Pediatric respiratory tract cleaning method and system based on multi-mode vibration and physiological feedback

Abstract

The application provides a pediatric respiratory tract cleaning method and system based on multi-mode vibration and physiological feedback, and belongs to the technical field of pediatric respiratory medical equipment and intelligent diagnosis and treatment. The method comprises the following steps: collecting multi-modal input data, generating physiological characteristic signals through a CNN+LSTM fusion model, and determining initial adaptive conditions; combining medical images and respiratory dynamics parameters to construct an individualized respiratory tract simulation model and perform vibration simulation; collecting real-time physiological parameters and performing noise reduction processing to obtain feedback data; comparing the preset threshold to adjust the vibration parameters, and performing actual cleaning after reaching the standard. The application realizes accurate adaptation of vibration parameters through individualized simulation and physiological feedback closed loop, improves cleaning safety and efficiency, reduces the risk of airway damage, and adapts to the physiological characteristics of young children.

Description

Technical Field

[0001] This invention relates to the field of pediatric respiratory medical equipment and intelligent diagnosis and treatment technology, and in particular to a pediatric airway clearance method and system based on multi-mode vibration and physiological feedback. Background Technology

[0002] In pediatric clinical practice, respiratory secretion accumulation is a common respiratory problem in newborns, infants, and preschool children. Secretion accumulation can easily lead to poor ventilation, decreased blood oxygen saturation, and even complications such as pneumonia and respiratory failure. Therefore, airway clearance is a routine and crucial diagnostic and treatment procedure in pediatric respiratory medicine. Currently, commonly used pediatric airway clearance methods mainly include manual chest percussion, mechanical vibration clearance, and nebulized inhalation-assisted clearance. Among these, mechanical vibration clearance has become the mainstream method due to its convenience and high efficiency. However, the vibration parameters of existing vibration clearance devices are mostly fixed, simply differentiated based on the child's age and weight, without individualizing the device according to the child's physiological tolerance, respiratory anatomy, secretion status, and behavioral cooperation. Inappropriate parameters can easily lead to airway wall damage, child agitation and resistance, or insufficient vibration intensity resulting in incomplete secretion clearance, thus causing insufficient adaptation of the initial vibration parameters.

[0003] Therefore, this invention proposes a pediatric airway clearance method and system based on multi-mode vibration and physiological feedback. Summary of the Invention

[0004] This invention provides a pediatric airway clearance method and system based on multi-mode vibration and physiological feedback to solve the aforementioned technical problems.

[0005] This invention provides a pediatric airway clearance method based on multi-mode vibration and physiological feedback, comprising:

[0006] Step 1: Acquire multimodal input data from pediatric patients and input it into a pre-trained neural network model to generate multidimensional physiological representation signals to determine the initial adaptation conditions for airway clearance;

[0007] Step 2: Based on the multi-dimensional physiological characterization signals and the synchronously acquired real-time medical image stream and respiratory dynamic parameters, dynamically construct an individualized airway simulation model corresponding to the pediatric patient, and control the multi-mode vibration module to apply vibration to the layered body surface area corresponding to the individualized airway simulation model according to the initial adaptation conditions for simulation. The layered body surface area includes: core vibration target area, auxiliary vibration target area and edge avoidance area.

[0008] Step 3: Real-time physiological parameters of pediatric patients during the airway clearance simulation process are collected, and the collected real-time physiological parameters are denoised and standardized to obtain feedback data for adjustment and judgment. The real-time physiological parameters include at least respiratory rate, blood oxygen saturation, airway resistance and breath sound signal.

[0009] Step 4: Compare and analyze the feedback data with the preset pediatric safety physiological threshold and secretion clearance judgment threshold, and adjust the vibration mode and corresponding vibration parameters of the multi-mode vibration module in real time until the airway clearance standard is met, and then perform actual airway clearance on the pediatric patient.

[0010] Preferably, the initial adaptation conditions for airway clearance are determined, including:

[0011] The multimodal input data is input into a pre-trained neural network model to generate a multi-dimensional physiological representation signal that includes physiological tolerance, cleanup risk and behavioral adaptability. The multi-dimensional physiological representation signal is a high-dimensional vector, and each dimension corresponds to a feature weight and a risk level label.

[0012] Based on the multi-dimensional physiological characterization signals, the K-means clustering algorithm is used to perform risk stratification, dividing pediatric patients into three physiological tolerance risk layers: low, medium, and high. Each risk layer corresponds to a preset vibration parameter constraint range.

[0013] The local pediatric airway clearance case knowledge base is invoked to match the historical cases with the highest similarity to the current patient's characteristic signals, and the vibration parameter constraint range is optimized and corrected.

[0014] Based on real-time environmental parameters and equipment response parameters for airway clearance, the corrected vibration parameter constraint range is dynamically fine-tuned to obtain initial adaptation conditions. These initial adaptation conditions include: initial range of vibration frequency, initial range of vibration intensity, initial sequence of vibration direction, and corresponding safety warning threshold and clearance efficiency threshold.

[0015] Preferably, the multimodal input data includes at least: static basic physiological parameters: age, weight, developmental stage label, basic respiratory rate, estimated viscosity of respiratory secretions, electroencephalogram (EEG) signals, and near-infrared brain functional imaging signals;

[0016] Behavioral status data: agitation score, cooperation score.

[0017] Preferably, a dynamically constructed individualized respiratory simulation model for pediatric patients includes:

[0018] The real-time medical image stream is time-aligned and reconstructed in three dimensions to extract dynamic geometric features of the airway, including airway lumen diameter, airway wall thickness, volume and distribution density of secretions, and airway deformation field during the respiratory cycle. At the same time, the respiratory dynamic parameters are feature extracted to obtain airway resistance spectrum, respiratory flow waveform and compliance change curve.

[0019] The extracted dynamic geometric features are used as the initial geometric structure of the simulation model, and the feature extraction results of respiratory dynamic parameters are used as the input boundary of the fluid dynamics simulation. In addition, the tolerance features in the multi-dimensional physiological characterization signals are used as the safety constraint threshold for airway deformation, limiting the upper limit of airway wall stress and secretion migration speed during the simulation process, thus constructing a finite element simulation framework.

[0020] The subsequent frame sequence of the real-time synchronized medical image stream is used to invert the airway deformation field based on the non-rigid image registration algorithm, update the geometric structure and material properties of the finite element simulation framework, and modify the fluid dynamic boundary conditions in combination with the real-time changes of respiratory dynamic parameters to obtain an individualized airway simulation model.

[0021] Preferably, the simulation is performed by controlling the multi-mode vibration module to apply vibration to the layered surface region corresponding to the individualized respiratory tract simulation model according to the initial adaptation conditions, including:

[0022] Based on the three-dimensional structure and secretion distribution of the individualized respiratory tract simulation model, a layered body surface region and vibration depth range corresponding to the respiratory tract target region are generated, and the initial vibration parameter combination of the multi-mode vibration module is generated according to the initial adaptation conditions.

[0023] The control multi-mode vibration module applies vibration to the layered body surface region according to the initial vibration parameter combination, and at the same time collects the simulation response data of the individualized respiratory tract simulation model in real time, including airway wall stress distribution, secretion displacement field, airway pressure change and vibration energy attenuation coefficient in respiratory tissue;

[0024] The simulation response data is compared with the preset vibration efficiency threshold and safety constraint threshold. If the simulation response data does not meet the standards, the parameter combination of the multi-mode vibration module is dynamically adjusted based on the optimization direction of the vibration parameters retrieved from the individualized airway simulation model, and the simulation is repeated until the simulation response data meets the preset vibration efficiency and safety constraint conditions. The optimized multi-mode vibration parameter combination and the corresponding layered body surface area are then output as the initial vibration control scheme for subsequent actual airway cleaning.

[0025] Preferably, when feedback data shows that the physiological state of a pediatric patient exceeds a preset pediatric safety physiological threshold, the vibration intensity is reduced and switched to a gentle vibration mode;

[0026] When feedback data shows that the respiratory secretions have not reached the clearance standard, the vibration frequency and direction are adjusted according to the loosening of the secretions until the feedback data shows that the respiratory secretions have reached the clearance standard and the pediatric patient's physiological state is maintained within the preset pediatric safe physiological threshold range. Then, the vibration cleaning stops, and the simulation process of one respiratory clearance is completed.

[0027] Preferably, based on real-time environmental parameters and equipment response parameters of airway clearance, the corrected vibration parameter constraint range is dynamically fine-tuned, including:

[0028] The device response parameters of the mechanical vibration unit and the acoustic vibration unit in the multi-mode vibration module are collected, and a state reliability matrix is ​​constructed, wherein the state reliability matrix contains only two off-diagonal elements;

[0029] Based on the state confidence matrix, the first comprehensive fit degree of the mechanical vibration unit and the acoustic vibration unit are calculated respectively, and the unit corresponding to the maximum value of the two first comprehensive fit degrees is selected as the optimal reference unit. If the first comprehensive fit degrees of the two units are equal, one unit is randomly selected as the optimal reference unit.

[0030] In response to real-time environmental parameters of the airway clearance scenario, the second comprehensive fit of the optimal reference unit is calculated, and the second comprehensive fit is used as a fine-tuning driving factor to dynamically fine-tune the corrected vibration parameter constraint range.

[0031] Preferably, generating the initial vibration parameter combination of the multi-mode vibration module according to the initial adaptation conditions includes:

[0032] The first feature of the three-dimensional structure of the target region of the airway and the second feature of the secretion distribution in the individualized airway simulation model are extracted. The first feature includes airway anatomical hierarchy, airway wall thickness, and pleural tissue layer thickness. The second feature includes secretion three-dimensional spatial coordinates, viscosity layering, and vertical distance and oblique penetration distance of secretion from the body surface.

[0033] Based on the anatomical landmarks on the patient's body surface, and combined with the vibration attenuation coefficients of multiple tissues such as skin, subcutaneous tissue, muscle, and chest wall between the respiratory target area and the body surface, a three-dimensional targeted mapping model of body surface-airway is constructed. The mapping model generates a layered body surface region that precisely corresponds to the respiratory target area. The layered body surface region includes a core vibration target area that matches the core location of secretions, an auxiliary vibration target area that matches the diffusion range of secretions, and an edge avoidance area for avoiding organs and sensitive tissues.

