Multi-modal respiratory signal acquisition system based on personalized wearing and calibration

Abstract

The invention relates to the technical field of medical equipment, in particular to a multi-modal respiratory signal acquisition system for realizing appearance and data dual individuation for multi-body-type respiratory patient crowds, which can generate an individualized patient respiratory data set through wearable waistcoat design and respiratory data weight calibration. The system mainly comprises a waistcoat body, an acquisition module, a communication module, a storage module and a personalized weight software module. The acquisition module comprises A-type ultrasound, an electroencephalogram sensor, a myoelectricity sensor, an inertial sensor, a flow meter and a body surface tension band. In the collecting process, firstly, a patient wears a waistcoat, the length of a bandage is adjusted, sensor collecting point positions are adjusted to medical reference positions, and then the sensors are synchronously triggered through the communication module to conduct real-time data collection. And a personalized weight software module is used for generating weight calibration for validity and significance of each modal data according to the basic medical record of the patient, and finally, the data and the weight are jointly stored in a storage module. According to the invention, the comprehensiveness and precision of data can be improved, and patient personalized acquisition which is widely applicable is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical equipment, in particular to a wearable multi-modal respiratory signal acquisition system suitable for a respiratory patient population with multiple body types and multiple diseases. The system realizes high-precision comprehensive perception of neural regulation, muscle kinematics and gas dynamics on the whole time sequence of respiration by compactly integrating multiple complementary biosignal sensors in an elastic waistcoat body through the dual personalization design of ergonomic shape adaptation structure design and data weight intelligent calibration algorithm based on patient medical records, and generates a structured personalized patient respiratory dataset with saliency annotation, serving precise diagnosis and treatment. BACKGROUND

[0002] Existing respiratory monitoring equipment usually uses single modal signals such as chest and abdominal movement, airflow and blood oxygen, which is difficult to comprehensively reflect the multi-dimensional signal complex correlation of neural regulation, muscle synergistic contraction, thoracic dynamics and gas exchange during respiration. And the traditional equipment has the problems of poor body adaptation, fixed sensor layout and low data fusion accuracy, which cannot provide high-value data sets with high information density and clinical direction for individualized diagnosis and treatment.

[0003] At the same time, even if it has multi-modal acquisition capability, the existing multi-modal data acquisition system often ignores the direct influence of patient's basic pathological characteristics on the effectiveness of specific modal signals, lacks dynamic signal saliency calibration mechanism for specific diseases, such as uneven ventilation caused by chronic obstructive pulmonary disease (COPD), muscle electrical signal attenuation or diaphragm dysfunction caused by neuromuscular diseases (such as ALS), resulting in redundant data or missing key features, and low overall utilization.

[0004] Therefore, it is urgent to develop a respiratory signal acquisition system with individual adaptability and intelligent data calibration capability to effectively improve the accuracy and efficiency of respiratory disease diagnosis and individualized treatment plan development. SUMMARY

[0005] The present application proposes a multi-modal respiratory signal acquisition system, which integrates spatially variable arrangement of multi-source sensors through customized wearable waistcoat, and realizes dual personalization of shape adaptation and data value through personalized weight calibration algorithm based on knowledge graph. The system can synchronously acquire bioelectric, motion and airflow signals of respiration under a unified time reference, thereby reflecting respiratory intention, action and efficiency, and optimizing the clinical relevance and information quality of the data set by deeply integrating signal weight and raw data.

[0006] A multi-modal respiratory signal acquisition system, characterized by: a wearable waistcoat body, a multi-modal respiratory signal sensor acquisition module, an upper computer data communication module, a personalized weight software module and a data storage module.

[0007] The wearable vest body provides three basic sizes based on ergonomics to meet the needs of the general population. The body is made of 5mm thick black nylon fabric with good elasticity, breathability and durability. Adjustable buckle straps with wide range of adjustment are used to fix the waist and shoulders. The key sensor basic positioning is as follows: A-type ultrasonic and electromyographic sensor electrode patches are distributed near the corresponding diaphragm projection area of the human body. Inertial sensors and body surface tension belts are placed on the midline where the human chest and abdomen have significant movement amplitude. In addition, each sensor is attached with a magic tape, which can be adjusted according to the specific body type and body position of the patient to ensure the accuracy of positioning and the reliability of the signal.

