-warning mode classification based on a pulse neural network and memristor hardware, thereby comprehensively reducing the energy consumption of a wearable arrhythmia classification algorithm from two dimensions of an algorithm and a circuit.
Drawings
FIG. 1 shows a technical circuit diagram of an anomaly early warning method for low measurement load electrocardiographic data;
FIG. 2 shows a flow chart of an anomaly early warning method for low measurement load electrocardiographic data according to the present application;
FIG. 3 shows a flowchart of a method of constructing a trace fingerprint of an abnormal heart rhythm in accordance with the present application;
FIG. 4 is a flow chart of a method of constructing a heart rhythm abnormality pre-warning model in accordance with the present application;
fig. 5 shows a block diagram of an anomaly early warning system for low measurement load electrocardiographic data according to the present application.
Detailed Description
In order that the above-recited objects, features and advantages of the present application will be more clearly understood, a more particular description of the application will be rendered by reference to the appended drawings and appended detailed description. It should be noted that, without conflict, the embodiments of the present application and features in the embodiments may be combined with each other.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application, however, the present application may be practiced in other ways than those described herein, and therefore the scope of the present application is not limited to the specific embodiments disclosed below.
Fig. 1 and 2 show a technical roadmap and a flow chart of an abnormality early warning method for low measurement load electrocardiographic data.
As shown in fig. 1 and 2, the first aspect of the present application provides an anomaly early warning method for low measurement load electrocardiographic data, which includes:
s202, acquiring low-load electrocardiograph monitoring data, preprocessing the low-load electrocardiograph monitoring data, and extracting abnormal sign features based on an empirical mode decomposition method;
s204, constructing a heart rhythm abnormal track fingerprint according to the abnormal symptom characteristics, and representing the mode of the heart rhythm abnormal overall dynamic characteristics through the heart rhythm abnormal track fingerprint;
s206, coding a heart rhythm abnormality track fingerprint by using a pulse neural network, training the pulse neural network to construct a mode identification strategy, building a memristor model, and importing weight parameters after training of the pulse neural network into the memristor model to generate a heart rhythm abnormality early warning model;
s208, acquiring current low-load electrocardio monitoring data, importing a heart rhythm abnormality early warning model, realizing mode classification through output current, judging whether the heart rhythm is normal or not, and generating early warning information according to a judging result.
In the early stage of arrhythmia, the amplitude of the abnormal feature is small, and the abnormal feature is smoothed over time until the abnormal feature is often submerged in noise. The signals obtained by low-load electrocardiograph monitoring are often accompanied by a large amount of noise and interference, and the early onset duration of ventricular arrhythmia has a transient characteristic, so that the arrhythmia signals captured by the sensor are easily regarded as noise and missed, and the abnormal sign characteristics are extracted based on an empirical mode decomposition method, specifically comprising the following steps: acquiring low-load electrocardio monitoring data, preprocessing the low-load electrocardio monitoring data, windowing the preprocessed low-load electrocardio monitoring data, acquiring historical experience data according to a big data method, and performing simulation evaluation through the historical experience data to acquire window length; carrying out stabilization treatment on nonstationary electrocardiograph monitoring data in a window, and carrying out experimental mode decomposition to obtain a preset number of eigenmode functions, so as to ensure that each eigenmode function component decomposed contains local characteristic information on different time scales of an original signal; performing time-frequency conversion on each different mode to obtain a corresponding spectrogram; based on the characteristic that the electrocardiosignal has rich frequency information in the normal-to-abnormal change process, a plurality of frequency points with the largest amplitude in a spectrogram corresponding to each eigenmode are selected to be used as suspected abnormal symptom characteristics, and an abnormal symptom characteristic plane matrix (the eigenmodes with different behaviors are listed as main frequency values corresponding to each eigenmode) is formed.
FIG. 3 shows a flowchart of a method for constructing a trace fingerprint of an abnormal heart rhythm in the present application.