[0034] Based on the spatial depth of secretions in the respiratory tract, the vibration attenuation characteristics of multi-layered tissues, and the safe distance of thoracic organs, the vibration effect depth range is calculated. The vibration effect depth range includes the minimum effective effect depth to reach the secretion layer, the maximum safe effect depth without damaging thoracic organs and airway walls, and the multi-level targeted effect depth corresponding to the viscosity stratification of secretions.

[0035] Based on the layered surface region and vibration depth range, and combined with the determined initial adaptation conditions including vibration frequency range, vibration intensity range, vibration direction sequence, safety warning threshold, and cleaning efficiency threshold, the depth penetration parameters of the mechanical vibration unit and the surface targeting parameters of the acoustic vibration unit are matched respectively to generate a multi-mode vibration module initial vibration parameter combination that is adapted to the layered surface region and vibration depth range.

[0036] This invention provides a pediatric airway clearing system based on multi-mode vibration and physiological feedback, comprising:

[0037] The initial condition determination module is used to acquire multimodal input data from pediatric patients and input it into a pre-trained neural network model to generate multidimensional physiological representation signals to determine the initial adaptation conditions for airway clearance.

[0038] The simulation module is used to dynamically construct an individualized airway simulation model for a pediatric patient based on the multi-dimensional physiological characterization signals and synchronously acquired real-time medical image streams and respiratory dynamic parameters. The module also controls the multi-mode vibration module to apply vibration to the layered body surface regions corresponding to the individualized airway simulation model according to the initial adaptation conditions. The layered body surface regions include: a core vibration target area, an auxiliary vibration target area, and an edge avoidance area.

[0039] The feedback judgment module is used to collect real-time physiological parameters of pediatric patients during the airway clearance simulation process, and to perform noise reduction and standardization processing on the collected real-time physiological parameters to obtain feedback data for adjustment judgment. The real-time physiological parameters include at least respiratory rate, blood oxygen saturation, airway resistance and breath sound signal.

[0040] The comparative analysis module is used to compare and analyze the feedback data with preset pediatric safety physiological thresholds and secretion clearance judgment thresholds, and adjust the vibration mode and corresponding vibration parameters of the multi-mode vibration module in real time until the airway clearance standard is met, and then perform actual airway clearance on the pediatric patient.

[0041] Compared with the prior art, the beneficial effects of this application are as follows:

[0042] The technical solution of this invention takes into account the physiological tolerance, respiratory tract anatomy and behavioral characteristics of pediatric patients, and realizes the individualized design of vibration parameters. Then, through subsequent simulation and dynamic adjustment, it further effectively improves the safety and efficiency of pediatric airway clearance and reduces the risk of airway tissue damage.

[0043] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0044] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0045] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0046] Figure 1 This is a flowchart of a pediatric airway clearance method based on multi-mode vibration and physiological feedback in an embodiment of the present invention;

[0047] Figure 2 This is a structural diagram of a pediatric airway clearing system based on multi-mode vibration and physiological feedback, as described in an embodiment of the present invention. Detailed Implementation

[0048] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0049] This invention provides a pediatric airway clearance method based on multi-mode vibration and physiological feedback, such as... Figure 1 As shown, it includes:

[0050] Step 1: Acquire multimodal input data from pediatric patients and input it into a pre-trained neural network model to generate multidimensional physiological representation signals to determine the initial adaptation conditions for airway clearance;

[0051] Step 2: Based on the multi-dimensional physiological characterization signals and the synchronously acquired real-time medical image stream and respiratory dynamic parameters, dynamically construct an individualized airway simulation model corresponding to the pediatric patient, and control the multi-mode vibration module to apply vibration to the layered body surface area corresponding to the individualized airway simulation model according to the initial adaptation conditions for simulation. The layered body surface area includes: core vibration target area, auxiliary vibration target area and edge avoidance area.

[0052] Step 3: Real-time physiological parameters of pediatric patients during the airway clearance simulation process are collected, and the collected real-time physiological parameters are denoised and standardized to obtain feedback data for adjustment and judgment. The real-time physiological parameters include at least respiratory rate, blood oxygen saturation, airway resistance and breath sound signal.

[0053] Step 4: Compare and analyze the feedback data with the preset pediatric safety physiological threshold and secretion clearance judgment threshold, and adjust the vibration mode and corresponding vibration parameters of the multi-mode vibration module in real time until the airway clearance standard is met, and then perform actual airway clearance on the pediatric patient.

[0054] Preferably, when feedback data shows that the physiological state of a pediatric patient exceeds a preset pediatric safety physiological threshold, the vibration intensity is reduced and switched to a gentle vibration mode;

[0055] When feedback data shows that the respiratory secretions have not reached the clearance standard, the vibration frequency and direction are adjusted according to the loosening of the secretions until the feedback data shows that the respiratory secretions have reached the clearance standard and the pediatric patient's physiological state is maintained within the preset pediatric safe physiological threshold range. Then, the vibration cleaning stops, and the simulation process of one respiratory clearance is completed.

[0056] In this embodiment, step 1:

[0057] Collect multimodal input data:

[0058] Static baseline physiological parameters: age 2 years, weight 12 kg, developmental stage label (early childhood), baseline respiratory rate 26 breaths / minute, initial airway resistance , viscosity of secretions (Moderate), EEG signals ( Wave power spectral density 0.3 Wave power spectral density 0.5), near-infrared brain functional imaging signal (prefrontal cortex oxygen saturation 92%);

[0059] Behavioral status data: agitation score of 3 (Ramsay scale), cooperation score of 7 (0-10 scale).

[0060] Model feature extraction: The above multimodal input data is input into a pre-trained CNN+LSTM fusion model (training hyperparameters: initial learning rate 0.001, Adam optimizer, MSE loss function, 500 iterations). The model outputs multi-dimensional physiological representation signals (5-dimensional vector): physiological tolerance 0.75 (medium risk), clearance risk 0.4 (medium risk), behavioral adaptability 0.7 (low risk), airway anatomy adaptability 0.65 (medium risk), and secretion feature adaptability 0.55 (medium risk).

[0061] Risk stratification and parameter optimization: Using the K-means clustering algorithm (initialized with the elbow rule of cluster centers, maximum iterations 100), combined with EEG and near-infrared signal characteristics, children were divided into a medium physiological tolerance risk layer, corresponding to preset vibration parameter constraint ranges: frequency 15-25Hz, intensity 3-6. ;

[0062] The local MySQL pediatric airway clearance case knowledge base was used, and a cosine similarity algorithm (similarity ≥ 0.9) was employed to match one historical case (2 years old, moderate secretions, medium risk level) with a similarity of 0.92 to the child's characteristic signal. The optimal vibration parameters for this case were a frequency of 20Hz and an intensity of 4.5. Based on this, the vibration parameter constraint range was optimized to: frequency 18-22Hz, intensity 4-5. ;

[0063] The equipment response parameters were collected as follows: mechanical vibration unit response delay was 8ms, energy conversion efficiency was 0.90, and temperature drift coefficient was 0.04; acoustic vibration unit response delay was 6ms, energy conversion efficiency was 0.88, and temperature drift coefficient was 0.03. A state reliability matrix was constructed, and the first comprehensive fitness was calculated. =0.86、 =0.84, the mechanical vibration element is selected as the optimal reference element;

[0064] Calculate the second overall fitness: First calculate =0.97, then calculate =0.89; with As the driving factor, fine-tune the optimized parameter range (medium risk layer). =0.03, medium viscosity =0.04), yielding the initial adaptation conditions: frequency 17.5-22.6Hz, intensity 3.85-5.22. The initial vibration direction sequence is [45°, 50°, 45°], the safety warning threshold (airway wall stress ≤ 5MPa), and the cleaning efficiency threshold (secretion migration speed ≥ 0.05m / s).

[0065] Step 2: Processing of medical images and respiratory dynamic parameters:

[0066] Real-time CT image streams of the children were acquired simultaneously (10 frames / second), and time alignment (0.1s / frame) and 3D reconstruction were performed using Mimics software. Dynamic geometric features were extracted: airway lumen diameter 5mm, airway wall thickness 1.2mm, secretion volume 2.5mL, distribution density 1.05g / cm³, and airway deformation field during respiratory cycle 0.3mm.

[0067] Respiratory dynamic parameters (50Hz) were collected, and features were extracted using wavelet transform to obtain airway resistance spectrum and respiratory flow waveform, and lung compliance was calculated.

[0068] Finite element simulation framework construction: The extracted dynamic geometric features were used as the initial geometric structure, respiratory dynamic parameters were used as the fluid dynamics input boundary, and the physiological tolerance feature (0.75) in the multi-dimensional physiological characterization signal was used as the airway deformation safety constraint threshold. The finite element simulation framework was constructed using ANSYS software. Material properties for early childhood were substituted: airway wall elastic modulus 9 MPa, secretion viscosity 400. Vibration attenuation coefficients of multiple tissues (skin 0.15, subcutaneous fat 0.20, chest wall muscle 0.30, pleura 0.05, chest wall 0.10).

[0069] Model updates and simulations:

[0070] Based on the Demons non-rigid image registration algorithm (iteration step size 0.2, convergence threshold 0.005), the airway deformation field is inverted, the geometric structure and material properties of the simulation framework are updated in real time, and the fluid dynamic boundary conditions are corrected by combining the changes in respiratory dynamic parameters to obtain an individualized airway simulation model.

[0071] Based on this model, a layered body surface region is generated (core vibration target area: 5th intercostal space along the midclavicular line, auxiliary vibration target area: 4th intercostal space along the anterior axillary line, and edge avoidance area: midline of the sternum and both sides of the spine), and the vibration depth range is calculated (3.5-4.5cm for moderately viscous secretions).

[0072] The multi-mode vibration control module outputs an initial vibration parameter combination (frequency 20Hz, intensity 4.5) according to the initial adaptation conditions. The mechanical vibration unit rotates at 1200 r / min, the acoustic vibration unit resonates at 1000 Hz, the core target area uses a composite vibration mode, and the auxiliary target area uses an acoustic vibration mode. Vibration is applied and simulation response data is collected: airway wall stress 3.2 MPa, secretion displacement 0.4 mm, airway pressure change 1.2 kPa, vibration energy attenuation coefficient 0.3.

