A method and system for intelligently identifying intestinal movement based on BRS acceleration

Abstract

The present application discloses a method and system for intelligently identifying intestinal movements based on BRS acceleration, which relates to the field of intestinal movement information processing, including: obtaining abdominal feature data of a target patient in a target time period to obtain a target observation signal; the abdominal feature data includes abdominal wall vibration data collected by a BRS acceleration sensor; performing a blind source separation operation on the target observation signal to obtain a target source signal; identifying the vibration source category corresponding to each target source signal to screen the target intestinal component; extracting the feature vector of the target intestinal component to obtain a target feature vector; matching the target feature vector with a preset normal value, and generating a first feedback signal when the target feature vector does not meet the preset normal value, thereby realizing non-invasive detection of the patient's intestinal movement, which can be used for long-term continuous monitoring of the patient's intestinal movement.

Description

Technical Field

[0001] The present application relates to the field of intestinal movement information processing, and specifically to a method and system for intelligently identifying intestinal movements based on BRS acceleration. Background Art

[0002] Continuous monitoring of intestinal motility has important clinical significance in evaluating intestinal dysfunction, monitoring treatment effects, and guiding enteral nutrition. How to obtain intestinal motility data and how to identify abnormal movement events from intestinal motility data are the keys to achieving continuous monitoring of intestinal motility.

[0003] Commonly used methods for identifying intestinal movement include imaging examination methods, radionuclide scanning, endoscopy, and other identification technologies such as capsule endoscopy. However, it is difficult for the above methods to continuously monitor the intestinal movement of patients for a long time. For example, during capsule endoscopy, the capsule continuously obtains intestinal movement characteristic data at different locations as the digestive tract moves. However, the capsule endoscopy detection method has the risk of capsule retention and secondary intestinal obstruction, which makes it difficult to continuously monitor the intestinal movement of patients through this method.

[0004] At present, there is a MEMS technology used in the prior art to detect the abdominal wall vibration of the patient. For example, the MEMS technology-based abdominal wall monitoring system with patent application number CN201210217518.1 converts the abdominal wall vibration data into an electrical signal through a micro-vibration sensor such as a piezoelectric film vibration sensor, uses a bias circuit, an LED device and a photodetector to complete the process of converting the electrical signal into an optical signal and from the optical signal into an electrical signal, and then uses a power amplifier to amplify the processed electrical signal to obtain the abdominal wall vibration data that can be used for display, thereby realizing long-term continuous monitoring of the patient's abdominal wall vibration. However, the abdominal wall vibration data obtained by this method includes not only the abdominal wall vibration caused by intestinal movement, but also other vibration sources such as breathing and heartbeat that cause abdominal wall vibration, which leads to the difficulty of identifying whether the intestinal movement belongs to an abnormal movement event according to the abdominal wall vibration data collected by this method, and the accuracy is low. Therefore, there is an urgent need for a method for processing and analyzing the electrical signal data of the patient's abdominal wall vibration, so as to identify and separate the abdominal wall vibration caused by intestinal movement from the patient's abdominal wall vibration electrical signal data, and judge whether the patient's intestinal movement has an abnormal movement event according to the characteristics of the electrical signal data corresponding to the intestinal movement. Summary of the invention

[0005] The purpose of the present invention is to solve the technical problem of how to capture abnormal intestinal movement events from the vibration characteristics of the patient's abdominal wall, and to provide a method and system for intelligent recognition of intestinal movement based on BRS acceleration.

[0006] According to a first aspect of the present invention, the present invention claims a method for intelligently identifying intestinal movements based on BRS acceleration, comprising:

[0007] Obtain abdominal characteristic data of a target patient in a target time period to obtain a target observation signal;

[0008] Wherein, the abdominal characteristic data includes abdominal wall vibration data of the target patient collected by a BRS acceleration sensor;

[0009] Performing a blind source separation operation on the target observation signal to obtain a target source signal;

[0010] Identifying the vibration source category corresponding to each of the target source signals to screen the target intestinal component;

[0011] Wherein, the vibration source categories include heartbeat, breathing and intestines;

[0012] Extracting the characteristic vector of the target intestinal component to obtain a target characteristic vector;

[0013] The target feature vector is matched with a preset normal value, and when the target feature vector does not conform to the preset normal value, a first feedback signal is generated, wherein the first feedback signal is used to indicate that the target observation signal belongs to an abnormal motion event.

[0014] In one embodiment of the present application, the number of each of the abdominal characteristic data collection devices is not less than 3, and after generating the first feedback signal, the method further includes:

[0015] The direction of the target intestinal segment is estimated by using beam forming to obtain the target direction;

[0016] Wherein, the target intestinal segment is the intestinal segment that generates the target observation signal;

[0017] Determine the distance from each target observation signal acquisition device to the target intestinal segment by triangulation to obtain the target distance;

[0018] The position of the corresponding target intestinal segment is obtained according to all the target directions and all the target distances.

[0019] In an embodiment of the present application, the step of identifying the vibration source category corresponding to each of the target source signals further includes:

[0020] Obtaining an initial value of a candidate category probability set according to the probability that the vibration source of each of the target source signals belongs to all of the vibration source categories;

[0021] Taking the maximum value of the candidate category probability from the candidate category probability set as the target category probability;

[0022] Setting the vibration source category corresponding to the target category probability as the vibration source of the target source signal corresponding to the target category probability;

[0023] Remove the candidate category probability that is the same as the vibration source category of the target category probability and the candidate category probability that is the same as the target source signal of the target category probability from the candidate category probability set to obtain an updated candidate category probability set;

[0024] Determine whether the element in the updated candidate category probability set is 0. If not, go to the step of taking the maximum value of the candidate category probability from the candidate category probability set. If so, vibration source identification is completed.

[0025] In one embodiment of the present application, before determining whether the element in the updated candidate category probability set is 0, after setting the vibration source category corresponding to the target category probability to the vibration source of the target source signal corresponding to the target category probability, it also includes: determining whether the vibration source category corresponding to the target category probability is intestines; if not, continuing to execute the step of obtaining the updated candidate category probability set; if so, vibration source identification is completed.

[0026] In an embodiment of the present application, the vibration source further includes abdominal muscles, and the probability that the vibration source according to each of the target source signals belongs to all the vibration source categories further includes:

[0027] Obtaining the probability that the vibration source of each target source signal belongs to a heartbeat;

[0028] Obtaining the probability that the vibration source of each target source signal belongs to breathing;

[0029] Obtaining the probability that the vibration source of each target source signal belongs to the intestine;

[0030] Obtaining the probability that the vibration source of each target source signal belongs to the abdominal wall muscle;

[0031] Among them, the probability that the vibration source of each target source signal belongs to the abdominal wall muscle is also included:

[0032] Matching each of the target source signals with the abdominal wall vibration data corresponding to the preset abdominal muscle state, and combining the successfully matched muscle states into corresponding candidate motion sequences in chronological order;

[0033] Acquire the motion video of the target patient in the target time period to obtain a target motion video;

[0034] identifying a target motion of the target patient in the target motion video;

[0035] Get the setting position of the BRS acceleration sensor and obtain the target detection position;

[0036] Acquire the abdominal muscle state corresponding to the target detection position and matching each of the target actions, and combine the abdominal muscle states corresponding to all the target actions into a target motion sequence in chronological order;

[0037] Calculating the similarity between the target motion sequence and each of the candidate motion sequences respectively to obtain motion similarity;

[0038] The probability that the vibration source corresponding to the target source signal belongs to the abdominal muscles is obtained according to the motion similarity.