According to the embodiment of the application, the fingerprint of the arrhythmia track is constructed according to the characteristics of the abnormal symptoms, specifically:
s302, acquiring an abnormal symptom feature plane matrix, performing backward difference on the abnormal symptom feature plane matrix to obtain a heart rhythm abnormal symptom feature residual error feature map, and acquiring suspected abnormal symptom features under a window;
s304, presetting a large-scale time window, combining all suspected abnormal sign residual feature images into a 'track runway' body containing rhythm abnormal track features in a time direction under the large-scale time window, wherein each track in the 'track runway' body represents a frequency change track corresponding to a certain eigenvector in the large-scale time window, namely a rhythm abnormal residual track vector;
s306, accumulating the frequency points corresponding to each eigenmode in the direction of a time axis to form a heart rhythm abnormal residual error track accumulated vector in the scale time window, and normalizing each heart rhythm abnormal residual error track accumulated vector to a gray level image value range;
s308, presetting a gray threshold, performing binarization processing according to the gray threshold, and constructing a fingerprint containing arrhythmia residual error track information in a large-scale time window.
It should be noted that, the local feature of a small scale time window is still obtained by empirical mode decomposition, if the index is used as the basis of early warning, the false alarm rate is easily kept high. Therefore, the key to improving the early warning accuracy is to construct a mode with the overall dynamic characteristic of describing the abnormal heart rhythm on a larger scale time window. In the embodiment of the application, the thought of early warning is utilized by utilizing the overall dynamic change trend of the longitudinal time axis direction of an individual, and the early warning of the arrhythmia is realized by referring to the principle that a pilot depends on runway light to distinguish a runway when landing at night based on the time stealth of the early abnormal characteristics and the signals of the adjacent time windows have the characteristics of strong correlation. The characteristic residual error characteristic map of the abnormal heart rhythm sign is similar to scattered lighting of an airport runway, the length of the airport runway is similar to that of the airport runway under a preset large-scale time window, and each abnormal heart rhythm sign residual error track vector represents the variation caused by abnormal heart rhythm in the time window, so that the degree of abnormal heart rhythm in the time window can be estimated by using the index.
And converting the arrhythmia early warning problem into the pattern recognition problem of the abnormal fingerprint. The heart rhythm abnormal track fingerprint comprises a plurality of abnormal residual characteristic information in adjacent time windows in longitudinal time and spatial characteristic information of each eigenmode and corresponding frequency information in a spatial cross section, so the heart rhythm abnormal track fingerprint has very good characteristic robustness. The former makes full use of the association constraint characteristic of the time direction to improve the fault tolerance space of local abnormal characteristics, namely, although the abnormal characteristics extracted by a certain time window are wrong, the error can be eliminated due to the association constraint of the front and back time window information (when a pilot identifies a runway, the pilot does not depend on a certain lamp but a line formed by the whole lamp). The latter repeatedly digs the correlation of the characteristic space, namely the eigenvoice and the corresponding frequency change information under different scales, namely the error of the abnormal characteristic extracted from a certain space position can be eliminated due to the association constraint of the eigenvoice (when a pilot recognizes running, the interference of other lights of the runway can be discharged based on the prior knowledge that the runway lights have specific constraint in space).
It should be noted that, the fingerprint of the abnormal track of heart rhythm is a set of different eigenmodes of the abnormal feature in a larger scale time window, each "pixel value" of the fingerprint characterizes the sum of the changes of the abnormal track of the eigenmode in the scale time window, and different "pixel" distributions form different abnormal modes. Therefore, when in decision, the idea of big data relevance is combined, and a corresponding pattern recognition strategy is constructed according to the sustainability of abnormal features on a longitudinal time axis and the uniformity of spatial cross section distribution, so that early warning of arrhythmia is realized.
Encoding the fingerprint of the abnormal cardiac rhythm track by using a pulse neural network, and training the pulse neural network to construct a pattern recognition strategy, wherein the pattern recognition strategy specifically comprises the following steps: encoding the fingerprint of the abnormal cardiac rhythm track through a pulse neural network, wherein the pulse neural network consists of an input layer, a suppression layer and an output layer; the number of neurons contained in the input layer is the matrix point number of the fingerprint of the abnormal cardiac rhythm track, the input neurons are used for coding pixels to a time window, and a coding strategy adopts pulse rate coding; the output layer is composed of excited neurons with variable quantity, the mode classification task is executed, the quantity of the inhibitory neurons of the inhibitory layer is the same as that of the output layer, and the network utilizes the inhibitory neurons to promote the excited layer to realize the winner general eating algorithm. The synaptic connection between the input and output layers in the model has a full connection with positive weight, and two unidirectional connections of forward synapses and backward synapses exist between the output layer and the inhibiting layer; training the impulse neural network, wherein the learning rule adopts impulse time-dependent plasticity, and the weight of synapse changes correspondingly along with the time difference between the front impulse transmitted by the front neuron and the rear impulse transmitted by the rear neuron, so as to perform pattern recognition.