[0073] The simulation response data was compared with the preset thresholds (secretion migration velocity ≥ 0.05 m / s, displacement ≥ 0.3 mm, airway wall stress ≤ 5 MPa), and all requirements were met. The vibration parameter combination and the layered body surface area were output as the initial vibration control scheme.

[0074] Step 3:

[0075] Real-time physiological parameter acquisition: The feedback judgment module is activated, and real-time physiological parameters of the child are acquired at a frequency of 100Hz through the chest wall motion sensor, pulse oximeter, pulmonary function tester, and surface auscultation sensor: respiratory rate 25 breaths / minute, blood oxygen saturation 96%, and airway resistance 1.8. The characteristic value of breath sounds with wet rales is 0.8 (initial value 1.0).

[0076] Data denoising and standardization:

[0077] Breath sound signals were decomposed and denoised using 5-layer decomposition of db5 wavelet to remove high-frequency noise (800-2000Hz) and retain effective wet rales.

[0078] Airway resistance was reduced using moving average noise reduction (window size 5), and respiratory rate and blood oxygen saturation were reduced using median filtering noise reduction (window size 3).

[0079] All processed physiological parameters were Z-score standardized to obtain the following feedback data: respiratory rate standardized value 0.2, blood oxygen saturation standardized value 0.3, airway resistance standardized value 0.4, and wet rales characteristic value standardized value 0.2.

[0080] Step 4:

[0081] Comparative analysis: The feedback data was compared with the preset thresholds, and the results are as follows: the physiological parameters (respiratory rate 25 breaths / minute, blood oxygen saturation 96%) were all within the safe threshold range; the airway resistance decreased by 10% (not reaching ≥30%) compared with the initial value, the characteristic value of moist rales decreased by 20% (not reaching ≥50%), the secretions did not meet the clearance standard, and the child did not show increased agitation or signs of airway damage.

[0082] Parameter adjustment: Based on the loosening of secretions (moderate loosening: 20% decrease in wet rales + 10% decrease in resistance), vibration parameters were adjusted using the gradient descent method (learning rate 0.02): frequency was adjusted to 21Hz (single adjustment 1Hz≤2Hz), and intensity was adjusted to 4.7. (Single adjustment 0.2) The vibration direction is adjusted to [48°, 52°, 48°] (single adjustment ≤ 10°) to maintain the composite vibration mode.

[0083] Continuous adjustment and actual clearance: Repeat steps 3-4, collect feedback data and adjust parameters every 10 seconds. After 3 minutes, the feedback data shows: respiratory rate 24 breaths / minute, blood oxygen saturation 97%, airway resistance decreased by 35% (reaching the clearance threshold), and the characteristic value of wet rales decreased by 55% (reaching the clearance threshold), meeting the airway clearance standard (physiological parameters meet the standard + secretion clearance meets the standard).

[0084] Cleaning complete: Control the multi-mode vibration module to stop vibrating, and record the cleaning parameters (frequency 21.5Hz, intensity 4.8). (Vibration duration 3 minutes), add this case to the local case knowledge base, and complete this pediatric airway cleanup.

[0085] In this embodiment, no adverse events such as airway damage or increased agitation occurred during the airway clearance process for the child. After clearance, the airway resistance decreased from 2.0. Reduced to 1.3 (Decrease of 35%), moist rales basically disappeared, blood oxygen saturation was maintained at 97%-98%, respiratory rate returned to the normal range (22-30 breaths / minute), clearance efficiency was improved by 40% compared with traditional mechanical sputum removal, and the child's cooperation was not significantly reduced.

[0086] In this embodiment, the pre-trained neural network model adopts a fusion model of CNN (Convolutional Neural Network) + LSTM (Long Short-Term Memory Network). The training data is: a clinical dataset of pediatric airway clearance aged 0-6 years, with a sample size of ≥1000 cases, covering the stages of infancy (0-1 years), early childhood (1-3 years), and preschool age (3-6 years). It includes data of children with different secretion viscosity (mild / moderate / severe), respiratory disease type (bronchitis / pneumonia / laryngostomia), and behavioral status (cooperative / mild agitation / severe agitation). Each data point contains complete multimodal input data and corresponding vibration parameters, clearance effect, and physiological response annotations. The annotations are jointly completed by two or more associate chief physicians of pediatric respiratory department.

[0087] Multidimensional physiological representation signals include features such as physiological tolerance, clearance risk, and behavioral adaptability. Each dimension is independent of the others and can comprehensively reflect the child's overall adaptability to airway clearance procedures.

[0088] In this embodiment, the initial adaptation conditions for airway clearance are determined by multi-dimensional physiological characterization signals through preset mapping rules. The mapping rules are the correspondence between feature signals trained based on clinical data and the initial adaptation conditions, and are stored in the model's database.

[0089] The ultrasound imaging equipment acquires dynamic images of the child's airway and chest cavity at a rate of 10 frames per second, forming a real-time medical image stream. The image data is transmitted to the simulation modeling module after being standardized in DICOM format. The respiratory dynamic parameters include airway resistance, respiratory flow, lung compliance, airway pressure, etc. The acquisition frequency is synchronized with the medical image stream at 50Hz.

[0090] A personalized respiratory tract simulation model was constructed using the finite element method. The geometric structure and mechanical parameters of the model were highly matched with the respiratory tract characteristics of the child.

[0091] The multi-mode vibration module includes a mechanical vibration unit (driven by an eccentric wheel motor to achieve low-frequency deep vibration) and an acoustic vibration unit (driven by a piezoelectric ceramic sheet to achieve high-frequency surface vibration). The vibration parameters of the module are adjusted in real time by the control program of the host computer, and the vibration acts on the surface of the child's body.

[0092] First, multi-dimensional physiological characterization signals, real-time medical image streams, and respiratory dynamics parameters are simultaneously imported into ANSYS finite element simulation software. This software performs three-dimensional reconstruction of the image stream, extracts features from the respiratory dynamics parameters, and constructs an individualized respiratory tract simulation model by combining the physiological tolerance features in the multi-dimensional physiological characterization signals. Subsequently, the host computer's control program generates vibration control commands based on the initial adaptation conditions and sends them to the multi-mode vibration module. This module controls the application of vibration to the layered body surface areas corresponding to the individualized respiratory tract simulation model. The vibration effect is then simulated in the simulation software, and the model's response data is recorded in real time during the simulation.

[0093] Respiratory rate was collected using a chest wall motion sensor, blood oxygen saturation was collected using a pulse oximeter, airway resistance was collected using a pulmonary function testing device, and breath sound signals were collected using a surface auscultation sensor. All data were collected at a frequency of 100Hz. Wavelet transform denoising algorithm was used for breath sound signals, moving average denoising algorithm was used for airway resistance parameters, and median filtering denoising algorithm was used for respiratory rate and blood oxygen saturation. All denoising algorithms were implemented through a signal processing program on a host computer.

[0094] Standardization processing refers to the process of converting real-time physiological parameters with different dimensions and numerical ranges after noise reduction into standardized data with unified dimensions and numerical ranges, eliminating the dimensional differences between parameters and facilitating subsequent comparative analysis. This is an existing technology.

[0095] The feedback data is a dataset containing standardized respiratory rate, blood oxygen saturation, airway resistance, and respiratory sound signal feature values. The respiratory sound signal feature values ​​are extracted using Mel-frequency cepstral coefficients and combined with other parameters to form unified feedback data. Specifically, during the simulation, four core real-time physiological parameters of the child are collected synchronously by various medical sensors and devices. The collected raw parameter data is transmitted to the signal processing module of the host computer. This module sequentially calls the noise reduction algorithm and the Z-score normalization algorithm to process the raw data, remove noise, and achieve standardization, ultimately forming feedback data. The feedback data is transmitted to the comparison and analysis module in real time at a frequency of 50Hz.

[0096] Preset pediatric safety physiological thresholds refer to critical values ​​of physiological parameters established based on pediatric medical guidelines and clinical data to ensure the physiological safety of children. These thresholds serve as the basis for determining whether a child's physiological state is safe. They are categorized according to the child's age group and stored in a database on a host computer. The specific values ​​for the preset pediatric safety physiological thresholds are: Infancy (0-1 year): respiratory rate 30-40 breaths / minute, blood oxygen saturation ≥95%, and airway resistance... During early childhood (1-3 years old), the respiratory rate should be 25-30 breaths / minute, blood oxygen saturation ≥95%, and airway resistance should be [not specified]. There were no abnormally sharp wheezing sounds in the breath sound signal.

[0097] The secretion clearance judgment threshold refers to the critical value established based on clinical data to judge the effectiveness of airway secretion clearance. It is the basis for judging whether secretions have reached the clearance standard. In practice, it is defined by indicators such as respiratory sound signal characteristics and airway resistance change rate. For example, the secretion clearance judgment threshold is: airway resistance decreases by ≥30% from the initial value, the wet rales characteristic value in the respiratory sound signal decreases by ≥50%, and there are no obvious respiratory sounds related to secretion accumulation.

[0098] The airway clearance standard refers to the state standard that simultaneously meets the preset pediatric safety physiological threshold and the secretion clearance judgment threshold. It is the basis for determining whether actual airway clearance can be performed. For example, when the feedback data shows that the child's various physiological parameters are all within the safety threshold range and the secretion-related indicators reach the clearance judgment threshold, the airway clearance standard is met.

[0099] Vibration mode refers to the operating mode of the multi-mode vibration module, including mechanical vibration mode, acoustic vibration mode, and mechanical + acoustic composite vibration mode. Vibration parameters refer to specific indicators characterizing vibration features; for example, vibration parameters include vibration frequency (Hz), vibration intensity, vibration direction, and vibration duration. If feedback data and preset dual thresholds are imported into a comparison and analysis program on a host computer, the program performs real-time comparison and analysis of the feedback data and thresholds. If the feedback data does not meet the airway clearance criteria, the vibration mode and parameters of the multi-mode vibration module are adjusted in real-time according to the direction of deviation. For example, if secretions have not reached the clearance threshold and the child's physiological state is safe, the frequency of mechanical vibration is adjusted from 20Hz to 25Hz, and the intensity is adjusted from... Adjust to 6 If the child's physiological parameters are close to the safety threshold, immediately switch to a gentle sound wave vibration mode to reduce the vibration intensity. When the comparative analysis results show that the feedback data meets the airway clearance criteria, stop the parameter adjustment in the simulation, and determine the vibration mode and parameters at this time as the execution parameters for actual clearance, controlling the multi-mode vibration module to perform actual airway clearance on the child.