[0039] In an embodiment of the present application, the respectively calculating the similarity between the target motion sequence and each of the candidate motion sequences further includes:

[0040] Respectively calculating the similarity between the muscle state type of the target motion sequence and the muscle state type of each candidate motion sequence to obtain type similarity;

[0041] Respectively calculating the similarity between the muscle state amplitude of the target motion sequence and the muscle state amplitude of each candidate motion sequence to obtain amplitude similarity;

[0042] A linear weighted result of the type similarity and the amplitude similarity is calculated to obtain the motion similarity.

[0043] In an embodiment of the present application, the weight of the type similarity is greater than the weight of the amplitude similarity.

[0044] In an embodiment of the present application, the method for calculating the candidate category probability set further includes:

[0045] Respectively obtaining the probability that the vibration sources of all the target source signals belong to the same vibration source category to obtain a signal probability vector;

[0046] Each of the signal probability vectors is normalized, and the candidate category probability set is obtained according to the normalization result.

[0047] In one embodiment of the present application, the abdominal characteristic data also includes bowel sounds.

[0048] According to a second aspect of the present invention, the present invention claims protection for an intelligent intestinal movement recognition system based on BRS acceleration, comprising:

[0049] A data acquisition module, the data acquisition module is used to acquire abdominal characteristic data of a target patient in a target time period to obtain a target observation signal; wherein the abdominal characteristic data includes abdominal wall vibration data of the target patient collected by a BRS acceleration sensor;

[0050] A blind source separation module, wherein the blind source separation module is used to perform a blind source separation operation on the target observation signal to obtain a target source signal;

[0051] A source identification module, the source identification module is used to identify the vibration source category corresponding to each of the target source signals to screen the target intestinal components; wherein the vibration source categories include heartbeat, breathing and intestinal tract;

[0052] A feature extraction module, wherein the feature extraction module is used to extract a feature vector of the target intestinal component to obtain a target feature vector;

[0053] An abnormality judgment module, the abnormality judgment module is used to match the target feature vector with a preset normal value, and when the target feature vector does not meet the preset normal value, a first feedback signal is generated, and the first feedback signal is used to indicate that the target observation signal belongs to an abnormal motion event.

[0054] In one embodiment of the present application, the number of each abdominal feature data collection device is not less than 3, and the system also includes a source positioning module, which is used to use beamforming to estimate the direction of the target intestinal segment to obtain the target direction; wherein the target intestinal segment is the intestinal segment that generates the target observation signal; the distance from each target observation signal collection device to the target intestinal segment is determined by a triangulation positioning method to obtain the target distance; and the position of the corresponding target intestinal segment is obtained according to all the target directions and all the target distances.

[0055] In one embodiment of the present application, the source identification module further includes:

[0056] A category probability calculation module, the category probability calculation module is used to obtain an initial value of a candidate category probability set according to the probability that the vibration source of each of the target source signals belongs to all the vibration source categories;

[0057] A vibration category identification module, the vibration category identification module is used to take the maximum value of the candidate category probability from the candidate category probability set as the target category probability; set the vibration source category corresponding to the target category probability to the vibration source of the target source signal corresponding to the target category probability; remove the candidate category probabilities that are the same as the vibration source category of the target category probability and the candidate category probabilities that are the same as the target source signal of the target category probability from the candidate category probability set to obtain the updated candidate category probability set; determine whether the elements in the updated candidate category probability set are 0, if not, go to the step of taking the maximum value of the candidate category probability from the candidate category probability set, if so, vibration source identification is completed.

[0058] In one embodiment of the present application, the vibration category identification module also includes determining whether the vibration source category corresponding to the target category probability is the intestine. If not, continue to execute the step of obtaining the updated candidate category probability set. If so, the vibration source identification is completed.

[0059] In one embodiment of the present application, the vibration source further includes abdominal muscles, and the category probability calculation module further includes:

[0060] A heartbeat probability calculation module, the heartbeat probability calculation module is used to obtain the probability that the vibration source of each target source signal belongs to a heartbeat;

[0061] A breathing probability calculation module, the breathing probability calculation module is used to obtain the probability that the vibration source of each target source signal belongs to breathing;

[0062] An intestinal probability calculation module, the intestinal probability calculation module is used to obtain the probability that the vibration source of each target source signal belongs to the intestinal tract;

[0063] A muscle probability calculation module, the muscle probability calculation module is used to obtain the probability that the vibration source of each target source signal belongs to the abdominal wall muscle;

[0064] Wherein, the muscle probability calculation module also includes:

[0065] A signal conversion module, the signal conversion module is used to match each of the target source signals with the abdominal wall vibration data corresponding to the preset abdominal muscle state, and combine the successfully matched muscle states into a corresponding candidate motion sequence in chronological order;

[0066] An image acquisition module, the image acquisition module is used to acquire a motion video of the target patient in the target time period to obtain a target motion video;

[0067] An action recognition module, the action recognition module is used to recognize the target action of the target patient in the target motion video;

[0068] A detection position acquisition module, wherein the detection position acquisition module is used to acquire the setting position of the BRS acceleration sensor to obtain the target detection position;

[0069] An action conversion module, the action conversion module is used to obtain the abdominal muscle state corresponding to the target detection position and matching each target action, and combine the abdominal muscle states corresponding to all the target actions into a target motion sequence in chronological order;

[0070] A probability calculation module is used to calculate the similarity between the target motion sequence and each of the candidate motion sequences respectively to obtain motion similarity; and obtain the probability that the vibration source corresponding to the target source signal belongs to the abdominal muscles according to the motion similarity.

[0071] In one embodiment of the present application, the probability calculation module further comprises: respectively calculating the similarity between the muscle state type of the target motion sequence and the muscle state type of each candidate motion sequence to obtain type similarity;

[0072] The similarity between the muscle state amplitude of the target motion sequence and the muscle state amplitude of each candidate motion sequence is calculated respectively to obtain amplitude similarity.

[0073] In an embodiment of the present application, the weight of the type similarity is greater than the weight of the amplitude similarity.

[0074] In one embodiment of the present application, the category probability calculation module also includes a normalization processing module, which is used to respectively obtain the probability that the vibration sources of all the target source signals belong to the same vibration source category to obtain a signal probability vector; normalize each of the signal probability vectors, and obtain the candidate category probability set based on the normalization processing result.

[0075] In one embodiment of the present application, the abdominal characteristic data also includes bowel sounds.

[0076] This application has the following beneficial effects:

[0077] 1. Use the BRS accelerometer to collect the patient's abdominal wall vibration data, and remove the influence of heartbeat, breathing, etc. on intestinal movement through blind source separation and vibration source identification of the separated source signal. Extract the abdominal wall vibration data caused by intestinal movement from the mixed signal to reduce the influence of other vibration sources on the identification of intestinal movement, improve the identification accuracy of intestinal movement, and then realize non-invasive detection of patients' intestinal movement, which can be used for long-term continuous monitoring of patients' intestinal movement.