Fig. 4 shows a flowchart of a method of constructing a heart rhythm abnormality pre-warning model in accordance with the present application.
According to the embodiment of the application, a memristor model is established, and weight parameters after the training of the impulse neural network are introduced into the memristor model to generate a heart rhythm abnormality early warning model, which is specifically as follows:
s402, establishing a memristor model with a memory function and a programmable function according to track fingerprints under normal and abnormal evolution rules of the heart rhythm;
s404, a memristor synaptic module with positive and negative weights and capable of preventing leakage current is arranged in the memristor model, and a synaptic array is generated by utilizing the synaptic module, so that a pulse neural network is constructed;
s406, converting each pixel of the track fingerprint into voltage input, and inputting the voltage input into a pulse neural network, and performing off-line training on the weight parameters of an output layer of the pulse neural network based on the normal track fingerprint of the heart rhythm, so that the magnitude of current output by the output layer can distinguish normal track fingerprint from abnormal track fingerprint;
s408, in the training process, the word line and the bit line of the synaptic array are controlled through the switch control time sequence of the switch tube, the memristor model is cleared or weight is written, the weight parameters after the training of the impulse neural network are programmed to the memristor model, and the arrhythmia early warning model is generated.
It should be noted that, the memristor is used as a programmable low-power consumption memory element, and can simulate biological synapses. In the training process of the impulse neural network, the network weight needs to be updated continuously, and the memristor and the impulse neural network can be fused well, so that the energy efficiency of the system is optimized in two layers of an algorithm model and hardware. And converting the pulse sequence formed by each pixel of the fingerprint spectrum into post-synaptic current through the synaptic sensing of the brain computing circuit, and realizing mode classification by using a memristive neural network. The specific implementation process is as follows: each pixel '1' or '0' of the abnormal fingerprint is converted into a voltage signal with the same range, the voltage signal is input into an input layer of the pulse neural network, and a recognizable mode is output after passing through an output layer and a suppression layer of the pulse neural network, so that mode classification is realized, namely whether the heart rhythm of a detected person is normal or not is judged by judging the magnitude of output current of the neural network.
According to the embodiment of the application, the current low-load electrocardio monitoring data is acquired, a heart rhythm abnormality early warning model is imported, mode classification is realized through output current, whether the heart rhythm is normal is judged, early warning information is generated according to a judging result, and the method specifically comprises the following steps:
acquiring current low-load electrocardiograph monitoring data, constructing a heart rhythm abnormal track fingerprint, converting each pixel into a voltage signal with the same range, inputting the voltage signal into an input layer of a heart rhythm early-warning model, and carrying out mode discrimination through the magnitude of output current after passing through an output layer and a suppression layer to realize mode classification;
judging whether the heart rhythm corresponding to the current low-load electrocardio monitoring data is normal or not according to the mode classification result, generating early warning information if the heart rhythm is abnormal, and sending and displaying the early warning information according to a preset method.
Fig. 5 shows a block diagram of an anomaly early warning system for low measurement load electrocardiographic data according to the present application.
The second aspect of the present application also provides an anomaly early warning system 5 for low measurement load electrocardiographic data, the system comprising: the memory 51 and the processor 52, wherein the memory includes an abnormality early warning method program for low measurement load electrocardiographic data, and the abnormality early warning method program for low measurement load electrocardiographic data realizes the following steps when executed by the processor:
acquiring low-load electrocardio monitoring data, preprocessing the low-load electrocardio monitoring data, and extracting abnormal sign features based on an empirical mode decomposition method;
constructing a heart rhythm abnormal track fingerprint according to the abnormal symptom characteristics, and representing the mode of the heart rhythm abnormal overall dynamic characteristics through the heart rhythm abnormal track fingerprint;
encoding the fingerprint of the abnormal cardiac rhythm track by using a pulse neural network, training the pulse neural network to construct a mode identification strategy, establishing a memristor model, and importing weight parameters after training the pulse neural network into the memristor model to generate an abnormal cardiac rhythm early warning model;
and acquiring current low-load electrocardio monitoring data, importing a heart rhythm abnormality early warning model, realizing mode classification through output current, judging whether the heart rhythm is normal or not, and generating early warning information according to a judging result.