[0100] In this embodiment, the gentle vibration mode refers to the low-intensity, low-frequency vibration mode of the multi-mode vibration module, which is suitable for gentle cleaning when the child's physiological state is close to or exceeds the safety threshold. In practice, a preset dedicated vibration mode is used. For example, the gentle vibration mode is a single sound wave vibration with a frequency of 500Hz and an intensity of 1... The vibration direction is parallel to the body surface to avoid strong impact on airway tissues.

[0101] The degree of loosening of secretions refers to the extent to which respiratory secretions are loosened under vibration, as judged by indicators such as the characteristics of breath sound signals and changes in airway resistance in the feedback data. It serves as the basis for adjusting the vibration frequency and direction. In practice, it is judged by changes in the characteristic value of wet rales in the breath sound signal and the rate of decrease in airway resistance. For example, a 20% decrease in the characteristic value of wet rales and a 10% decrease in airway resistance are judged as mild loosening of secretions; a 40% decrease in the characteristic value of wet rales and a 20% decrease in airway resistance are judged as moderate loosening of secretions.

[0102] The beneficial effects of the above technical solution are: by taking into account the physiological tolerance, respiratory tract anatomical features and behavioral characteristics of pediatric patients, the individualized design of vibration parameters is achieved, and then the safety and efficiency of pediatric airway clearance are further effectively improved through subsequent simulation and dynamic adjustment, reducing the risk of airway tissue damage.

[0103] This invention provides a pediatric airway clearance method based on multi-mode vibration and physiological feedback, which determines the initial adaptation conditions for airway clearance, including:

[0104] The multimodal input data is input into a pre-trained neural network model to generate a multi-dimensional physiological representation signal that includes physiological tolerance, cleanup risk and behavioral adaptability. The multi-dimensional physiological representation signal is a high-dimensional vector, and each dimension corresponds to a feature weight and a risk level label.

[0105] Based on the multi-dimensional physiological characterization signals, the K-means clustering algorithm is used to perform risk stratification, dividing pediatric patients into three physiological tolerance risk layers: low, medium, and high. Each risk layer corresponds to a preset vibration parameter constraint range.

[0106] The local pediatric airway clearance case knowledge base is invoked to match the historical cases with the highest similarity to the current patient's characteristic signals, and the vibration parameter constraint range is optimized and corrected.

[0107] Based on real-time environmental parameters and equipment response parameters for airway clearance, the corrected vibration parameter constraint range is dynamically fine-tuned to obtain initial adaptation conditions. These initial adaptation conditions include: initial range of vibration frequency, initial range of vibration intensity, initial sequence of vibration direction, and corresponding safety warning threshold and clearance efficiency threshold.

[0108] Preferably, the multimodal input data includes at least: static basic physiological parameters: age, weight, developmental stage label, basic respiratory rate, estimated viscosity of respiratory secretions, electroencephalogram (EEG) signals, and near-infrared brain functional imaging signals;

[0109] Behavioral status data: agitation score, cooperation score.

[0110] A high-dimensional vector refers to a vector with more than 3 dimensions. It is a mathematical expression of multi-dimensional physiological representation signals. In practice, multi-dimensional physiological representation signals are converted into N-dimensional vectors (N≥5). Each dimension of the vector corresponds to a specific feature dimension. For example, physiological tolerance, clearance risk, behavioral adaptability, airway anatomical adaptability, and secretion feature adaptability are taken as 5 dimensions to form a 5-dimensional high-dimensional vector. Each element of the vector is the feature quantization value of the corresponding dimension.

[0111] Feature weights refer to the weight values ​​corresponding to each dimension in a high-dimensional vector, reflecting the degree of influence of the feature of that dimension on the initial adaptation conditions of airway clearance. In practice, feature weights are determined by the analytic hierarchy process (AHP). For example, for young infants, the feature weight of physiological tolerance is set to 0.4, the feature weight of clearance risk is set to 0.3, and the feature weight of behavioral adaptability is set to 0.3, with the sum of the weight values ​​being 1.

[0112] Risk level labels refer to labels that annotate each dimension of a high-dimensional vector, reflecting the degree of risk of that dimension's characteristics. In practice, they are divided into three levels: low, medium, and high. For example, a quantitative value of 0.8 for the physiological tolerance dimension is labeled as low risk, 0.5-0.8 as medium risk, and ≤0.5 as high risk; a quantitative value of 0.2 for the cleanup risk dimension is labeled as high risk, 0.2-0.5 as medium risk, and ≥0.5 as low risk.

[0113] In practice, the K-means clustering algorithm was used to perform cluster analysis on the high-dimensional vectors, dividing the children into three physiological tolerance risk layers: low, medium, and high. For example, the low-risk layer consists of children with high physiological tolerance, low clearance risk, and high behavioral cooperation; the medium-risk layer consists of children with moderate levels of various characteristics; and the high-risk layer consists of children with low physiological tolerance, high clearance risk, and low behavioral cooperation.

[0114] The vibration parameter constraint range refers to the preset range of vibration parameter values ​​for each physiological tolerance risk level. It is the initial constraint boundary of the vibration parameters. In implementation, different ranges are divided according to the risk level. For example, the vibration frequency constraint range for the low-risk level is 20-30Hz, and the vibration intensity is... The medium-risk layer is 15-25Hz, with an intensity of The high-risk layer is 10-20Hz, with an intensity of .

[0115] The local pediatric airway clearance case knowledge base refers to a database stored on a local server containing a large number of pediatric airway clearance clinical cases. It serves as a reference for parameter optimization. Each case in the knowledge base contains multi-dimensional physiological characteristics, vibration parameters, clearance effects, and physiological response data of the child. In implementation, a relational database (MySQL) is used to build this knowledge base, which supports rapid retrieval and matching of cases.

[0116] The historical case with the highest similarity to the current patient's multidimensional physiological representation signals refers to the clinical case retrieved from the case knowledge base that has the highest similarity to the current patient's signals. In practice, a cosine similarity algorithm is used to calculate the similarity between the current patient's signals and the case signals in the knowledge base. The calculation formula is as follows: Where A is the high-dimensional vector of the current patient and B is the high-dimensional vector of the case. The closer the similarity value is to 1, the higher the similarity. For example, cases with a cosine similarity ≥ 0.9 are retrieved from the knowledge base as the historical cases with the highest similarity.

[0117] Real-time environmental parameters refer to the on-site environmental parameters when the child's airway is cleared. These parameters affect the transmission of vibration energy and the child's physiological state. They are collected by environmental sensors during implementation. For example, real-time environmental parameters include the ambient temperature (°C), relative humidity (%), and air flow velocity (m / s) of the ward.

[0118] Equipment response parameters refer to the state parameters of the multi-mode vibration module during operation, reflecting the working stability of the equipment. In practice, they are collected through the module's built-in sensors. For example, equipment response parameters include the vibration unit's response delay (ms), energy conversion efficiency (%), temperature drift coefficient, etc.

[0119] In this embodiment, age and weight are collected using an electronic scale and a height and weight meter; baseline respiratory rate is collected using a chest wall motion sensor; developmental stage labels are marked according to age: infancy (0-1 year), toddlerhood (1-3 years), and preschool age (3-6 years); the estimated viscosity of respiratory secretions is determined by a combination of clinical examination and equipment analysis; electroencephalogram (EEG) signals and near-infrared brain function imaging signals are collected using a dedicated portable device, and all parameters are transmitted to a data acquisition terminal for integration after collection.

[0120] Agitation rating is a quantitative assessment of a child's agitation level using a standardized rating scale. It reflects the child's behavioral state. The Ramsay Agitation Rating Scale, specifically designed for pediatrics, is used in this assessment. This scale divides agitation into 1-6 points. For example, 1 indicates lethargy and no response to stimuli; 2 indicates wakefulness and quiet cooperation; 3 indicates wakefulness and mild agitation; 4 indicates moderate agitation and limb movement; 5 indicates severe agitation and violent struggle; and 6 indicates mania and inability to control. The rating is determined by medical staff based on the child's actual behavior.

[0121] The cooperation score refers to the quantitative assessment of a child's cooperation level during the procedure using a self-made pediatric airway clearance cooperation score scale. The scale is scored from 0 to 10 points, assessing the child's physical cooperation, emotional state, and response to instructions. Each dimension is scored from 0 to 3 points, and the scores are summed to obtain the final score. For example, a child who shows no physical resistance, is calm, and responds to simple instructions receives a score of 8; a child who shows strong physical resistance, cries, and ignores instructions receives a score of 2. The score is assessed by medical staff before the procedure.

[0122] The beneficial effects of the above technical solution are as follows: the process of determining the initial adaptation conditions has been meticulously designed; multi-dimensional physiological representation signals are quantified by high-dimensional vector quantization; scientific risk stratification is achieved by combining cluster analysis; parameters are optimized and corrected by utilizing a clinical case knowledge base; and environmental and equipment parameters are dynamically fine-tuned, making the initial adaptation conditions more closely match the individual characteristics of the children and the actual operating scenarios. This effectively improves the accuracy and rationality of the initial adaptation conditions, laying a good parameter foundation for subsequent simulation and actual cleaning.

[0123] This invention provides a pediatric airway clearance method based on multi-mode vibration and physiological feedback, which dynamically constructs an individualized airway simulation model corresponding to the pediatric patient, including:

[0124] The real-time medical image stream is time-aligned and reconstructed in three dimensions to extract dynamic geometric features of the airway, including airway lumen diameter, airway wall thickness, volume and distribution density of secretions, and airway deformation field during the respiratory cycle. At the same time, the respiratory dynamic parameters are feature extracted to obtain airway resistance spectrum, respiratory flow waveform and compliance change curve.