[0078] 2. BRS accelerometer has the advantages of high sensitivity, wide frequency response, anti-interference and low power consumption. Thanks to the high electromechanical coupling coefficient of piezoelectric materials and the high Q value of thin film structures, BRS accelerometer can achieve acceleration detection in the μg level, and its resonant frequency can be designed to hundreds of kHz, covering a wide frequency band of DC-10kHz. The frequency output is insensitive to electromagnetic interference, bias drift, etc. The resonator power consumption can be as low as μW level to improve the collection quality and efficiency of the patient's abdominal wall vibration data.

[0079] 3. When the target observation signal belongs to an abnormal motion event, the position of the intestinal segment that generates the target observation signal is determined through beamforming technology and triangulation positioning method, so as to locate the location of the abnormal motion event.

[0080] 4. The maximum value of the element in the candidate category probability set indicates, on the one hand, that the probability of the vibration source of the target source signal belonging to the vibration source category is the highest among all vibration source categories, and on the other hand, that the probability of the target source signal belonging to the vibration source category is the highest among all target source signals belonging to the vibration source category. In order of category probability from large to small, the corresponding vibration source category is set for each of the target source signals in turn, taking into account both the probability ranking of the same target source signal belonging to different vibration source categories and the probability ranking of different target source signals belonging to the same vibration source category, thus achieving multi-dimensional probability comparison and improving the recognition accuracy of abdominal wall vibration caused by intestinal movement.

[0081] 5. After each vibration source category is identified, it is prioritized to determine whether the vibration source category currently identified is the intestine. If so, the vibration source category identification process is stopped to reduce the amount of calculation for screening the target source signal corresponding to the intestine and improve the recognition efficiency.

[0082] 6. Assume that each target source signal is the abdominal wall vibration data generated by the change of the abdominal muscle state, and match it with the abdominal wall vibration data corresponding to the preset abdominal muscle state to obtain the candidate motion sequence corresponding to each target source signal. Match the patient's movements during the test according to the abdominal muscle states corresponding to all target movements, so as to generate the patient's target motion sequence, and obtain the probability that the vibration source of the target source signal corresponding to the candidate motion sequence belongs to the abdominal muscles according to the similarity between each candidate motion sequence and the target motion sequence, so as to reduce the impact of the patient's movement on the intestinal movement recognition, improve the intestinal movement recognition accuracy due to the patient's movement during the test, and expand the application scenarios of long-term monitoring of the patient's intestinal movement.

[0083] 7. The differences in body size, abdominal fat thickness, and abdominal muscle size among different patients result in the same abdominal muscle state type when completing the same target action, but the amplitude of abdominal wall vibration is different. When calculating each reference motion sequence and target motion sequence, the similarity between the reference motion sequence and the target motion sequence corresponding to the target source signal generated by the real abdominal muscle will be reduced. Setting the weight value of amplitude similarity to be smaller than the weight value of type similarity is conducive to reducing the adverse effects of individual differences among patients on screening intestinal components from all target source signals and improving the recognition accuracy of intestinal motion.

[0084] 8. When calculating the probability that the target source signal belongs to different vibration source types, different evaluation methods can be used to evaluate the probability that the target source signal belongs to each vibration source type. Each signal probability vector is processed by normalization to reduce the adverse effects of different probability evaluation methods on vibration source identification, eliminate the dimension, and reduce the difficulty of model design for calculating different vibration source types of the target source signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0086] Figure 1 This is a flow chart of a method for intelligently identifying intestinal movements based on BRS acceleration involved in an embodiment of the present application;

[0087] Figure 2 A flow chart of vibration source identification according to an embodiment of the present application;

[0088] Figure 3 This is a schematic diagram of a calculation process in which the vibration source of the target source signal involved in the embodiment of the present application belongs to the abdominal muscles;

[0089] Figure 4 This is a schematic diagram of the structure of an intelligent intestinal movement recognition system based on BRS acceleration involved in an embodiment of the present application.

[0090] Figure 5 This is a schematic diagram of the structure of an electronic device involved in an embodiment of the present application. DETAILED DESCRIPTION

[0091] The present invention provides a method and system for intelligent identification of intestinal movement based on BRS acceleration. In order to make the above-mentioned purposes, features and advantages of the present application more obvious and understandable, the technical solutions in the embodiments of the present application will be clearly and completely described in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, with reference to the terms "one embodiment", "some embodiments", "implementation", "embodiment", "illustrative embodiment", "example", "specific example" or "some examples", the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the present application for protection, but only indicates that the specific features, structures or characteristics described in combination with the embodiment or example are included in at least one embodiment or example of the present invention. Moreover, the specific features, structures or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0092] It should be noted that similar reference numerals and letters represent similar items in the following figures, so once an item is defined in one figure, it does not need to be further defined and explained in the subsequent figures. At the same time, in the description of this application, the terms "first", "second" and other relational terms are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0093] According to the first aspect of the present invention, the present invention claims a method for intelligently identifying intestinal movements based on BRS acceleration, referring to the attached Figure 1 As shown, including:

[0094] S1: Obtain abdominal feature data of a target patient in a target time period to obtain a target observation signal.

[0095] The abdominal feature data includes a first abdominal sub-feature and a second abdominal sub-feature, but it is not limited thereto. The first abdominal sub-feature is the abdominal wall vibration data of the target patient, and the second abdominal sub-feature is the bowel sound of the target patient.

[0096] It should be noted that the Bulk Acoustic Wave Resonator Accelerometer Sensor (hereinafter referred to as BRS accelerometer) uses piezoelectric and capacitive principles to convert the mechanical vibration of the patient's abdominal surface into electrical signals. BRS accelerometers have the advantages of high sensitivity, wide frequency response, anti-interference and low power consumption. Thanks to the high electromechanical coupling coefficient of piezoelectric materials and the high Q value of thin film structures, BRS accelerometers can achieve acceleration detection in the μg range, and its resonant frequency can be designed to hundreds of kHz, covering a wide frequency band of DC-10kHz. The frequency output is insensitive to electromagnetic interference, bias drift, etc. The resonator power consumption can be as low as μW to improve the quality and efficiency of collecting abdominal wall vibration data of patients.

[0097] In this embodiment, before the detection begins, at least three BRS accelerometers are set at different positions on the abdomen of the target patient, and the sensitivity of the BRS accelerometer is not less than 10mV / g, the frequency response range is 0.01-20 Hz, and the noise is not more than 10 μg / √Hz. The BRS accelerometer can be fixed to the abdomen of the target patient by means of adhesive and / or strapping to obtain the abdominal wall vibration caused by intestinal peristalsis of the patient. At the same time, at least three silicon-based micro-microphones are set on the abdomen of the target patient. The sensitivity of the silicon-based micro-microphone is not less than 10mV / Pa, the frequency response range is 20-2000 Hz, and the noise is not more than 30 dBA. It can be fixed by adhesive and / or strapping to obtain the patient's bowel sounds. After setting up the BRS accelerometer array and the silicon-based micro-microphone array, the power is turned on, and the abdominal wall vibration data and bowel sounds of the target patient are collected at the same time to obtain the observation signal.