The third aspect of the present application also provides a computer readable storage medium, where the computer readable storage medium includes an anomaly early warning method program for low measurement load electrocardiographic data, where the anomaly early warning method program for low measurement load electrocardiographic data implements the steps of the anomaly early warning method for low measurement load electrocardiographic data according to any one of the above steps when the anomaly early warning method program for low measurement load electrocardiographic data is executed by a processor.
In the several embodiments provided by the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above described device embodiments are only illustrative, e.g. the division of the units is only one logical function division, and there may be other divisions in practice, such as: multiple units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. In addition, the various components shown or discussed may be coupled or directly coupled or communicatively coupled to each other via some interface, whether indirectly coupled or communicatively coupled to devices or units, whether electrically, mechanically, or otherwise.
The units described above as separate components may or may not be physically separate, and components shown as units may or may not be physical units; can be located in one place or distributed to a plurality of network units; some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in each embodiment of the present application may be integrated in one processing unit, or each unit may be separately used as one unit, or two or more units may be integrated in one unit; the integrated units may be implemented in hardware or in hardware plus software functional units.
Those of ordinary skill in the art will appreciate that: all or part of the steps for implementing the above method embodiments may be implemented by hardware related to program instructions, and the foregoing program may be stored in a computer readable storage medium, where the program, when executed, performs steps including the above method embodiments; and the aforementioned storage medium includes: a mobile storage device, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk or an optical disk, or the like, which can store program codes.
Alternatively, the above-described integrated units of the present application may be stored in a computer-readable storage medium if implemented in the form of software functional modules and sold or used as separate products. Based on such understanding, the technical solutions of the embodiments of the present application may be embodied in essence or a part contributing to the prior art in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: a removable storage device, ROM, RAM, magnetic or optical disk, or other medium capable of storing program code.
The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (2)
1. An anomaly early warning method for low measurement load electrocardiographic data is characterized by comprising the following steps:
acquiring low-load electrocardio monitoring data, preprocessing the low-load electrocardio monitoring data, and extracting abnormal sign features based on an empirical mode decomposition method;
constructing a heart rhythm abnormal track fingerprint according to the abnormal symptom characteristics, and representing the mode of the heart rhythm abnormal overall dynamic characteristics through the heart rhythm abnormal track fingerprint;
encoding the fingerprint of the abnormal cardiac rhythm track by using a pulse neural network, training the pulse neural network to construct a mode identification strategy, establishing a memristor model, and importing weight parameters after training the pulse neural network into the memristor model to generate an abnormal cardiac rhythm early warning model;
acquiring current low-load electrocardio monitoring data, importing a heart rhythm abnormality early warning model, realizing mode classification through output current, judging whether the heart rhythm is normal or not, and generating early warning information according to a judging result;
the method for extracting the abnormal sign features based on the empirical mode decomposition specifically comprises the following steps:
acquiring low-load electrocardio monitoring data, preprocessing the low-load electrocardio monitoring data, windowing the preprocessed low-load electrocardio monitoring data, acquiring historical experience data according to a big data method, and performing simulation evaluation through the historical experience data to acquire window length;
carrying out stabilization treatment on nonstationary electrocardiograph monitoring data in a window, and carrying out experimental mode decomposition to obtain a preset number of eigenmode functions, so as to ensure that each eigenmode function component decomposed contains local characteristic information on different time scales of an original signal;
performing time-frequency conversion on each different mode to obtain a corresponding spectrogram;
based on the characteristic that the electrocardiosignals have rich frequency information in the normal-to-abnormal change process, selecting a plurality of frequency points with the largest amplitude in a spectrogram corresponding to each eigenmode as suspected abnormal symptom characteristics to form an abnormal symptom characteristic plane matrix;
constructing a heart rhythm abnormal track fingerprint according to the abnormal symptom characteristics, wherein the fingerprint comprises the following specific steps:
acquiring an abnormal symptom characteristic plane matrix, performing backward difference on the abnormal symptom characteristic plane matrix to obtain a heart rhythm abnormal symptom characteristic residual error characteristic graph, and acquiring suspected abnormal symptom characteristics under a window;
presetting a large-scale time window, combining all suspected abnormal symptom residual feature images into a 'track runway' body containing rhythm abnormal track features in a time direction under the large-scale time window, wherein each track in the 'track runway' body represents a frequency change track corresponding to a certain intrinsic mode in the large-scale time window, namely a rhythm abnormal residual track vector;
accumulating the frequency points corresponding to each eigenmode in the direction of a time axis to form a heart rhythm abnormal residual error track accumulated vector in the scale time window, and normalizing each heart rhythm abnormal residual error track accumulated vector to a gray level image value range;
presetting a gray threshold, performing binarization processing according to the gray threshold, and constructing a fingerprint containing arrhythmia residual error track information in a large-scale time window;
encoding the fingerprint of the abnormal cardiac rhythm track by using a pulse neural network, and training the pulse neural network to construct a pattern recognition strategy, wherein the pattern recognition strategy specifically comprises the following steps:
acquiring a heart rhythm abnormal track fingerprint, wherein the heart rhythm abnormal track fingerprint is a set of different eigenmodes of abnormal sign characteristics in a large time window, each pixel value in the fingerprint represents the sum of the changes of the eigenmode abnormal tracks in the large-scale time window, and different pixel distribution forms different abnormal modes;
encoding the fingerprint of the abnormal cardiac rhythm track through a pulse neural network, wherein the pulse neural network consists of an input layer, a suppression layer and an output layer;
the number of neurons contained in the input layer is the matrix point number of the fingerprint of the abnormal cardiac rhythm track, the input neurons are used for coding pixels to a time window, and a coding strategy adopts pulse rate coding;
the output layer consists of a variable number of excited neurons, the mode classification task is executed, and the number of the inhibitory neurons of the inhibition layer is the same as that of the output layer;
the synaptic connection between the input and output layers in the model has a full connection with positive weight, and two unidirectional connections of forward synapses and backward synapses exist between the output layer and the inhibiting layer;
training the impulse neural network, wherein the learning rule adopts impulse time-dependent plasticity, and the synaptic weight correspondingly changes along with the time difference between the front impulse transmitted by the front neuron and the rear impulse transmitted by the rear neuron, so as to perform pattern recognition;
the method comprises the steps of establishing a memristor model, importing weight parameters after pulse neural network training is completed into the memristor model, and generating a heart rhythm abnormality early warning model, wherein the method specifically comprises the following steps:
establishing a memristor model with a memory function and a programmable function according to track fingerprint patterns under normal and abnormal evolution rules of the heart rhythm;
setting positive and negative weights in the memristor model, and generating a synaptic array by using the memristor synaptic module to construct a pulse neural network;
each pixel of the track fingerprint is converted into voltage input and is used as input into a pulse neural network, and the weight parameters of the output layer of the pulse neural network are trained offline based on the normal track fingerprint of the heart rhythm, so that the current output by the output layer can distinguish normal track fingerprint from abnormal track fingerprint;
in the training process, a word line and a bit line of a synaptic array are controlled through a switching tube switching control time sequence, a memristor model is cleared or weight is written, weight parameters after pulse neural network training is completed are programmed into the memristor model, and a heart rhythm abnormality early warning model is generated;
acquiring current low-load electrocardio monitoring data, importing a heart rhythm abnormality early warning model, realizing mode classification through output current, judging whether the heart rhythm is normal or not, and generating early warning information according to a judging result, wherein the method specifically comprises the following steps of:
acquiring current low-load electrocardiograph monitoring data, constructing a heart rhythm abnormal track fingerprint, converting each pixel into a voltage signal with the same range, inputting the voltage signal into an input layer of a heart rhythm early-warning model, and carrying out mode discrimination through the magnitude of output current after passing through an output layer and a suppression layer to realize mode classification;
judging whether the heart rhythm corresponding to the current low-load electrocardio monitoring data is normal or not according to the mode classification result, generating early warning information if the heart rhythm is abnormal, and sending and displaying the early warning information according to a preset method.