[0125] The extracted dynamic geometric features are used as the initial geometric structure of the simulation model, and the feature extraction results of respiratory dynamic parameters are used as the input boundary of the fluid dynamics simulation. In addition, the tolerance features in the multi-dimensional physiological characterization signals are used as the safety constraint threshold for airway deformation, limiting the upper limit of airway wall stress and secretion migration speed during the simulation process, thus constructing a finite element simulation framework.

[0126] The subsequent frame sequence of the real-time synchronized medical image stream is used to invert the airway deformation field based on the non-rigid image registration algorithm, update the geometric structure and material properties of the finite element simulation framework, and modify the fluid dynamic boundary conditions in combination with the real-time changes of respiratory dynamic parameters to obtain an individualized airway simulation model.

[0127] In this embodiment, the timing alignment uses a timestamp matching method to add an acquisition timestamp to each frame of image. The image stream is sorted and calibrated according to the order of the timestamps, and the image stream with uneven acquisition time intervals is calibrated into an ordered image sequence with equal time intervals (0.1s / frame).

[0128] Mimics medical imaging 3D reconstruction software was used. This software can segment, register and reconstruct 2D images such as CT and ultrasound. For example, the airway ultrasound image stream of the child was imported into Mimics software, the contour of the airway tissue was extracted by threshold segmentation, and then 3D reconstruction was performed to obtain the 3D geometric model of the airway.

[0129] Dynamic geometric features include airway lumen diameter (variation with the respiratory cycle, mm), airway wall thickness (mm), secretion volume (mL), and distribution density (…). ), airway deformation field during the respiratory cycle (deformation of various parts of the airway, mm).

[0130] The airway deformation field refers to the field distribution formed by the deformation and direction of various anatomical parts of the airway during the inspiratory and expiratory phases of the respiratory cycle. It reflects the dynamic deformation characteristics of the airway. In practice, it is obtained by comparing and analyzing three-dimensional reconstruction models at different stages of the respiratory cycle. For example, the deformation of the airway lumen diameter during inhalation is 2 mm, and the deformation during exhalation is 1.5 mm. The deformation and direction of each part are mapped onto the three-dimensional model to form the airway deformation field.

[0131] Signal processing and feature extraction algorithms are used, such as wavelet transform, to extract the time-domain and frequency-domain features of respiratory dynamic parameters, and obtain the corresponding feature curves and values.

[0132] Airway resistance spectrum refers to the spectral curve of airway resistance as a function of respiratory rate and airflow velocity, reflecting the resistance characteristics of the airway under different respiratory states. In practice, the raw airway resistance data collected by the pulmonary function testing instrument is converted into a frequency domain resistance spectrum curve by Fourier transform. For example, the airway resistance spectrum shows that the airway resistance is 2.0 kPa・s / L at a respiratory rate of 20 breaths / minute.

[0133] The respiratory flow waveform refers to the curve showing the change in gas flow rate in the airway over time during respiration, reflecting the flow characteristics of respiration. In practice, it is directly collected and generated by a pulmonary function testing instrument. For example, the flow rate shows an upward trend during the inspiratory phase, with a peak value of 0.5 L / s, and a downward trend during the expiratory phase, with a peak value of -0.4 L / s, forming a symmetrical waveform curve.

[0134] The compliance curve refers to the curve showing how the compliance of the lungs and airways changes with the respiratory cycle. Compliance reflects the expandability of the airways and lung tissue, and is calculated in practice through the relationship between respiratory pressure and volume. The calculation formula is as follows: ,in, For compliance, For volume change, For pressure changes, plot the compliance change curve over time based on the calculation results.

[0135] In this embodiment, ANSYS finite element simulation software was used to construct the finite element simulation framework, which includes elements such as the geometric mesh of the respiratory tract, fluid dynamics boundaries, material properties, and safety constraint thresholds.

[0136] The preferred non-rigid image registration algorithm is the Demons algorithm, with an iteration step size of 0.2, a convergence threshold of 0.005, a pixel grayscale matching weight of 0.7, and a spatial smoothing weight of 0.3. Considering the characteristics of the pediatric respiratory tract, such as narrow airway diameter and rapid deformation, the pixel search range for image registration is limited to a 5×5 neighborhood to improve registration speed and accuracy. A supplementary algorithm execution flowchart is provided to clarify the steps of image frame input → grayscale matching → deformation field calculation → airway structure update → convergence determination.

[0137] The input boundary of fluid dynamics simulation refers to using the feature extraction results of respiratory dynamics parameters as the input conditions of fluid dynamics simulation. These include parameters such as pressure, flow rate, and velocity at the boundary, which serve as the basis for calculating gas flow in the simulation model. For example, the respiratory flow waveform and airway resistance spectrum can be used as the inlet and outlet boundary conditions of fluid dynamics simulation.

[0138] In this embodiment, the material properties of respiratory tract tissues in infancy / early childhood / preschool age and the vibration attenuation coefficient of multilayer tissues were determined by vibration experiments using pediatric ex vivo tissue specimens and biomimetic models. The experimental sample size was ≥30 cases per age group, as shown in Table 1.

[0139] Table 1 Parameter Quantization Table

[0140]

[0141] Fluid dynamics boundary conditions refer to the conditions that limit the flow characteristics of fluid (gas) in the simulation model, including inlet conditions, outlet conditions, wall conditions, etc. During implementation, they are dynamically corrected according to the real-time changes of respiratory dynamic parameters. For example, during the inhalation phase, the flow rate at the inlet boundary is set to a positive peak value, and during the exhalation phase, it is set to a negative peak value.

[0142] The beneficial effects of the above technical solution are as follows: by constructing an individualized airway simulation model, the dynamic geometric and hydrodynamic characteristics of the airway of pediatric patients are accurately restored; the accuracy of the geometric structure is ensured by temporal alignment and three-dimensional reconstruction; the real-time dynamic updating of the model is achieved by using a non-rigid image registration algorithm; and physiological tolerance is incorporated into the framework as a safety constraint threshold, making the model more in line with the physiological characteristics of pediatric patients, providing an accurate virtual environment for subsequent vibration simulation pre-runs, and effectively improving the reliability and predictability of the simulation.

[0143] This invention provides a pediatric airway clearance method based on multi-mode vibration and physiological feedback. The method involves controlling a multi-mode vibration module to apply vibration to the layered surface regions corresponding to the individualized airway simulation model according to the initial adaptation conditions, and performing simulation. The method includes:

[0144] Based on the three-dimensional structure and secretion distribution of the individualized respiratory tract simulation model, a layered body surface region and vibration depth range corresponding to the respiratory tract target region are generated, and the initial vibration parameter combination of the multi-mode vibration module is generated according to the initial adaptation conditions.

[0145] The control multi-mode vibration module applies vibration to the layered body surface region according to the initial vibration parameter combination, and at the same time collects the simulation response data of the individualized respiratory tract simulation model in real time, including airway wall stress distribution, secretion displacement field, airway pressure change and vibration energy attenuation coefficient in respiratory tissue;

[0146] The simulation response data is compared with the preset vibration efficiency threshold and safety constraint threshold. If the simulation response data does not meet the standards, the parameter combination of the multi-mode vibration module is dynamically adjusted based on the optimization direction of the vibration parameters retrieved from the individualized airway simulation model, and the simulation is repeated until the simulation response data meets the preset vibration efficiency and safety constraint conditions. The optimized multi-mode vibration parameter combination and the corresponding layered body surface area are then output as the initial vibration control scheme for subsequent actual airway cleaning.

[0147] In this embodiment, the vibration depth range refers to the depth range at which the vibration energy of the multi-mode vibration module can effectively act on the respiratory tract. It is determined based on the spatial distribution of secretions. In practice, it is calculated by combining the distance from the body surface to the secretions and the tissue vibration attenuation characteristics. For example, if the secretions are located 3-5cm below the body surface, the vibration depth range is set to 3-5cm to ensure that the vibration energy can reach the secretion layer.

[0148] The initial vibration parameter combination refers to the initial working parameter combination of the multi-mode vibration module determined based on the initial adaptation conditions, the layered surface area, and the vibration depth range. It includes the vibration mode and specific vibration parameters. For example, the initial vibration parameter combination is a mechanical + acoustic composite vibration mode, with a mechanical vibration frequency of 20Hz and an intensity of 5. The sound wave vibration frequency is 1000Hz and the intensity is 2. The vibration direction is at a 45° angle to the body surface.

[0149] Simulation response data refers to the parameter data output by the individualized airway simulation model after it is subjected to vibration during the simulation process, which can reflect the vibration effect and safety status. It is the basis for judging whether the vibration parameters meet the standards. For example, simulation response data includes airway wall stress distribution, secretion displacement field, airway pressure change, and vibration energy attenuation coefficient in airway tissue.

[0150] The stress distribution of the airway wall refers to the magnitude and distribution of stress on various parts of the airway wall under vibration. It reflects the degree of impact of vibration on the airway wall. In practice, it is calculated by finite element simulation software. For example, the stress in the core vibration zone of the airway wall under vibration is 3MPa, and the stress in the auxiliary vibration zone is 1.5MPa, both of which are within the safe threshold range.

[0151] The secretion displacement field refers to the field distribution formed by the displacement amount and direction of various parts of respiratory secretions under vibration. It reflects the loosening and migration effect of secretions. In practice, it is calculated by finite element simulation software. For example, the displacement of the core area of ​​secretions under vibration is 0.5 mm, showing a migration trend towards the airway outlet.

[0152] The vibration energy attenuation coefficient in respiratory tissues refers to the proportion of vibration energy attenuated when passing through tissues such as skin, subcutaneous tissue, chest wall, and airway wall. It reflects the effective transmission degree of vibration energy. In practice, it is determined by in vitro tissue model experiments and substituted into the simulation model. For example, the vibration energy attenuation coefficient of skin is 0.2, that is, muscle tissue is 0.3, and chest wall is 0.1. That is, vibration energy is attenuated according to the corresponding coefficient for each layer of tissue it passes through.

[0153] The vibration efficiency threshold is a preset critical value that can determine the efficiency of vibration in cleaning secretions. It is the basis for judging whether vibration parameters have an effective cleaning effect. In practice, it is defined by indicators such as secretion displacement field and migration speed. For example, the vibration efficiency threshold is secretion migration speed ≥ 0.05m / s and displacement ≥ 0.3mm.