[0008] Preferably, the adjustable buckle strap adopts a segmented ratchet locking structure with high reliability and easy operation. The adjustment range can adapt to patients with chest circumference of 65-160cm, covering children to obese adults. The buckle is equipped with a precise spring-loaded ratchet mechanism, providing uniform locking force and preventing accidental loosening. The outside of the strap is provided with clear and wear-resistant centimeter-level scale marks to facilitate medical staff to quickly locate the initial adjustment position. The inner layer of the strap is composed of a high-friction coefficient silicone anti-slip stripe array, which can effectively ensure the stability of the sensor application and the anti-slip performance of the overall system, meeting the respiratory monitoring needs in multiple scenarios.

[0009] The multi-modal respiratory signal sensor acquisition module includes A-type ultrasonic, electroencephalogram sensor, electromyography sensor, inertial sensor, flow meter and body surface tension belt. The electroencephalogram sensor (preferably dry electrode) captures the frontal lobe area respiratory-related intention electrical signal (such as preparatory inspiration potential), the A-type ultrasonic patch and the electromyography electrode patch array synchronously monitor the diaphragm displacement and surface electromyography activity, providing direct observation of neural drive and muscle response, the inertial sensor (IMU, including three-axis accelerometer and gyroscope) and body surface tension belt (containing strain sensor) detect the three-dimensional movement amplitude, speed and mode of the thorax and abdomen, and the flow meter (integrated in the lightweight mask) monitors the airflow velocity at the mouth and nose. The above multi-source sensors work together to realize comprehensive perception of the complete respiratory time sequence from neural intention to muscle action to gas flow.

[0010] Preferably, A-type ultrasonic is introduced as the core innovative sensing unit in the acquisition module, which can use high-frequency ultrasonic beams to penetrate the body surface and directly detect the one-dimensional displacement signal (distance-time curve) generated by the vertical movement of the diaphragm. Based on the diaphragm electrical excitation signal collected by the electromyography sensor, the shortcomings of the electromyography signal being easily affected by subcutaneous fat thickness and muscle fiber orientation are overcome, further improving the detection accuracy and reliability of diaphragm mechanical activity, especially suitable for obese or atrophic patients.

[0011] The upper computer data communication module utilizes software development to have a friendly interactive interface visual upper computer software, each sensor acquisition unit transmits data to the upper computer through low-power Bluetooth 5.0 or a higher version and stores locally or in the cloud, and the upper computer provides real-time waveform display, data quality monitoring and acquisition control functions.

[0012] The personalized weight software module analyzes the medical records text in the electronic health records (EHR) of the patient based on the constructed respiratory disease knowledge graph technology (integrating authoritative medical literature, clinical guidelines and expert consensus) to identify diagnosis, symptoms, basic signs and key pathological features, and then dynamically assigns the significance weight of each modality data in different stages (such as the inspiration phase and the expiration phase) or specific events (such as coughing and sighing) in the entire respiratory cycle, and the weight value ranges from 0.0 to 1.0, and the total is 1.0. For example, for amyotrophic lateral sclerosis (ALS) patients, because the diaphragmatic electromyographic signal is weak but the intention signal is relatively preserved, the electromyographic signal weight is dynamically increased to 0.8, and the airflow signal weight is reduced to 0.2 because there may be ventilation disorders; and for COPD patients, the weights of airflow and chest and abdominal movement signals may be increased.

[0013] The data storage module adopts high-speed and large-capacity storage devices (such as solid state disks SSD) to uniformly record the respiratory modality original data after time stamp alignment and the corresponding significance weight vector generated by the software module, and saves them as patient personalized respiratory signal data sets in a standardized format, including metadata (patient ID, time, medical record abstract, weight configuration).

[0014] Compared with the prior art, the advantages of the present application are:

[0015] 1. The present application adopts double personalization design, which ensures signal quality and reliability through body type adaptive waistcoat design, and improves the clinical value of data through medical record driven weight calibration, and at the same time solves the individualized needs of physical wear comfort and data value extraction targeting and accuracy.

[0016] 2. The present application realizes highly integrated multi-modal collaborative acquisition, integrates at least six types of time and space synchronous complementary respiratory signals covering neural intention (EEG), muscle electrical activity (EMG), core organ movement (A-mode ultrasound), body surface movement (inertia, body surface tension belt) and respiratory efficiency (airflow) signals, overcomes the limitations of single modality, and constructs a multi-dimensional respiratory signal comprehensive model reflecting the respiratory function of the patient, providing a rich data basis for accurate analysis. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is a front view of the overall appearance of the present application, Figure 2 is a back view of the overall appearance of the present application.