2. An anomaly early warning system for low measurement load electrocardiograph data is characterized in that the system comprises: the system comprises a memory and a processor, wherein the memory comprises an abnormality early warning method program for low-measurement-load electrocardiograph data, and the abnormality early warning method program for the low-measurement-load electrocardiograph data realizes the following steps when being executed by the processor:
acquiring low-load electrocardio monitoring data, preprocessing the low-load electrocardio monitoring data, and extracting abnormal sign features based on an empirical mode decomposition method;
constructing a heart rhythm abnormal track fingerprint according to the abnormal symptom characteristics, and representing the mode of the heart rhythm abnormal overall dynamic characteristics through the heart rhythm abnormal track fingerprint;
encoding the fingerprint of the abnormal cardiac rhythm track by using a pulse neural network, training the pulse neural network to construct a mode identification strategy, establishing a memristor model, and importing weight parameters after training the pulse neural network into the memristor model to generate an abnormal cardiac rhythm early warning model;
acquiring current low-load electrocardio monitoring data, importing a heart rhythm abnormality early warning model, realizing mode classification through output current, judging whether the heart rhythm is normal or not, and generating early warning information according to a judging result;
the method for extracting the abnormal sign features based on the empirical mode decomposition specifically comprises the following steps:
acquiring low-load electrocardio monitoring data, preprocessing the low-load electrocardio monitoring data, windowing the preprocessed low-load electrocardio monitoring data, acquiring historical experience data according to a big data method, and performing simulation evaluation through the historical experience data to acquire window length;
carrying out stabilization treatment on nonstationary electrocardiograph monitoring data in a window, and carrying out experimental mode decomposition to obtain a preset number of eigenmode functions, so as to ensure that each eigenmode function component decomposed contains local characteristic information on different time scales of an original signal;
performing time-frequency conversion on each different mode to obtain a corresponding spectrogram;
based on the characteristic that the electrocardiosignals have rich frequency information in the normal-to-abnormal change process, selecting a plurality of frequency points with the largest amplitude in a spectrogram corresponding to each eigenmode as suspected abnormal symptom characteristics to form an abnormal symptom characteristic plane matrix;
constructing a heart rhythm abnormal track fingerprint according to the abnormal symptom characteristics, wherein the fingerprint comprises the following specific steps:
acquiring an abnormal symptom characteristic plane matrix, performing backward difference on the abnormal symptom characteristic plane matrix to obtain a heart rhythm abnormal symptom characteristic residual error characteristic graph, and acquiring suspected abnormal symptom characteristics under a window;
presetting a large-scale time window, combining all suspected abnormal symptom residual feature images into a 'track runway' body containing rhythm abnormal track features in a time direction under the large-scale time window, wherein each track in the 'track runway' body represents a frequency change track corresponding to a certain intrinsic mode in the large-scale time window, namely a rhythm abnormal residual track vector;
accumulating the frequency points corresponding to each eigenmode in the direction of a time axis to form a heart rhythm abnormal residual error track accumulated vector in the scale time window, and normalizing each heart rhythm abnormal residual error track accumulated vector to a gray level image value range;
presetting a gray threshold, performing binarization processing according to the gray threshold, and constructing a fingerprint containing arrhythmia residual error track information in a large-scale time window;
encoding the fingerprint of the abnormal cardiac rhythm track by using a pulse neural network, and training the pulse neural network to construct a pattern recognition strategy, wherein the pattern recognition strategy specifically comprises the following steps:
acquiring a heart rhythm abnormal track fingerprint, wherein the heart rhythm abnormal track fingerprint is a set of different eigenmodes of abnormal sign characteristics in a large time window, each pixel value in the fingerprint represents the sum of the changes of the eigenmode abnormal tracks in the large-scale time window, and different pixel distribution forms different abnormal modes;
encoding the fingerprint of the abnormal cardiac rhythm track through a pulse neural network, wherein the pulse neural network consists of an input layer, a suppression layer and an output layer;