[0154] The safety constraint threshold refers to a preset critical value that can determine whether vibration will cause damage to respiratory tissues. It is the basis for judging whether vibration parameters are safe. In practice, it is defined by indicators such as airway wall stress and tissue deformation. For example, the safety constraint threshold is airway wall stress ≤ 5MPa and airway wall deformation ≤ 1mm.

[0155] The optimization direction of the inverted vibration parameters refers to deriving the adjustment direction of the vibration parameters in reverse based on the deviation between the simulated response data and the threshold. In practice, stochastic gradient descent (SGD) is used. Considering the sensitivity of pediatric vibration parameters, an upper limit is set for the parameter adjustment step size (frequency ≤ 2 Hz / cycle, intensity ≤ 0.5). To avoid physiological damage caused by parameter mutations, for example, if the secretion migration speed does not reach the efficiency threshold, the inversion result is to increase the vibration frequency and intensity; if the airway wall stress is close to the safety threshold, the inversion result is to reduce the vibration intensity.

[0156] The initial vibration control scheme refers to the vibration control plan output after the simulation meets the target, which is used for subsequent actual airway clearance. It includes the defined layered body surface areas, vibration modes, vibration parameters, and execution procedures. For example, the initial vibration control scheme uses a combination of mechanical and acoustic vibration in the core vibration target area, with a frequency of 22Hz and an intensity of 5.5. The auxiliary vibration target area uses acoustic vibration at a frequency of 800Hz and an intensity of 1.5. The vibration lasts for 5 seconds per cycle, with a 2-second interval.

[0157] The beneficial effects of the above technical solution are as follows: it realizes the pre-simulation and iterative optimization of vibration parameters based on the simulation process; improves the accuracy of vibration through the design of targeted layered body surface areas and vibration depth range; realizes the scientific optimization of parameters through the collection and comparison of simulation response data; avoids invalid and dangerous vibrations; provides a proven, safe and effective initial vibration control scheme for actual airway cleaning; effectively reduces safety risks in actual operation; and improves cleaning efficiency.

[0158] This invention provides a pediatric airway clearance method based on multi-mode vibration and physiological feedback. Based on real-time environmental parameters and equipment response parameters during airway clearance, the method dynamically fine-tunes the corrected vibration parameter constraint range, including:

[0159] The device response parameters of the mechanical vibration unit and the acoustic vibration unit in the multi-mode vibration module are collected, and a state reliability matrix is ​​constructed, wherein the state reliability matrix contains only two off-diagonal elements;

[0160] Based on the state confidence matrix, the first comprehensive fit degree of the mechanical vibration unit and the acoustic vibration unit are calculated respectively, and the unit corresponding to the maximum value of the two first comprehensive fit degrees is selected as the optimal reference unit. If the first comprehensive fit degrees of the two units are equal, one unit is randomly selected as the optimal reference unit.

[0161] In response to real-time environmental parameters of the airway clearance scenario, the second comprehensive fit of the optimal reference unit is calculated, and the second comprehensive fit is used as a fine-tuning driving factor to dynamically fine-tune the corrected vibration parameter constraint range.

[0162] In this embodiment, the device response parameters include the response delay of each unit. (Unit: ms), Vibration output noise power spectral density N (Unit: dB / Hz), Energy conversion efficiency (Dimensionless, value range 0~1), Phase shift (Unit: °C) Temperature drift coefficient (Dimensionless, representing the drift ratio relative to the module's reference operating temperature);

[0163] In this embodiment, the two off-diagonal elements are , ,in, The degree of cooperative state correlation between mechanical vibration unit M and acoustic vibration unit S; The degree of cooperative state correlation between acoustic vibration unit S and mechanical vibration unit M;

[0164] The formula for calculating matrix elements is: , dimensionless, taking values ​​from 0 to 1, where, For the credibility of unit u itself, , The energy conversion efficiency of unit u; The temperature drift coefficient of unit u. The larger the value, the more stable the unit's own working state;

[0165] The cooperative delay between element u and element v. , The baseline response delay calibrated at the module's factory settings; For reference phase offset; The reference noise power spectral density; The larger the value, the worse the synchronization between the two units;

[0166] In this embodiment, the vibration energy transfer efficiency It is a dimensionless physical quantity, and its value range is usually 0.6~0.95. Its magnitude is mainly affected by the vibration attenuation of chest wall tissue and the fit between the multi-mode vibration module and the body surface. In pediatric and medical external vibration sputum clearance related equipment, fixed empirical values ​​(such as 0.75~0.85) are often used.

[0167] In this embodiment, the first comprehensive fit is calculated as follows:

[0168] ;

[0169] ;

[0170] in, , and The first preset dimensionless coefficient is calibrated by the module manufacturer. Used to weigh the impact of the unit's own credibility. Used to weigh the impact of the degree of synergy between two units, such as ;

[0171] , All are dimensionless parameters, with values ​​ranging from 0 to 1. The larger the value, the better the overall working state of the corresponding unit and its cooperative state with another unit, and the more suitable it is as a reference unit for subsequent fine-tuning of vibration parameters. They are the first comprehensive fit degree of mechanical vibration unit M and acoustic vibration unit S, respectively.

[0172] In this embodiment, real-time environmental parameters include relative humidity H (dimensionless, ranging from 0 to 1), ambient temperature T (unit: °C), and air velocity. (Unit: m / s);

[0173] In this embodiment, the second comprehensive fit The calculation formula is as follows:

[0174] ,in, , , The second preset dimensionless coefficient is calibrated based on the standard environmental conditions of the pediatric ward. Weigh the impact of the equipment's own adaptability, and weigh the impact of environmental parameters, such as... , ;

[0175] (Dimensionless, ranging from 0 to +∞), is an environment adaptation correction term. =0.5 (50%) =24℃ =0.2m / s is the standard environmental reference parameter for pediatric airway cleanup. The closer it is to 1, the more suitable the environmental conditions are for the transmission of vibrational energy.

[0176] In this embodiment, the specific process of dynamic fine-tuning is as follows:

[0177] The original frequency constraint range (Unit: Hz) Fine-tuning The calculation formula is:

[0178] ,in, The frequency fine-tuning coefficient (dimensionless) is negatively correlated with the physiological tolerance risk level of pediatric patients, with a low-risk level. =0.05, medium risk level =0.03, high-risk level =0.01;

[0179] The original strength constraint range (unit: Fine-tuning to The calculation formula is:

[0180] ,in, The intensity fine-tuning coefficient (dimensionless) is positively correlated with the viscosity of respiratory secretions, and its value ranges from 0.02 to 0.06.

[0181] The initial sequence of the original vibration direction (Unit: °) Fine-tuning to The calculation formula is: ,in, The phase offset of the optimal reference cell (unit: °). This is a sign function used to correct vibration direction deviations caused by phase shifts; This is the j-th original vibration direction; This represents the j-th vibration direction after fine-tuning.

[0182] The original safety warning threshold (Dimensionless, such as blood oxygen saturation threshold) fine-tuned to The calculation formula is:

[0183] ,in, The safety factor (dimensionless) is positively correlated with the patient's physiological tolerance risk level, with the high-risk level... =0.08, medium risk level =0.05, low-risk layer =0.02;

[0184] The original cleaning efficiency threshold (Dimensionless) fine-tuning to The calculation formula is: ,in, The efficiency coefficient (dimensionless) is positively correlated with the viscosity of secretions, and its value ranges from 0.03 to 0.07.

[0185] In this embodiment, .

[0186] The beneficial effects of the above technical solution are as follows: the state reliability matrix accurately represents the self and cooperative state of the vibration unit, and the combination of dual comprehensive adaptability realizes the comprehensive consideration of equipment state and environmental conditions. The second comprehensive adaptability is used as a fine-tuning driving factor and substituted into the quantitative calculation formula to realize the accurate and dynamic fine-tuning of the vibration parameter constraint range, so that the parameter range is more in line with the actual working state of the equipment and the on-site environmental conditions, effectively improving the accuracy and practicality of the initial adaptability conditions.

[0187] This invention provides a pediatric airway clearance method based on multi-mode vibration and physiological feedback, which generates an initial vibration parameter combination for the multi-mode vibration module according to the initial adaptation conditions, including:

[0188] The first feature of the three-dimensional structure of the target region of the airway and the second feature of the secretion distribution in the individualized airway simulation model are extracted. The first feature includes airway anatomical hierarchy, airway wall thickness, and pleural tissue layer thickness. The second feature includes secretion three-dimensional spatial coordinates, viscosity layering, and vertical distance and oblique penetration distance of secretion from the body surface.

[0189] Based on the anatomical landmarks on the patient's body surface, and combined with the vibration attenuation coefficients of multiple tissues such as skin, subcutaneous tissue, muscle, and chest wall between the respiratory target area and the body surface, a three-dimensional targeted mapping model of body surface-airway is constructed. The mapping model generates a layered body surface region that precisely corresponds to the respiratory target area. The layered body surface region includes a core vibration target area that matches the core location of secretions, an auxiliary vibration target area that matches the diffusion range of secretions, and an edge avoidance area for avoiding organs and sensitive tissues.

[0190] Based on the spatial depth of secretions in the respiratory tract, the vibration attenuation characteristics of multi-layered tissues, and the safe distance of thoracic organs, the vibration effect depth range is calculated. The vibration effect depth range includes the minimum effective effect depth to reach the secretion layer, the maximum safe effect depth without damaging thoracic organs and airway walls, and the multi-level targeted effect depth corresponding to the viscosity stratification of secretions.

[0191] Based on the layered surface region and vibration depth range, and combined with the determined initial adaptation conditions including vibration frequency range, vibration intensity range, vibration direction sequence, safety warning threshold, and cleaning efficiency threshold, the depth penetration parameters of the mechanical vibration unit and the surface targeting parameters of the acoustic vibration unit are matched respectively to generate a multi-mode vibration module initial vibration parameter combination that is adapted to the layered surface region and vibration depth range.

[0192] In this embodiment, the first feature refers to the three-dimensional structural features of the target region of the airway in the individualized airway simulation model. It is the core parameter characterizing the anatomical structure of the airway. In practice, it is directly extracted from the simulation model through the feature extraction function of the finite element simulation software. For example, the first feature includes airway anatomical layers, airway wall thickness, and pleural tissue layer thickness. Each feature parameter is quantified in millimeters.