[0018] Figure 3 The figure is a schematic diagram of the internal sensor distribution of the jacket, and the precise installation position and coverage area of each sensor on the jacket lining are marked in detail.

[0019] In the figure, 1 is an electroencephalogram sensor, 2 is a respiratory flowmeter mask, 3 is a wearable jacket body, 4 is a segmented adjustable strap, 5 is a ratchet buckle, 6 is an upper computer communication module, 7 is an electromyography sensor electrode patch, 8 is a body surface tension belt, 9 is an A-type ultrasonic patch, and 10 is an inertial sensor. DETAILED DESCRIPTION

[0020] The technical solutions of the present application will be described clearly and completely in combination with the drawings and examples. Obviously, the described examples are only a part of the examples of the present application, rather than all the examples. Based on the examples in the present application, all other examples obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0021] The present application provides a technical solution: a multi-modal respiratory signal acquisition system based on individualized wearing and calibration, comprising: 1-electroencephalogram sensor, 2-respiratory flowmeter mask, 3-wearable jacket body, 4-segmented adjustable strap, 5-ratchet buckle, 6-upper computer communication module, 7-electromyography sensor electrode patch, 8-body surface tension belt, 9-A-type ultrasonic patch, and 10-inertial sensor. Each sensor is connected to the data communication module located on the front chest of the jacket through a flexible cable or a wireless way.

[0022] Installation and initialization: the respiratory signal acquisition system needs to be installed, worn and initialized before use. First, the electroencephalogram sensor headband is worn on the corresponding position of the patient's head (Fp1 / Fp2), the flowmeter mask is worn on the patient's mouth and nose and the sealing property is ensured, the wearable jacket body is worn on the patient's chest and abdomen, the shoulder strap and the side waist strap are preliminarily positioned, the tightness is adjusted to the appropriate scale by operating the segmented ratchet buckle strap (pulling the end of the strap and hearing the ratchet engagement sound), and the inner layer of silicone anti-slip stripes is fully attached to the skin without obvious compression, so as to ensure that each sensor achieves good and stable contact with the skin and reduces motion artifacts. After the acquisition system is powered on, the initialization process is performed, including self-checking of each sensor, automatic pairing and connection confirmation with the upper computer through Bluetooth, and signal baseline calibration for individual baseline noise.

[0023] The electroencephalogram sensor is arranged at the Fp1 / Fp2 site according to the 10-20 international standard system to capture the respiratory-related cortical potential, the A-mode ultrasonic patch is attached to the human body near the anterior axillary line of the 7th-10th intercostal space (the common surface projection area of the diaphragm), the electromyogram electrode patch is distributed on the lower intercostal muscles and costal margins on the bilateral anterior axillary lines, the electrode spacing is referred to the medical measurement standard position, the inertial sensor is fixed to the midpoint of the sternum xiphoid midline and the umbilical cord to monitor the chest wall movement, the body surface tension belt is wrapped around the chest and abdomen and kept in a moderate and natural tight state (without limiting the respiratory movement), and the flow meter is connected to the oronasal mask through a flexible pipeline to ensure that the air flow monitoring channel is completely sealed and has no leakage.

[0024] The positions of the sensors in the wearable vest on the preset can be adjusted according to the patient's body shape and anatomical landmarks (such as ribs, navel) through the magic tape, so as to maximize the patient's individualized wearing comfort and signal acquisition optimization requirements.

[0025] Multi-modal respiratory signal synchronous acquisition: after initialization, the electroencephalogram sensor and the electromyogram sensor acquire microvolt-level bioelectric signals, the A-mode ultrasonic sensor transmits and receives ultrasonic waves to acquire one-dimensional direction movement displacement signals of the diaphragm, the inertial sensor measures the movement acceleration of the chest wall in the vertical direction, the body surface tension belt detects the circumferential strain change amount of the chest wall, and the flow meter acquires the air flow velocity signal through the pressure difference principle. Through the upper computer software interface, a "start acquisition" instruction is sent, the instruction is broadcasted through the internal Bluetooth communication module, and all multi-modal sensors are synchronously triggered to perform real-time acquisition of respiratory signals. The hardware clock built in each sensor receives the synchronization signal, and the high-precision time sequence consistency of all channel signals in the respiratory cycle is ensured.