the number of neurons contained in the input layer is the matrix point number of the fingerprint of the abnormal cardiac rhythm track, the input neurons are used for coding pixels to a time window, and a coding strategy adopts pulse rate coding;
the output layer consists of a variable number of excited neurons, the mode classification task is executed, and the number of the inhibitory neurons of the inhibition layer is the same as that of the output layer;
the synaptic connection between the input and output layers in the model has a full connection with positive weight, and two unidirectional connections of forward synapses and backward synapses exist between the output layer and the inhibiting layer;
training the impulse neural network, wherein the learning rule adopts impulse time-dependent plasticity, and the synaptic weight correspondingly changes along with the time difference between the front impulse transmitted by the front neuron and the rear impulse transmitted by the rear neuron, so as to perform pattern recognition;
the method comprises the steps of establishing a memristor model, importing weight parameters after pulse neural network training is completed into the memristor model, and generating a heart rhythm abnormality early warning model, wherein the method specifically comprises the following steps:
establishing a memristor model with a memory function and a programmable function according to track fingerprint patterns under normal and abnormal evolution rules of the heart rhythm;
setting positive and negative weights in the memristor model, and generating a synaptic array by using the memristor synaptic module to construct a pulse neural network;
each pixel of the track fingerprint is converted into voltage input and is used as input into a pulse neural network, and the weight parameters of the output layer of the pulse neural network are trained offline based on the normal track fingerprint of the heart rhythm, so that the current output by the output layer can distinguish normal track fingerprint from abnormal track fingerprint;
in the training process, a word line and a bit line of a synaptic array are controlled through a switching tube switching control time sequence, a memristor model is cleared or weight is written, weight parameters after pulse neural network training is completed are programmed into the memristor model, and a heart rhythm abnormality early warning model is generated;
acquiring current low-load electrocardio monitoring data, importing a heart rhythm abnormality early warning model, realizing mode classification through output current, judging whether the heart rhythm is normal or not, and generating early warning information according to a judging result, wherein the method specifically comprises the following steps of:
acquiring current low-load electrocardiograph monitoring data, constructing a heart rhythm abnormal track fingerprint, converting each pixel into a voltage signal with the same range, inputting the voltage signal into an input layer of a heart rhythm early-warning model, and carrying out mode discrimination through the magnitude of output current after passing through an output layer and a suppression layer to realize mode classification;
judging whether the heart rhythm corresponding to the current low-load electrocardio monitoring data is normal or not according to the mode classification result, generating early warning information if the heart rhythm is abnormal, and sending and displaying the early warning information according to a preset method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310754754.5A CN116530996B (en) | 2023-06-26 | 2023-06-26 | Low-measurement-load-oriented electrocardiographic data abnormality early warning method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310754754.5A CN116530996B (en) | 2023-06-26 | 2023-06-26 | Low-measurement-load-oriented electrocardiographic data abnormality early warning method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116530996A CN116530996A (en) | 2023-08-04 |
| CN116530996B true CN116530996B (en) | 2023-08-29 |
Family
ID=87443875
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310754754.5A Active CN116530996B (en) | 2023-06-26 | 2023-06-26 | Low-measurement-load-oriented electrocardiographic data abnormality early warning method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116530996B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107516075A (en) * | 2017-08-03 | 2017-12-26 | 安徽华米信息科技有限公司 | Detection method, device and the electronic equipment of electrocardiosignal |
| WO2018217677A1 (en) * | 2017-05-22 | 2018-11-29 | Adaptive, Intelligent And Dynamic Brain Corporation (Aidbrain) | Method, module and system for analysis of brain electrical activity |
| KR20190052587A (en) * | 2017-11-08 | 2019-05-16 | 삼성전자주식회사 | Neural network device and operation method of the same |