[0098] S2: Preprocess the target observation signal to improve the quality of the target observation signal, reduce noise, and enhance signal characteristics.

[0099] It should be noted that the preprocessing includes bandpass filtering, removing the mean to eliminate baseline drift, removing the power frequency to eliminate 50 / 60Hz interference and signal enhancement, which is of course not limited to this. Among them, the bandpass range of the bandpass filter is preset according to the frequencies of all preset vibration sources. The frequencies of the preset vibration sources include heart rate, breathing rate and intestinal movement frequency, etc., which is of course not limited to this.

[0100] In this embodiment, step S2 further includes the following sub-steps: S21, performing blind source separation operation on the target observation signal to obtain a target source signal. S22, identifying the vibration source category corresponding to each target source signal to screen the target intestinal component.

[0101] It should be noted that the vibration source categories include heartbeat, breathing and intestines, but are certainly not limited thereto.

[0102] In this embodiment, blind source separation is performed on the first abdominal sub-feature to obtain abdominal vibration data caused by different vibration sources, that is, to obtain the corresponding first target source signal. It is determined whether the vibration source of each of the first target source signals is the intestine, and the first target source signal corresponding to the intestinal movement is screened out according to the judgment result. The target intestinal component can be obtained based on the first target source signal corresponding to the intestinal movement. Blind source separation is performed on the second abdominal sub-feature to obtain bowel sounds and sound signals generated by other vibration sources, that is, to obtain the corresponding second target source signal. It is determined whether the vibration source of each of the second target source signals is the intestine, and the second target source signal corresponding to the intestinal movement is screened out according to the judgment result. The intestinal movement component can also be obtained based on the first source signal and the second source signal corresponding to the intestine.

[0103] It should be noted that, during the detection process, if the first abdominal sub-feature of the target patient is acquired through only one BRS acceleration sensor, the vibration data generated by different vibration sources in the first abdominal sub-feature can be separated by a single-channel blind source separation method to obtain the corresponding first target source signal, such as the SSA-ICA algorithm. If the first abdominal sub-feature of the target patient is acquired through multiple BRS acceleration sensors, the vibration data generated by different vibration sources in the first abdominal sub-feature can be separated by a multi-channel blind source separation method, such as the FastICA algorithm and the SOBI algorithm. The blind source separation of the abdominal sub-feature is similar to the above method and will not be described in detail here.

[0104] In this embodiment, taking the implementation of multi-channel blind source separation by FastICA algorithm as an example, step S21 also includes the following sub-steps: S211, after completing the preprocessing, the target observation signals collected by all BRS acceleration sensors are centralized. Calculate the mean of each target observation signal and centralize all the target observation signals. S212, whiten all the target observation signals after centering so that the dimensions of the observation data matrix composed of the target observation signals after whitening are independent of each other and have unit variance. This can be achieved by performing eigenvalue decomposition or singular value decomposition on the covariance matrix, but it is certainly not limited to this. S213, initialize the weight matrix. A unit matrix or a random matrix can be selected as the initial value of the weight matrix, but it is certainly not limited to this. S214, in units of rows, perform a linear transformation on the observation data matrix composed of the target observation signals after whitening according to the weight vector corresponding to each row in the weight matrix. Calculate the product of the transpose of the weight vector and the observation data matrix. S215, perform a nonlinear transformation on the observation data matrix after linear transformation. You can choose nonlinear functions such as tanh and Gaussian for processing. S216. Update the weights. S217. Normalize the updated weight vector to ensure that its length is 1. S218. Orthogonalize each weight vector to ensure that the new independent components are orthogonal to the previously obtained independent components. You can choose the Gram-Schmidt orthogonal method, or other orthogonal methods. S219. Determine whether the 1-norm of the change between the updated weight vector and the weight vector before the update is less than the first preset threshold. If the 1-norm of the change is less than the first preset threshold, the separation is completed; if the 1-norm of the change is not less than the first preset threshold, return to step S214 to continue execution.

[0105] It should be noted that, in step S216, the weight vector updating method includes:

[0106] ;

[0107] ;

[0108] Wherein, ω1 represents the weight vector before updating, ω2 represents the weight vector after updating, E represents the desired function to be solved, z represents the observation data matrix composed of all the target observation signals after whitening processing, g represents the nonlinear function, Represents the derivative of the g function, a1 is a constant, and the value of a1 can be chosen to be 1.

[0109] It should be noted that the blind source separation of the abdomen sub-features is similar to the above method and will not be described in detail here.

[0110] It should be noted that step S22 can extract the feature vector of each of the separated target source signals through a neural network model, and classify each of the target source signals using a classification model according to the corresponding feature vector to obtain the vibration source of each of the target source signals, thereby further screening the target intestinal component corresponding to the intestinal movement.

[0111] S3: Extract the feature vector of the target intestinal component to obtain a target feature vector.

[0112] It should be noted that a neural network model may be used to extract the feature vector of the target observation signal, or a statistical method may be used to extract the feature vector of the target observation signal, or other feasible extraction methods may be used.

[0113] In this embodiment, the feature vector includes a feature vector obtained according to time domain features, frequency domain features and time-frequency features. Among them, the time domain features include amplitude, zero crossing rate and waveform entropy, etc., of course, it may not be limited to this. The frequency domain features include dominant frequency, band energy and spectrum entropy, etc., of course, it may not be limited to this. The time-frequency features include wavelet coefficients and Hilbert-Huang transform, etc., of course, it may not be limited to this.

[0114] S4: Match the target feature vector with the preset normal value. When the target feature vector does not meet the preset normal value, a first feedback signal is generated, and step S5 is continued. When the target feature vector meets the preset normal value, a second feedback signal is generated to complete the identification of the intestinal movement of the target patient within the target time period. Among them, the first feedback signal is used to indicate that the target observation signal belongs to an abnormal movement event, and the second feedback signal is used to indicate that the target observation signal does not belong to an abnormal movement event.

[0115] In this embodiment, step S4 also includes the following sub-steps: S41, estimating the intestinal parameters of the target patient according to the target feature vector to obtain target intestinal parameters. S42, obtaining the normal value range corresponding to each of the intestinal parameters. S43, matching each of the target intestinal parameters with the corresponding normal value range, if there is a target intestinal parameter that does not belong to the corresponding normal value range, generating the first feedback signal, if all the target intestinal parameters belong to the corresponding normal value range, generating the second feedback signal.

[0116] It should be noted that the intestinal parameters include the obvious degree of abdominal wall vibration, intestinal peristalsis frequency, intestinal peristalsis duration, propagation speed, bowel sound spectrum, bowel sound intensity and bowel sound duration, etc., of course, it may not be limited to this. The obvious degree of abdominal wall vibration, intestinal peristalsis frequency, intestinal peristalsis duration and propagation speed are estimated according to the target feature vector corresponding to the first abdominal sub-feature. The bowel sound spectrum, bowel sound intensity and bowel sound duration are estimated according to the target feature vector corresponding to the second abdominal sub-feature. The estimation of the intestinal parameters can be achieved by linear fitting, or by training a neural network model, using the target feature vector as the input of the neural network model, and the intestinal parameters as the output of the neural network model, and the neural network model is trained through a historical patient's physiological characteristic data set to achieve the estimation of the intestinal parameters.