[0026] Individualized weight calibration: during the data acquisition process, the individualized weight calibration process is started synchronously. The individualized weight software module obtains the patient's imported medical record data based on the pre-constructed knowledge graph technology, analyzes the expert experience knowledge contained therein, extracts the diagnosis keywords through the integrated natural language processing, maps the keywords to the pathological feature nodes in the knowledge graph, and simultaneously refers to the real-time acquisition data quality of each modality respiratory signal, dynamically calculates and assigns the weight calibration scheme of each modality respiratory signal, and generates a one-dimensional weight calibration vector.

[0027] Data processing and storage: the original respiratory data obtained by acquisition is transmitted to the buffer zone of the upper computer. Firstly, the built-in filtering processing algorithm is used to perform real-time or offline preprocessing on the multi-modal respiratory signals to improve the signal-to-noise ratio, and then the temporary cache is performed with the patient ID and acquisition time stamp as the index. The storage module uses the on-board or externally connected solid state drive (SSD) to save the pre-processed multi-modal respiratory signal original waveform, individualized weight parameter vector and medical record associated metadata in the form of independent data packets according to the accurate time sequence, and generates a standardized, traceable and individualized patient respiratory data set.

[0028] The solution provided by the present application completes two specific works by designing a personalized respiratory signal acquisition system: first, the body size adaptation structure based on ergonomics is innovatively adopted to solve the wearing flexibility problem, and the modal weight calibration based on knowledge graph is adopted to solve the data value optimization problem, realizing double personalized design, synchronously solving the individualized needs of the comfort and stability of physical wearing and the accuracy and effectiveness of data value extraction, providing solid and reliable adaptive basic data support for the precise diagnosis and treatment of respiratory diseases; second, six modal respiratory signal sensors such as A-type ultrasonic, electromyography, electroencephalogram, inertial sensor, body surface tension belt and flow meter are efficiently integrated on the wearable platform, the bioelectric, motion and airflow signals reflecting the whole process of respiration are synchronously collected, the patient's respiratory function model containing high-fidelity original signals and feature parameters after weight weighting is constructed, and fine multi-dimensional perception and quantitative evaluation on the neural-muscle-airflow time sequence are realized.

Claims

1. A multimodal respiratory signal acquisition system based on personalized wearable devices and calibration, characterized in that: Wearable vest, multimodal respiratory signal sensor acquisition module, host computer data communication module, personalized weighting software module and storage module.

2. The wearable vest as described in claim 1 is designed based on ergonomics to fit the size of the general population. It is made of 5mm thick black nylon fabric and custom-cut. Adjustable distributed ratchet buckle straps are placed on the upper and lower sides of the back. Type A ultrasound and electromyography sensor electrode patches are distributed near the corresponding human diaphragm. Inertial sensors and body surface tension bands are placed at the corresponding human chest and abdomen midline.

3. The multimodal respiratory signal sensor acquisition module as described in claim 1 includes A-mode ultrasound, an electroencephalogram (EEG) sensor, an electromyography (EMG) sensor, an inertial sensor, a flow meter, and a body surface tension band, all of which are standard components. The EEG electrodes are arranged in a 10-20 system at Fp1 / Fp2 sites to capture the electrical signals of respiratory intention. The A-mode ultrasound and EMG electrodes synchronously monitor diaphragmatic displacement and electromyographic activity. The inertial sensor and body surface tension band detect three-dimensional thoracic motion. The flow meter is connected to a mask interface to monitor airflow dynamic parameters. The multiple sensors work together to achieve comprehensive perception of the respiratory time series.

4. The host computer data communication module as described in claim 1 utilizes software development to create a visual host computer software with a user-friendly interactive interface. Each sensor acquisition unit transmits data to the host computer via Bluetooth Low Energy 5.0 or a higher version and stores it locally or in the cloud. The host computer provides real-time waveform display, data quality monitoring, and acquisition control functions.

5. The personalized weighting software module as described in claim 1, based on the constructed respiratory disease knowledge graph technology to analyze expert experience rules, analyzes the patient's electronic medical record text through natural language processing, identifies diagnosis, symptoms, basic signs and key pathological features, and then personally assigns significance weights to each modality of data in different stages or specific scenarios throughout the respiratory cycle, with weight values ​​ranging from 0.0 to 1.0 and a total of 1.

0.

6. The storage module as described in claim 1 uses a high-speed, large-capacity solid-state drive to uniformly record the original data of each respiratory modality after timestamp alignment and the corresponding saliency weight vector generated by the software module, and saves them as a patient-specific respiratory signal dataset in a standardized format.

Images