[0193] Airway anatomical hierarchy refers to the classification of the airway based on the anatomical structure of the respiratory tract, reflecting the location and diameter characteristics of the airway. In practice, it is divided according to the pediatric respiratory tract anatomy standard. For example, the pediatric airway is divided into level 1 trachea, level 2 main bronchus, level 3 lobar bronchus, and level 4 bronchioles. Different vibration parameter designs correspond to different levels of airway.

[0194] The thickness of the pleural tissue layers refers to the thickness of each layer of tissue between the surface of the pleural cavity and the target area of ​​the airway. It is measured through three-dimensional reconstruction of medical images during the implementation. For example, the thickness of the pleural tissue layers includes skin thickness of 0.3 cm, subcutaneous fat thickness of 0.2 cm, chest wall muscle thickness of 0.5 cm, and pleura thickness of 0.02 cm.

[0195] The second feature refers to the distribution characteristics of respiratory secretions in the individualized respiratory tract simulation model. It is a core parameter that characterizes the state and spatial location of secretions. In practice, it is extracted through the secretion analysis module of the simulation software. For example, the second feature includes the three-dimensional spatial coordinates of secretions, viscosity stratification, vertical distance of secretions from the body surface, and oblique penetration distance.

[0196] The three-dimensional spatial coordinates of secretions refer to the spatial coordinates of the core accumulation area of ​​secretions within a three-dimensional coordinate system established with the anatomical landmarks on the child's body surface as the origin. In practice, the coordinate positioning function of simulation software is used to determine the spatial coordinates of the secretions. For example, the three-dimensional spatial coordinates of secretions are X=6cm, Y=4cm, and Z=5cm, which accurately characterizes the spatial location of the secretions.

[0197] Viscosity stratification refers to the classification of accumulated secretions into levels based on their estimated viscosity. In practice, it is divided into three levels: mild, moderate, and severe. For example, the core area of ​​the secretion is severely viscous, the surrounding diffusion area is moderately viscous, and the edge area is mildly viscous. Different viscosity levels correspond to different vibration intensities and frequencies.

[0198] The vertical distance between the secretion and the body surface refers to the vertical straight-line distance between the core area of ​​the secretion and the corresponding position on the child's body surface. In practice, this distance is calculated using the distance calculation function of a three-dimensional coordinate system. For example, the vertical distance between the secretion and the body surface is 4cm, which is the core basis for designing the depth of vibration.

[0199] The oblique penetration distance of secretions from the body surface refers to the oblique straight-line distance between the core area of ​​the secretion and the vibration target area on the child's body surface. It takes into account the angular relationship between the body surface and the airway. In practice, it is calculated using a spatial distance formula in a three-dimensional coordinate system. For example, the oblique penetration distance of secretions from the body surface is 4.5cm.

[0200] Anatomical landmarks refer to points on the child's body surface that have clear anatomical features and can be used as positioning references. In practice, typical anatomical landmarks of the pediatric thoracic cavity are selected, such as the suprasternal notch, sternal angle, intersection of the midclavicular line and the 5th intercostal space, and inferior angle of the scapula, as the references for constructing a three-dimensional coordinate system and targeted mapping model.

[0201] The vibration attenuation coefficient of multi-layer tissues refers to the proportion of energy attenuation when vibration energy passes through different tissues in the thoracic cavity, reflecting the transmission efficiency of vibration energy. In practice, it is determined and calibrated through in vitro experiments on pediatric tissues. For example, the vibration attenuation coefficient of skin is 0.15, muscle tissue is 0.3, chest wall bones are 0.2, and pleura is 0.05. The smaller the attenuation coefficient, the higher the vibration energy transmission efficiency.

[0202] The body surface-airway three-dimensional targeted mapping model refers to a three-dimensional spatial mapping model constructed based on anatomical landmarks on the body surface and combined with multi-layer tissue vibration attenuation coefficients. It is constructed using three-dimensional modeling software to achieve a precise correspondence between the body surface area and the internal airway area. For example, the 5th-6th intercostal region along the midclavicular line on the body surface is mapped to the secretion accumulation area of ​​the right lower lobe bronchus.

[0203] The core vibration target area refers to the body surface area that precisely corresponds to the core accumulation location of secretions in the layered body surface region. It is the main area where vibration energy is applied. During implementation, the vibration intensity and frequency in this area are at their optimal values. For example, the area of ​​the 5th intercostal space along the midclavicular line corresponding to the core secretion area is the core vibration target area, with an area of ​​approximately 5cm × 5cm.

[0204] The auxiliary vibration target area refers to the area on the body surface corresponding to the spread of secretions in the layered body surface region. It is an auxiliary area for vibration energy. In practice, the vibration intensity in this area is slightly lower than that in the core vibration target area. For example, the area on the body surface corresponding to the spread area around the secretions in the midclavicular line between the 4th and 6th intercostal spaces is the auxiliary vibration target area, with an area of ​​approximately 8cm × 6cm.

[0205] The edge avoidance zone refers to the area on the body surface that needs to be avoided in the layered body surface area, corresponding to the thoracic organs and sensitive tissues. No vibration is applied to this area during the implementation to avoid organ damage. For example, the area between the 2nd and 4th intercostal spaces on the left sternal border, corresponding to the heart and major blood vessels, is the edge avoidance zone, and no vibration is applied throughout the process.

[0206] The vibration effect depth range refers to the effective depth range of vibration energy calculated based on the spatial depth of secretions, the vibration attenuation characteristics of multi-layer tissues, and the safe distance of thoracic organs. The implementation includes three sub-depth ranges. For example, the vibration effect depth range is 3-5cm, covering the spatial depth of secretions.

[0207] The minimum effective depth of action refers to the minimum depth to which vibration energy can reach the secretion layer. It is the lower limit of vibration effect. In practice, it is calculated by combining the vertical distance of the secretion from the body surface and the tissue attenuation coefficient. For example, if the vertical distance of the secretion from the body surface is 4cm, the minimum effective depth of action is set to 3.5cm after considering tissue attenuation.

[0208] The maximum safe depth of action refers to the maximum depth to which vibration energy will not damage the thoracic organs and airway walls. It is the upper limit of vibration action and is determined in practice by combining the location of the thoracic organs and the tolerance of the airway walls. For example, the maximum safe depth of action is set to 5cm to avoid vibration energy penetrating deep into the thoracic cavity and damaging the lung tissue and heart.

[0209] Multi-level targeting depth refers to the targeting depth corresponding to the stratification of secretion viscosity. In practice, different vibration depths are designed for secretions with different viscosities. For example, the multi-level targeting depth for severely viscous secretions is 4-5cm, for moderate secretions it is 3.5-4.5cm, and for mild secretions it is 3-4cm.

[0210] The depth penetration parameters of a mechanical vibration unit refer to the vibration parameters used by the unit to achieve deep vibration, enabling energy to reach the deep respiratory tract. In practice, mechanical vibration, due to its low frequency and large amplitude, possesses deep penetration capability. For example, depth penetration parameters include low frequencies of 10-30Hz and higher intensity of 3-8. .

[0211] The surface targeting parameters of the acoustic vibration unit refer to the vibration parameters used by the acoustic vibration unit to achieve surface-targeted vibration and act on the shallow secretions of the respiratory tract. In practice, acoustic vibration has surface targeting capability due to its high frequency and small amplitude. For example, the surface targeting parameters include high frequency 500-2000Hz and low intensity 1-3 .

[0212] The beneficial effects of the above technical solution are as follows: by extracting precise features of the respiratory tract and secretions, a three-dimensional targeted mapping model of the body surface and airway is constructed to achieve precise stratification of the vibration area. Combined with secretion features, the vibration depth range of multi-level targeting is calculated. At the same time, differentiated parameters are matched for different characteristics of mechanical and acoustic vibration units, realizing the targeted and stratified design of multi-mode vibration. This effectively reduces the ineffective attenuation of vibration energy in the body surface tissue, improves the utilization efficiency of vibration energy and the accuracy of secretion clearance. In addition, the design of the edge avoidance zone avoids damage to thoracic organs and sensitive tissues by vibration, further improving the safety and efficiency of pediatric airway clearance.

[0213] This invention provides a pediatric airway clearing system based on multi-mode vibration and physiological feedback, such as... Figure 2 As shown, it includes:

[0214] The initial condition determination module is used to acquire multimodal input data from pediatric patients and input it into a pre-trained neural network model to generate multidimensional physiological representation signals to determine the initial adaptation conditions for airway clearance.

[0215] The simulation module is used to dynamically construct an individualized airway simulation model for a pediatric patient based on the multi-dimensional physiological characterization signals and synchronously acquired real-time medical image streams and respiratory dynamic parameters. The module also controls the multi-mode vibration module to apply vibration to the layered body surface regions corresponding to the individualized airway simulation model according to the initial adaptation conditions. The layered body surface regions include: a core vibration target area, an auxiliary vibration target area, and an edge avoidance area.

[0216] The feedback judgment module is used to collect real-time physiological parameters of pediatric patients during the airway clearance simulation process, and to perform noise reduction and standardization processing on the collected real-time physiological parameters to obtain feedback data for adjustment judgment. The real-time physiological parameters include at least respiratory rate, blood oxygen saturation, airway resistance and breath sound signal.

[0217] The comparative analysis module is used to compare and analyze the feedback data with preset pediatric safety physiological thresholds and secretion clearance judgment thresholds, and adjust the vibration mode and corresponding vibration parameters of the multi-mode vibration module in real time until the airway clearance standard is met, and then perform actual airway clearance on the pediatric patient.

[0218] The beneficial effects of the above technical solution are: by taking into account the physiological tolerance, respiratory tract anatomical features and behavioral characteristics of pediatric patients, the individualized design of vibration parameters is achieved, and then the safety and efficiency of pediatric airway clearance are further effectively improved through subsequent simulation and dynamic adjustment, reducing the risk of airway tissue damage.

[0219] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A pediatric airway clearance method based on multi-mode vibration and physiological feedback, characterized in that, include: Step 1: Acquire multimodal input data from pediatric patients and input it into a pre-trained neural network model to generate multidimensional physiological representation signals to determine the initial adaptation conditions for airway clearance; Step 2: Based on the multi-dimensional physiological characterization signals and the synchronously acquired real-time medical image stream and respiratory dynamic parameters, dynamically construct an individualized airway simulation model corresponding to the pediatric patient, and control the multi-mode vibration module to apply vibration to the layered body surface area corresponding to the individualized airway simulation model according to the initial adaptation conditions for simulation. The layered body surface area includes: core vibration target area, auxiliary vibration target area and edge avoidance area. Step 3: Real-time physiological parameters of pediatric patients during the airway clearance simulation process are collected, and the collected real-time physiological parameters are denoised and standardized to obtain feedback data for adjustment and judgment. The real-time physiological parameters include at least respiratory rate, blood oxygen saturation, airway resistance and breath sound signal. Step 4: Compare and analyze the feedback data with the preset pediatric safety physiological threshold and secretion clearance judgment threshold, and adjust the vibration mode and corresponding vibration parameters of the multi-mode vibration module in real time until the airway clearance standard is met, and then perform actual airway clearance on the pediatric patient.

2. The pediatric airway clearance method based on multi-mode vibration and physiological feedback according to claim 1, characterized in that, Determine the initial adaptation conditions for airway clearance, including: The multimodal input data is input into a pre-trained neural network model to generate a multi-dimensional physiological representation signal that includes physiological tolerance, cleanup risk and behavioral adaptability. The multi-dimensional physiological representation signal is a high-dimensional vector, and each dimension corresponds to a feature weight and a risk level label. Based on the multi-dimensional physiological characterization signals, the K-means clustering algorithm is used to perform risk stratification, dividing pediatric patients into three physiological tolerance risk layers: low, medium, and high. Each risk layer corresponds to a preset vibration parameter constraint range. The local pediatric airway clearance case knowledge base is invoked to match the historical cases with the highest similarity to the current patient's characteristic signals, and the vibration parameter constraint range is optimized and corrected. Based on real-time environmental parameters and equipment response parameters for airway clearance, the corrected vibration parameter constraint range is dynamically fine-tuned to obtain initial adaptation conditions. These initial adaptation conditions include: initial range of vibration frequency, initial range of vibration intensity, initial sequence of vibration direction, and corresponding safety warning threshold and clearance efficiency threshold.

3. The pediatric airway clearance method based on multi-mode vibration and physiological feedback according to claim 1 or 2, characterized in that, The multimodal input data includes at least the following: static basic physiological parameters: age, weight, developmental stage label, basic respiratory rate, estimated viscosity of respiratory secretions, electroencephalogram (EEG) signals, and near-infrared brain functional imaging signals. Behavioral status data: agitation score, cooperation score.

4. The pediatric airway clearance method based on multi-mode vibration and physiological feedback according to claim 1, characterized in that, Dynamically construct individualized respiratory simulation models for corresponding pediatric patients, including: The real-time medical image stream is time-aligned and reconstructed in three dimensions to extract dynamic geometric features of the airway, including airway lumen diameter, airway wall thickness, volume and distribution density of secretions, and airway deformation field during the respiratory cycle. At the same time, the respiratory dynamic parameters are feature extracted to obtain airway resistance spectrum, respiratory flow waveform and compliance change curve. The extracted dynamic geometric features are used as the initial geometric structure of the simulation model, and the feature extraction results of respiratory dynamic parameters are used as the input boundary of the fluid dynamics simulation. In addition, the tolerance features in the multi-dimensional physiological characterization signals are used as the safety constraint threshold for airway deformation, limiting the upper limit of airway wall stress and secretion migration speed during the simulation process, thus constructing a finite element simulation framework. The subsequent frame sequence of the real-time synchronized medical image stream is used to invert the airway deformation field based on the non-rigid image registration algorithm, update the geometric structure and material properties of the finite element simulation framework, and modify the fluid dynamic boundary conditions in combination with the real-time changes of respiratory dynamic parameters to obtain an individualized airway simulation model.

5. The pediatric airway clearance method based on multi-mode vibration and physiological feedback according to claim 1, characterized in that, According to the initial adaptation conditions, the multi-mode vibration module is controlled to apply vibration to the layered body surface region corresponding to the individualized respiratory tract simulation model for simulation, including: Based on the three-dimensional structure and secretion distribution of the individualized respiratory tract simulation model, a layered body surface region and vibration depth range corresponding to the respiratory tract target region are generated, and the initial vibration parameter combination of the multi-mode vibration module is generated according to the initial adaptation conditions. The control multi-mode vibration module applies vibration to the layered body surface region according to the initial vibration parameter combination, and at the same time collects the simulation response data of the individualized respiratory tract simulation model in real time, including airway wall stress distribution, secretion displacement field, airway pressure change and vibration energy attenuation coefficient in respiratory tissue; The simulation response data is compared with the preset vibration efficiency threshold and safety constraint threshold. If the simulation response data does not meet the standards, the parameter combination of the multi-mode vibration module is dynamically adjusted based on the optimization direction of the vibration parameters retrieved from the individualized airway simulation model, and the simulation is repeated until the simulation response data meets the preset vibration efficiency and safety constraint conditions. The optimized multi-mode vibration parameter combination and the corresponding layered body surface area are then output as the initial vibration control scheme for subsequent actual airway cleaning.

6. The pediatric airway clearance method based on multi-mode vibration and physiological feedback according to claim 1, characterized in that, When feedback data shows that the physiological state of a pediatric patient exceeds the preset pediatric safety physiological threshold, the vibration intensity is reduced and switched to a gentle vibration mode. When feedback data shows that the respiratory secretions have not reached the clearance standard, the vibration frequency and direction are adjusted according to the loosening of the secretions until the feedback data shows that the respiratory secretions have reached the clearance standard and the pediatric patient's physiological state is maintained within the preset pediatric safe physiological threshold range. Then, the vibration cleaning stops, and the simulation process of one respiratory clearance is completed.

7. The pediatric airway clearance method based on multi-mode vibration and physiological feedback according to claim 2, characterized in that, Based on real-time environmental parameters and equipment response parameters for airway clearance, the corrected vibration parameter constraint range is dynamically fine-tuned, including: The device response parameters of the mechanical vibration unit and the acoustic vibration unit in the multi-mode vibration module are collected, and a state reliability matrix is ​​constructed, wherein the state reliability matrix contains only two off-diagonal elements; Based on the state confidence matrix, the first comprehensive fit degree of the mechanical vibration unit and the acoustic vibration unit are calculated respectively, and the unit corresponding to the maximum value of the two first comprehensive fit degrees is selected as the optimal reference unit. If the first comprehensive fit degrees of the two units are equal, one unit is randomly selected as the optimal reference unit. In response to real-time environmental parameters of the airway clearance scenario, the second comprehensive fit of the optimal reference unit is calculated, and the second comprehensive fit is used as a fine-tuning driving factor to dynamically fine-tune the corrected vibration parameter constraint range.

8. The pediatric airway clearance method based on multi-mode vibration and physiological feedback according to claim 5, characterized in that, The initial vibration parameter combination of the multi-mode vibration module is generated according to the initial adaptation conditions, including: The first feature of the three-dimensional structure of the target region of the airway and the second feature of the secretion distribution in the individualized airway simulation model are extracted. The first feature includes airway anatomical hierarchy, airway wall thickness, and pleural tissue layer thickness. The second feature includes secretion three-dimensional spatial coordinates, viscosity layering, and vertical distance and oblique penetration distance of secretion from the body surface. Based on the anatomical landmarks on the patient's body surface, and combined with the vibration attenuation coefficients of multiple tissues such as skin, subcutaneous tissue, muscle, and chest wall between the respiratory target area and the body surface, a three-dimensional targeted mapping model of body surface-airway is constructed. The mapping model generates a layered body surface region that precisely corresponds to the respiratory target area. The layered body surface region includes a core vibration target area that matches the core location of secretions, an auxiliary vibration target area that matches the diffusion range of secretions, and an edge avoidance area for avoiding organs and sensitive tissues. Based on the spatial depth of secretions in the respiratory tract, the vibration attenuation characteristics of multi-layered tissues, and the safe distance of thoracic organs, the vibration effect depth range is calculated. The vibration effect depth range includes the minimum effective effect depth to reach the secretion layer, the maximum safe effect depth without damaging thoracic organs and airway walls, and the multi-level targeted effect depth corresponding to the viscosity stratification of secretions. Based on the layered surface region and vibration depth range, and combined with the determined initial adaptation conditions including vibration frequency range, vibration intensity range, vibration direction sequence, safety warning threshold, and cleaning efficiency threshold, the depth penetration parameters of the mechanical vibration unit and the surface targeting parameters of the acoustic vibration unit are matched respectively to generate a multi-mode vibration module initial vibration parameter combination that is adapted to the layered surface region and vibration depth range.

9. A pediatric airway clearing system based on multi-mode vibration and physiological feedback, characterized in that, include: The initial condition determination module is used to acquire multimodal input data from pediatric patients and input it into a pre-trained neural network model to generate multidimensional physiological representation signals to determine the initial adaptation conditions for airway clearance. The simulation module is used to dynamically construct an individualized airway simulation model for a pediatric patient based on the multi-dimensional physiological characterization signals and synchronously acquired real-time medical image streams and respiratory dynamic parameters. The module also controls the multi-mode vibration module to apply vibration to the layered body surface regions corresponding to the individualized airway simulation model according to the initial adaptation conditions. The layered body surface regions include: a core vibration target area, an auxiliary vibration target area, and an edge avoidance area. The feedback judgment module is used to collect real-time physiological parameters of pediatric patients during the airway clearance simulation process, and to perform noise reduction and standardization processing on the collected real-time physiological parameters to obtain feedback data for adjustment judgment. The real-time physiological parameters include at least respiratory rate, blood oxygen saturation, airway resistance and breath sound signal. The comparative analysis module is used to compare and analyze the feedback data with preset pediatric safety physiological thresholds and secretion clearance judgment thresholds, and adjust the vibration mode and corresponding vibration parameters of the multi-mode vibration module in real time until the airway clearance standard is met, and then perform actual airway clearance on the pediatric patient.

Images