[0117] In this embodiment, the preset normal value can be preset in the normal value range corresponding to each intestinal parameter. This application does not further limit the specific value method of the preset normal value.

[0118] It should be noted that intestinal movement can cause the patient's abdominal wall to vibrate, especially for patients with less abdominal fat or weak abdominal wall. The abdominal wall vibration data caused by intestinal movement collected by the BRS accelerometer is more obvious. When the patient's intestinal movement is abnormal, the abdominal wall vibration data will also be abnormal. For example, intestinal obstruction can cause the intestines above the obstruction site to strengthen peristalsis in an attempt to push the intestinal contents through the obstruction site, which in turn causes abnormal abdominal wall vibration data of the patient. The BRS accelerometer is used to collect the patient's abdominal wall vibration data, and the blind source separation and vibration source identification of the separated source signal are performed to remove the influence of heartbeat, breathing, etc. on intestinal movement, and to extract the abdominal wall vibration data caused by intestinal movement from the mixed signal, so as to reduce the influence of other vibration sources on the recognition of intestinal movement caused by the vibration of the abdominal wall, improve the recognition accuracy of intestinal movement, and then realize non-invasive detection of patients' intestinal movement, which can be used for long-term continuous monitoring of patients' intestinal movement.

[0119] S5. Estimate the direction of the target intestinal segment using beam forming to obtain the target direction.

[0120] It should be noted that the target intestinal segment is the intestinal segment that generates the target observation signal.

[0121] In this embodiment, step S5 also includes: S51, estimating the autocorrelation matrix between the BRS acceleration sensors based on all the second source signals. S52, calculating the inverse matrix of the autocorrelation matrix. S53, constructing the corresponding steering vector according to the position between the BRS acceleration sensors. S54, performing beam scanning successively according to the preset step, and taking the scanning angle corresponding to the signal peak point in all beam scanning operations as the direction from the intestinal segment generating abdominal wall vibration to the corresponding BRS acceleration sensor. The method for determining the target direction corresponding to the intestinal segment generating bowel sounds based on bowel sounds is similar to the above method and will not be described in detail here.

[0122] S6. Determine the distance from each target observation signal acquisition device to the target intestinal segment by triangulation to obtain the target distance.

[0123] S7. Obtain the position of the corresponding target intestinal segment according to all the target directions and all the target distances. The position of the intestinal segment generating abdominal wall vibration and the position of the intestinal segment generating bowel sounds can be determined by using a time difference based method TDOA.

[0124] It should be noted that when the target observation signal belongs to an abnormal motion event, the position of the intestinal segment that generates the target observation signal is determined by beamforming technology and triangulation positioning method, so as to locate the location of the abnormal motion event, and imaging technology can be further used to intuitively simulate imaging according to the location of the abnormal motion event.

[0125] In a possible implementation, refer to the attached Figure 2 As shown, step S22 also includes the following sub-steps:

[0126] S221. According to the probability that the vibration source of each target source signal belongs to all the vibration source categories, a candidate category probability set is obtained. The set consisting of all the target source signals is denoted as S, where S={s1, s2, ..., s i ,…,s n}, where s i represents the i-th target source signal obtained by the blind source separation operation, and n represents the total number of target source signals obtained by the blind source separation operation. The candidate category probability set is initially denoted as P, and P is initialized as:

[0127] ;

[0128] Among them, p i，j Represents the target source signal s i The probability of belonging to the jth vibration source category, m represents the total number of all vibration source categories.

[0129] S222. Take the maximum value of the candidate category probability from the candidate category probability set as the target category probability. S223. Set the vibration source category corresponding to the target category probability as the vibration source of the target source signal corresponding to the target category probability. S224. Determine whether the vibration source category corresponding to the target category probability is the intestine. If not, continue to execute S225. If so, the vibration source identification is completed to reduce the amount of calculation for screening the target source signal corresponding to the intestine and improve the recognition efficiency. S225 removes the candidate category probability that is the same as the vibration source category of the target category probability and the candidate category probability that is the same as the target source signal of the target category probability from the candidate category probability set to obtain the updated candidate category probability set.

[0130] In this embodiment, if p i，j is the maximum value of the elements in the candidate category probability set, then p i，j As the target category probability, the vibration source of the i-th target source signal is set to the j-th vibration source category, that is, the vibration source identification of the i-th target source signal is completed. When updating the candidate category probability set, the i-th target source signal and the j-th vibration source category whose vibration source identification is completed are removed, that is, all elements in the i-th row and all elements in the j-th column of the candidate category probability set are deleted to obtain a new candidate category probability set. For example, if p 1，1 is the maximum value of the elements in the current candidate category probability set, then the vibration source of the first target source signal is the first vibration source category, and the updated candidate category probability set for:

[0131] .

[0132] S225, determine whether the element in the updated candidate category probability set is 0, if not, set the value of P to , return to step S222 to continue execution, if yes, vibration source identification is completed.

[0133] It should be noted that the maximum value of the elements in the candidate category probability set indicates, on the one hand, that the probability of the vibration source of the target source signal belonging to the vibration source category is the highest among all the vibration source categories, and on the other hand, it also indicates that the probability of the target source signal belonging to the vibration source category is the highest among all the target source signals belonging to the vibration source category. According to the order of category probability from large to small, the corresponding vibration source category is set for each of the target source signals in turn, which takes into account both the probability ranking of the same target source signal belonging to different vibration source categories and the probability ranking of different target source signals belonging to the same vibration source category, thereby realizing multi-dimensional probability comparison and improving the recognition accuracy of abdominal wall vibration caused by intestinal movement.

[0134] In a feasible implementation, step S221 further includes the following sub-steps: S2211, respectively calculating the probability that the vibration source of each target source signal belongs to the heartbeat. S2212, respectively calculating the probability that the vibration source of each target source signal belongs to breathing. S2213, respectively calculating the probability that the vibration source of each target source signal belongs to the intestine. S2214, respectively calculating the probability that the vibration source of each target source signal belongs to the abdominal wall muscle.

[0135] In this embodiment, refer to the attached Figure 3 As shown, step S2214 also includes the following sub-steps:

[0136] S22141. Match each of the target source signals with the abdominal wall vibration data corresponding to the preset abdominal muscle state, and combine the successfully matched muscle states into corresponding candidate motion sequences in chronological order. The abdominal muscle state includes a muscle state type and a muscle state amplitude, which may not be limited thereto. Among them, the muscle state type includes relaxation, contraction and stretching, which may not be limited thereto. Among them, muscle relaxation indicates a relaxed state of the muscle without generating force. The muscle state amplitude includes a degree of muscle contraction and a degree of muscle stretching, which may not be limited thereto.

[0137] It should be noted that the preset abdominal muscle state can be the abdominal wall vibration data of the preset abdominal muscles in different states, or it can be the abdominal wall vibration data of the abdominal muscles of historical patients in different states. A dynamic programming algorithm, a DTW algorithm, or a FastDTW algorithm can be selected to match the target source signal with the abdominal wall vibration data corresponding to each of the preset abdominal muscle states. For example, the first preset abdominal muscle state is recorded as F1={contraction, x1}, the second preset abdominal muscle state is recorded as F2={stretching, x2}, and the third preset abdominal muscle state is recorded as F3={relaxation, x3}. The first target source signal is matched with the vibration data of the above three preset abdominal muscle states to obtain the following matching results: in the time period t1-t2, the first target source signal successfully matches F1; in the time period t2-t3, the first target source signal successfully matches F3; in the time period t3-t4, the first target source signal successfully matches F2; ​​in the time period t4-t5, the first target source signal successfully matches F3. The candidate motion sequence corresponding to the first target source signal is recorded as [F1, F3, F2, F3].

[0138] S22142: Obtain a motion video of the target patient in the target time period to obtain a target motion video. The target motion video may be obtained through monitoring video.

[0139] S22143, identifying the target action of the target patient in the target motion video. The target action includes an action that causes a change in the state of the abdominal muscles, such as bending, turning sideways, and walking. The movement of the skeletal key points of the target patient in the target motion video over time can be analyzed to identify whether the target patient in the target motion video has the target action, for example, by using a recurrent neural network to analyze the characteristics of the skeletal key point sequence to classify the target patient's actions, and the target action can be screened out according to the classification result.

[0140] S22144. Obtain the setting position of the BRS acceleration sensor to obtain the target detection position.

[0141] S22145, obtaining the abdominal muscle state corresponding to the target detection position and matching each of the target actions, and combining the abdominal muscle states corresponding to all the target actions into a target motion sequence in chronological order.

[0142] It should be noted that each of the target actions is obtained by combining different abdominal muscle states. For example, when leaning to the left, the external oblique and internal oblique muscles on the right side begin to contract to help rotate the trunk when leaning to the side; when leaning further to the side, the rectus abdominis and internal oblique muscles on the left side will also be involved to provide additional support and control to prevent excessive side tilt; when maintaining the sideways posture, the abdominal muscles will continue to remain tense to support the body weight and maintain balance; when returning to the frontal posture, the abdominal muscles will further contract to help the body return to the center position while preventing the body from shaking; after returning to the frontal posture, the abdominal muscles will gradually relax and return to the initial state to maintain core stability.

[0143] In this embodiment, the abdominal muscle state corresponding to the first target action A1 is [F2, F3], and the abdominal muscle state corresponding to the second target action A2 is [F1, F3]. The action recognition result of the first target motion video includes: in the time period T1-T2, the action of the target patient is A2, and in the time period T2-T3, the action of the target patient is A1. Therefore, in the time period T1-T3, the target motion sequence corresponding to the target patient can be recorded as [F1, F3, F2, F3].

[0144] S22146. Calculate the similarity between the target motion sequence and each of the candidate motion sequences respectively to obtain motion similarity; obtain the probability that the vibration source of the target source signal belongs to the abdominal muscles according to the motion similarity. The greater the motion similarity, the greater the probability that the vibration source of the target source signal corresponding to the candidate motion sequence belongs to the abdominal muscles. The smaller the motion similarity, the smaller the probability that the vibration source of the target source signal corresponding to the candidate motion sequence belongs to the abdominal muscles.

[0145] It should be noted that the motion similarity can be calculated based on the Euclidean distance or the cosine similarity.

[0146] In this embodiment, the method for calculating the motion similarity S includes:

[0147] ;

[0148] ;

[0149] ;

[0150] Among them, S k represents the similarity between the muscle state type of the target motion sequence and the muscle state type corresponding to the candidate motion sequence, T represents the total number of moments in the target time period, represents the muscle state type at the t-th moment in the target motion sequence, represents the muscle state type at the tth moment in the candidate motion sequence, S x represents the similarity between the muscle state amplitude of the target motion sequence and the muscle state amplitude corresponding to the candidate motion sequence, represents the muscle state amplitude at the t-th moment in the target motion sequence, represents the muscle state amplitude at the t-th moment in the candidate motion sequence, α represents the weight of type similarity, β represents the weight of amplitude similarity, and the value of α is greater than the value of β.

[0151] It should be noted that the differences in body size, abdominal fat thickness, and abdominal muscle size among different patients result in the same abdominal muscle state type when completing the same target action, but the amplitude of abdominal wall vibration is different. When calculating each reference motion sequence and target motion sequence, the similarity between the reference motion sequence and the target motion sequence corresponding to the target source signal generated by the real abdominal muscle will be reduced. Setting the weight value of amplitude similarity to be smaller than the weight value of type similarity is conducive to reducing the adverse effects of individual differences among patients on screening intestinal components from all target source signals and improving the recognition accuracy of intestinal motion.

[0152] It should be noted that in the process of long-term monitoring of the patient's intestinal movements, the patient will inevitably move. When the patient is in the process of movement, the abdominal wall vibration data collected by the BRS acceleration sensor includes not only the abdominal wall vibration data generated by breathing, heartbeat and intestinal movement, but also the abdominal wall vibration data generated by the contraction and stretching of the abdominal muscles, which leads to the existence of the target source signal due to the vibration source category of the abdominal muscles in the blind source separation result. By assuming that each target source signal is the abdominal wall vibration data generated by the change of the abdominal muscle state, it is matched with the abdominal wall vibration data corresponding to the preset abdominal muscle state to obtain the candidate motion sequence corresponding to each target source signal. According to the abdominal muscle states corresponding to all target actions, the patient's movements during the detection process are matched to generate the patient's target motion sequence, and the probability that the vibration source of the target source signal corresponding to the candidate motion sequence belongs to the abdominal muscle is obtained according to the similarity between each candidate motion sequence and the target motion sequence, so as to reduce the influence of the patient's movement on the intestinal movement recognition, improve the intestinal movement recognition accuracy due to the movement of the patient during the detection process, and expand the application scenario of long-term monitoring of the patient's intestinal movement.

[0153] In a feasible implementation, the method for calculating the candidate category probability set P further includes:

[0154] S2216, respectively obtain the probability that the vibration sources of all the target source signals belong to the same vibration source category, and obtain a signal probability vector. The signal probability vector corresponding to the vibration sources of all the target source signals belonging to the jth vibration source category can be denoted as P j , P j =[p 1，j ,…,p i，j ,…,p n，j ].

[0155] S2217: Normalize each of the signal probability vectors, and obtain the candidate category probability set according to the normalization result.

[0156] In this embodiment, the normalization method includes:

[0157] ;

[0158] ;

[0159] Among them, ||P j || represents the signal probability vector P j Length, Indicates P j Normalized signal vector probability.

[0160] In this embodiment, the initial value of the candidate category probability set P can be expressed as:

[0161] .

[0162] It should be noted that, since different vibration source categories have different characteristics, such as the characteristics of heartbeat signals and respiratory signals are relatively fixed, the similarity of heartbeat signal characteristics and respiratory signal characteristics between different patients are relatively large, and the probability that the vibration source of the target source signal belongs to the heartbeat and / or the probability that the vibration source of the target source signal belongs to the breathing can be judged by training the classification model with historical patient data. However, the similarity of the motion characteristics of different patients during the monitoring process is small, and it is difficult to judge the probability that the vibration source of the target source signal belongs to the abdominal muscles through the motion characteristics of historical patients. Therefore, when calculating the probability that the target source signal belongs to different vibration source types, different evaluation methods can be used to evaluate the probability that the target source signal belongs to each vibration source type. Each signal probability vector is processed by normalization to reduce the adverse effects of different probability evaluation methods on vibration source identification, eliminate the dimension, and reduce the difficulty of model design for calculating different vibration source types of the target source signal.

[0163] According to the second aspect of the present invention, the present invention claims a BRS acceleration-based intestinal movement intelligent recognition system, referring to the attached Figure 4As shown, including:

[0164] A data acquisition module, the data acquisition module is used to acquire abdominal characteristic data of a target patient in a target time period to obtain a target observation signal; wherein the abdominal characteristic data includes abdominal wall vibration data of the target patient collected by a BRS acceleration sensor;

[0165] A blind source separation module, wherein the blind source separation module is used to perform a blind source separation operation on the target observation signal to obtain a target source signal;

[0166] A source identification module, the source identification module is used to identify the vibration source category corresponding to each of the target source signals to screen the target intestinal components; wherein the vibration source categories include heartbeat, breathing and intestinal tract;

[0167] A feature extraction module, wherein the feature extraction module is used to extract a feature vector of the target intestinal component to obtain a target feature vector;

[0168] An abnormality judgment module, the abnormality judgment module is used to match the target feature vector with a preset normal value, and when the target feature vector does not meet the preset normal value, a first feedback signal is generated, and the first feedback signal is used to indicate that the target observation signal belongs to an abnormal motion event.

[0169] In a feasible embodiment, the number of each abdominal feature data collection device is not less than 3, and the system also includes a source positioning module, which is used to use beamforming to estimate the direction of the target intestinal segment to obtain the target direction; wherein the target intestinal segment is the intestinal segment that generates the target observation signal; the distance from each target observation signal collection device to the target intestinal segment is determined by a triangulation positioning method to obtain the target distance; and the position of the corresponding target intestinal segment is obtained according to all the target directions and all the target distances.

[0170] In a feasible implementation manner, the source identification module further includes:

[0171] A category probability calculation module, the category probability calculation module is used to obtain an initial value of a candidate category probability set according to the probability that the vibration source of each of the target source signals belongs to all the vibration source categories;

[0172] A vibration category identification module, the vibration category identification module is used to take the maximum value of the candidate category probability from the candidate category probability set as the target category probability; set the vibration source category corresponding to the target category probability to the vibration source of the target source signal corresponding to the target category probability; remove the candidate category probabilities that are the same as the vibration source category of the target category probability and the candidate category probabilities that are the same as the target source signal of the target category probability from the candidate category probability set to obtain the updated candidate category probability set; determine whether the elements in the updated candidate category probability set are 0, if not, go to the step of taking the maximum value of the candidate category probability from the candidate category probability set, if so, vibration source identification is completed.

[0173] In a feasible implementation, the vibration category identification module further includes determining whether the vibration source category corresponding to the target category probability is the intestine. If not, continue to execute the step of obtaining the updated candidate category probability set. If so, the vibration source identification is completed.

[0174] In a feasible implementation manner, the vibration source further includes abdominal muscles, and the category probability calculation module further includes:

[0175] A heartbeat probability calculation module, the heartbeat probability calculation module is used to obtain the probability that the vibration source of each target source signal belongs to a heartbeat;

[0176] A breathing probability calculation module, the breathing probability calculation module is used to obtain the probability that the vibration source of each target source signal belongs to breathing;

[0177] An intestinal probability calculation module, the intestinal probability calculation module is used to obtain the probability that the vibration source of each target source signal belongs to the intestinal tract;

[0178] A muscle probability calculation module, the muscle probability calculation module is used to obtain the probability that the vibration source of each target source signal belongs to the abdominal wall muscle;

[0179] Wherein, the muscle probability calculation module also includes:

[0180] A signal conversion module, the signal conversion module is used to match each of the target source signals with the abdominal wall vibration data corresponding to the preset abdominal muscle state, and combine the successfully matched muscle states into a corresponding candidate motion sequence in chronological order;

[0181] An image acquisition module, the image acquisition module is used to acquire the motion video of the target patient in the target time period to obtain a target motion video;

[0182] An action recognition module, the action recognition module is used to recognize the target action of the target patient in the target motion video;

[0183] A detection position acquisition module, wherein the detection position acquisition module is used to acquire the setting position of the BRS acceleration sensor to obtain the target detection position;

[0184] An action conversion module, the action conversion module is used to obtain the abdominal muscle state corresponding to the target detection position and matching each target action, and combine the abdominal muscle states corresponding to all the target actions into a target motion sequence in chronological order;

[0185] A probability calculation module is used to calculate the similarity between the target motion sequence and each of the candidate motion sequences respectively to obtain motion similarity; and obtain the probability that the vibration source corresponding to the target source signal belongs to the abdominal muscles according to the motion similarity.

[0186] In a feasible implementation, the probability calculation module further comprises: respectively calculating the similarity between the muscle state type of the target motion sequence and the muscle state type of each candidate motion sequence to obtain type similarity;

[0187] Respectively calculating the similarity between the muscle state amplitude of the target motion sequence and the muscle state amplitude of each candidate motion sequence to obtain amplitude similarity;

[0188] A linear weighted result of the type similarity and the amplitude similarity is calculated to obtain the motion similarity.

[0189] In a feasible implementation manner, the weight of the type similarity is greater than the weight of the amplitude similarity.

[0190] In a feasible embodiment, the category probability calculation module also includes a normalization processing module, which is used to respectively obtain the probability that the vibration sources of all the target source signals belong to the same vibration source category to obtain a signal probability vector; normalize each of the signal probability vectors, and obtain the candidate category probability set according to the normalization processing result.

[0191] In a feasible implementation manner, the abdominal characteristic data also includes bowel sounds.

[0192] See attached Figure 5 As shown, an embodiment of the present application provides an electronic device, including: a processor and a memory, the processor and the memory are interconnected and communicate with each other through a communication bus and / or other forms of connection mechanisms (not shown), the memory stores a computer program executable by the processor, and when the computing device is running, the processor executes the computer program to execute the system in any optional implementation of the above embodiments.

[0193] The embodiment of the present application provides a storage medium, and when the computer program is executed by the processor, the system in any optional implementation of the above embodiment is executed. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read-Only Memory, referred to as EPROM), programmable read-only memory (Programmable Read-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, disk or optical disk.

[0194] In the embodiments provided in the present application, it should be understood that the disclosed system can be implemented in other ways. The system embodiments described above are merely schematic. For example, the division of the modules is only a logical function division, and can be implemented in another way. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some communication interface, system or unit, which can be electrical, mechanical or other forms.

[0195] In addition, the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple 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.

[0196] Furthermore, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0197] Flowcharts are used herein to illustrate the steps of the method of the embodiments of the present disclosure. It should be understood that the preceding or following steps are not necessarily performed precisely in order. On the contrary, various steps may be evaluated in reverse order or simultaneously. At the same time, other operations may also be added to these processes.

[0198] Unless otherwise defined, all terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which the present disclosure belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or highly formal sense, unless explicitly defined as such herein.

[0199] The above is a detailed introduction to a method and system for intelligently identifying intestinal movements based on BRS acceleration. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only an embodiment of the present application, which is only used to help understand the method and system for intelligently identifying intestinal movements based on BRS acceleration of the present application, and is not used to limit the scope of protection of the present application. At the same time, for those skilled in the art, the present application may have various changes and variations. Any modifications and equivalent substitutions made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for intelligently identifying intestinal movements based on BRS acceleration, characterized in that: include: Obtain abdominal characteristic data of a target patient in a target time period to obtain a target observation signal; Wherein, the abdominal characteristic data includes abdominal wall vibration data collected by a body acoustic wave resonator acceleration sensor; Performing a blind source separation operation on the target observation signal to obtain a target source signal; Identifying the vibration source category corresponding to each of the target source signals to screen the target intestinal component; Wherein, the vibration source categories include heartbeat, breathing and intestines; Extracting the characteristic vector of the target intestinal component to obtain a target characteristic vector; Matching the target feature vector with a preset normal value, and generating a first feedback signal when the target feature vector does not meet the preset normal value, wherein the first feedback signal is used to indicate that the target observation signal belongs to an abnormal motion event; The step of identifying the vibration source category corresponding to each of the target source signals further includes: Obtaining an initial value of a candidate category probability set according to the probability that the vibration source of each of the target source signals belongs to all of the vibration source categories; Taking the maximum value of the candidate category probability from the candidate category probability set as the target category probability; Setting the vibration source category corresponding to the target category probability as the vibration source of the target source signal corresponding to the target category probability; Remove the candidate category probability that is the same as the vibration source category of the target category probability and the candidate category probability that is the same as the target source signal of the target category probability from the candidate category probability set to obtain an updated candidate category probability set; Determine whether the element in the updated candidate category probability set is 0. If not, go to the step of taking the maximum value of the candidate category probability from the candidate category probability set. If so, vibration source identification is completed.

2. The method for intelligently identifying intestinal movements based on BRS acceleration according to claim 1, characterized in that: The number of each of the abdominal characteristic data collection devices is not less than 3, and after generating the first feedback signal, the method further includes: The direction of the target intestinal segment is estimated by using beam forming to obtain the target direction; Wherein, the target intestinal segment is the intestinal segment that generates the target observation signal; Determine the distance from each target observation signal acquisition device to the target intestinal segment by triangulation to obtain the target distance; The position of the corresponding target intestinal segment is obtained according to all the target directions and all the target distances.

3. The method for intelligently identifying intestinal movements based on BRS acceleration according to claim 2, characterized in that: Before determining whether the element in the updated candidate category probability set is 0, after setting the vibration source category corresponding to the target category probability to the vibration source of the target source signal corresponding to the target category probability, it also includes: determining whether the vibration source category corresponding to the target category probability is intestines; if not, continuing to execute the step of obtaining the updated candidate category probability set; if so, vibration source identification is completed.

4. The method for intelligently identifying intestinal movements based on BRS acceleration according to claim 2, characterized in that: The vibration source also includes abdominal muscles, and the probability that the vibration source according to each of the target source signals belongs to all the vibration source categories also includes: Obtaining the probability that the vibration source of each target source signal belongs to a heartbeat; Obtaining the probability that the vibration source of each target source signal belongs to breathing; Obtaining the probability that the vibration source of each target source signal belongs to the intestine; Obtaining the probability that the vibration source of each target source signal belongs to the abdominal wall muscle; Among them, the probability that the vibration source of each target source signal belongs to the abdominal wall muscle is also included: Matching each of the target source signals with the abdominal wall vibration data corresponding to the preset abdominal muscle state, and combining the successfully matched muscle states into corresponding candidate motion sequences in chronological order; Acquire the motion video of the target patient in the target time period to obtain a target motion video; identifying a target motion of the target patient in the target motion video; Obtaining the setting position of the BAW resonator acceleration sensor to obtain the target detection position; Acquire the abdominal muscle state corresponding to the target detection position and matching each of the target actions, and combine the abdominal muscle states corresponding to all the target actions into a target motion sequence in chronological order; Calculating the similarity between the target motion sequence and each of the candidate motion sequences respectively to obtain motion similarity; The probability that the vibration source corresponding to the target source signal belongs to the abdominal muscles is obtained according to the motion similarity.

5. The method for intelligently identifying intestinal movements based on BRS acceleration according to claim 4, characterized in that: The step of respectively calculating the similarity between the target motion sequence and each of the candidate motion sequences further includes: Respectively calculating the similarity between the muscle state type of the target motion sequence and the muscle state type of each candidate motion sequence to obtain type similarity; Respectively calculating the similarity between the muscle state amplitude of the target motion sequence and the muscle state amplitude of each candidate motion sequence to obtain amplitude similarity; A linear weighted result of the type similarity and the amplitude similarity is calculated to obtain the motion similarity.

6. The method for intelligently identifying intestinal movements based on BRS acceleration according to claim 5, characterized in that: The weight of the type similarity is greater than the weight of the amplitude similarity.

7. The method for intelligently identifying intestinal movements based on BRS acceleration according to claim 4, characterized in that: The method for calculating the candidate category probability set also includes: Respectively obtaining the probability that the vibration sources of all the target source signals belong to the same vibration source category to obtain a signal probability vector; Each of the signal probability vectors is normalized, and the candidate category probability set is obtained according to the normalization result.

8. A method for intelligently identifying intestinal movements based on BRS acceleration according to any one of claims 1 to 7, characterized in that: The abdominal characteristic data also includes bowel sounds.

9. An intelligent intestinal movement recognition system based on BRS acceleration, characterized in that: include: A data acquisition module, the data acquisition module is used to acquire abdominal characteristic data of a target patient to obtain a target observation signal; wherein the abdominal characteristic data includes abdominal wall vibration data collected by a body acoustic wave resonator acceleration sensor; A blind source separation module, wherein the blind source separation module is used to perform a blind source separation operation on the target observation signal to obtain a target source signal; A source identification module, the source identification module is used to identify the vibration source corresponding to each of the target source signals to screen the target intestinal components; wherein the vibration sources include heartbeat, breathing and intestinal movement; A feature extraction module, wherein the feature extraction module is used to extract a feature vector of the target intestinal component to obtain a target feature vector; An abnormality judgment module, the abnormality judgment module is used to match the target feature vector with a preset normal value, and when the target feature vector does not meet the preset normal value, a first feedback signal is generated, and the first feedback signal is used to indicate that the target observation signal belongs to an abnormal motion event The source identification module also includes: A category probability calculation module, the category probability calculation module is used to obtain an initial value of a candidate category probability set according to the probability that the vibration source of each of the target source signals belongs to all the vibration source categories; A vibration category identification module, the vibration category identification module is used to take the maximum value of the candidate category probability from the candidate category probability set as the target category probability; set the vibration source category corresponding to the target category probability to the vibration source of the target source signal corresponding to the target category probability; remove the candidate category probabilities that are the same as the vibration source category of the target category probability and the candidate category probabilities that are the same as the target source signal of the target category probability from the candidate category probability set to obtain the updated candidate category probability set; determine whether the elements in the updated candidate category probability set are 0, if not, go to the step of taking the maximum value of the candidate category probability from the candidate category probability set, if so, vibration source identification is completed.

Images