| CN111281367A (en) * | 2018-12-10 | 2020-06-16 | 绍兴图聚光电科技有限公司 | Anti-interference non-contact heart rate detection method based on face video |
| EP3922177A1 (en) * | 2020-06-10 | 2021-12-15 | Biosense Webster (Israel) Ltd | System and method to detect stable arrhythmia heartbeat and to calculate and detect cardiac mapping annotations |
| CN114052744A (en) * | 2021-11-02 | 2022-02-18 | 西安电子科技大学重庆集成电路创新研究院 | Electrocardiosignal classification method based on pulse neural network |
| CN114732415A (en) * | 2022-03-31 | 2022-07-12 | 南京邮电大学 | Electrocardiosignal abnormity prediction method and device based on SNN network |
| CN116186581A (en) * | 2023-01-09 | 2023-05-30 | 重庆大学 | Floor identification method and system based on graph pulse neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3766055A1 (en) * | 2018-03-14 | 2021-01-20 | Chronolife | System and method for processing multiple signals |
-
2023
- 2023-06-26 CN CN202310754754.5A patent/CN116530996B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018217677A1 (en) * | 2017-05-22 | 2018-11-29 | Adaptive, Intelligent And Dynamic Brain Corporation (Aidbrain) | Method, module and system for analysis of brain electrical activity |
| CN107516075A (en) * | 2017-08-03 | 2017-12-26 | 安徽华米信息科技有限公司 | Detection method, device and the electronic equipment of electrocardiosignal |
| KR20190052587A (en) * | 2017-11-08 | 2019-05-16 | 삼성전자주식회사 | Neural network device and operation method of the same |
| CN111281367A (en) * | 2018-12-10 | 2020-06-16 | 绍兴图聚光电科技有限公司 | Anti-interference non-contact heart rate detection method based on face video |
| EP3922177A1 (en) * | 2020-06-10 | 2021-12-15 | Biosense Webster (Israel) Ltd | System and method to detect stable arrhythmia heartbeat and to calculate and detect cardiac mapping annotations |
| CN114052744A (en) * | 2021-11-02 | 2022-02-18 | 西安电子科技大学重庆集成电路创新研究院 | Electrocardiosignal classification method based on pulse neural network |
| CN114732415A (en) * | 2022-03-31 | 2022-07-12 | 南京邮电大学 | Electrocardiosignal abnormity prediction method and device based on SNN network |
| CN116186581A (en) * | 2023-01-09 | 2023-05-30 | 重庆大学 | Floor identification method and system based on graph pulse neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116530996A (en) | 2023-08-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111160139B (en) | Electrocardiosignal processing method and device and terminal equipment | |
| CN110133610B (en) | Ultra-wideband radar action identification method based on time-varying distance-Doppler diagram | |
| WO2022257329A1 (en) | Brain machine interface decoding method based on spiking neural network | |
| CN111772619B (en) | Heart beat identification method based on deep learning, terminal equipment and storage medium | |
| CN107832737A (en) | Electrocardiogram interference identification method based on artificial intelligence | |
| Guo et al. | A cross-correlated delay shift supervised learning method for spiking neurons with application to interictal spike detection in epilepsy | |
| CN111329469A (en) | Arrhythmia prediction method | |
| WO2019216378A1 (en) | Calculation device, detection device, calculation method, and computer program | |
| Li et al. | Patient-specific seizure prediction from electroencephalogram signal via multi-channel feedback capsule network | |
| CN110491506A (en) | Auricular fibrillation prediction model and its forecasting system | |
| CN110522412A (en) | Method based on multiple dimensioned brain function network class EEG signals | |
| CN113349793A (en) | Arrhythmia classification method based on layered attention model | |
| CN113180679A (en) | Single-lead electrocardio abnormal signal identification method | |
| Buettner et al. | Heartbeat classification with spiking neural networks on the loihi neuromorphic processor | |
| EP4041073A1 (en) | Systems and methods for electrocardiogram diagnosis using deep neural networks and rule-based systems | |
| CN113995417A (en) | Electrocardiosignal abnormity prediction method and system based on LSTM self-encoder | |
| CN116977330B (en) | Atrial fibrillation auxiliary analysis method based on pulse neural network and context awareness | |
| CN116530996B (en) | Low-measurement-load-oriented electrocardiographic data abnormality early warning method and system | |
| Wang | ReNN: Rule-embedded neural networks | |
| CN116864140A (en) | Intracardiac branch of academic or vocational study postoperative care monitoring data processing method and system thereof | |
| Song et al. | Automatic identification of atrial fibrillation based on the modified Elman neural network with exponential moving average algorithm | |
| CN116504398A (en) | Methods and systems for arrhythmia prediction using a transducer-based neural network | |
| CN114732415A (en) | Electrocardiosignal abnormity prediction method and device based on SNN network | |
| CN112022149B (en) | Atrial fibrillation detection method based on electrocardiosignals | |
| Liu et al. | Automatic focal EEG identification based on deep reinforcement learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |