A method and device for constructing a millimeter wave radar-based and chronic obstructive pulmonary disease prediction model

Abstract

The present application relates to a kind of based on millimeter wave radar and slow obstructive pulmonary disease prediction model construction method and system, its method includes: based on the echo signal of millimeter wave radar of target doing step test, and according to the echo signal real-time determination target motion state;According to the motion state of target, adjust the filtering method of echo signal;Through variational mode decomposition method and space spectrum estimation method, extract the sign data of target from filtered echo signal, the sign data includes velocity, acceleration, respiratory rate and heartbeat frequency;Based on the sign data of multiple targets, construct slow obstructive pulmonary disease data set;Based on slow obstructive pulmonary disease data set, construct slow obstructive pulmonary disease prediction model by random forest method.The present application realizes unattended measurement sign data by the echo of millimeter wave radar, and constructs prediction model by random forest, to improve the accuracy of model prediction.

Description

Technical Field

[0001] This invention belongs to the field of millimeter radar and physiological data technology, specifically relating to a method and device for constructing a prediction model for COPD based on millimeter-wave radar. Background Technology

[0002] Chronic obstructive pulmonary disease (COPD) is the third leading cause of death in my country, leading to shortness of breath. It suffers from low awareness and low diagnosis rates. Current methods for initial detection of COPD include: patient-assessed questionnaires based on symptoms and exercise capacity, with scores used to assess risk, but these methods lack sensitivity and specificity; chest CT scans can quantify small airway lesions and alveolar destruction in the early stages of COPD, but involve radiation exposure, and frequent examinations may increase the patient's radiation risk, and the tests are relatively expensive; measuring forced expiratory volume in one second (FEV1) and forced vital capacity (FVC) can assess airflow limitation, with the FEV1 / FVC ratio being the gold standard for COPD diagnosis. This can be measured using a spirometer, but requires close patient cooperation, such as forced exhalation, which can be challenging for some patients (e.g., children) and demands high operator skill. Summary of the Invention

[0003] To address the problems raised in the background art, a method for constructing a COPD prediction model based on millimeter-wave radar is provided in a first aspect of the present invention. The method includes: using the echo signal of a millimeter-wave radar to perform a step test on a target, and determining the target's motion state in real time based on the echo signal; adjusting the filtering method of the echo signal according to the target's motion state; extracting the target's vital signs data from the filtered echo signal using variational mode decomposition and spatial spectrum estimation methods, wherein the vital signs data includes velocity, acceleration, respiratory rate, and heart rate; constructing a COPD dataset based on the vital signs data of multiple targets; and constructing a COPD prediction model based on the COPD dataset using a random forest method.

[0004] In some embodiments of the present invention, the method for adjusting the filtering of the echo signal according to the motion state of the target includes: if the target is stationary, filtering the phase and frequency of the echo signal.

[0005] In some embodiments of the present invention, the step of extracting the target's vital signs data from the filtered echo signal using variational mode decomposition and spatial spectrum estimation methods includes: extracting the target's respiratory signal and heartbeat signal from the filtered echo signal using variational mode decomposition; and extracting the target's respiratory rate and heartbeat rate from the filtered echo signal using spatial spectrum estimation.

[0006] Furthermore, extracting the target's respiratory and heartbeat signals from the filtered echo signal using the variational mode decomposition method includes: decomposing the filtered echo signal into multiple components and establishing a constrained variational model based on the multiple components; transforming the constrained variational model into an unconstrained variational model using a quadratic penalty term and Lagrange multipliers; iterating the unconstrained variational model multiple times based on the quadratic penalty term and Lagrange multipliers until a preset convergence condition is reached to obtain the optimal solution of the unconstrained variational model; and separating the respiratory and heartbeat signals based on the optimal solution.

[0007] Furthermore, the extraction of the target's respiratory rate and heart rate from the filtered echo signal using the spatial spectrum estimation method includes: constructing a covariance matrix for the filtered echo signal, and performing eigenvalue decomposition on the covariance matrix to obtain a signal subspace and a noise subspace; constructing a cost function based on the signal subspace and the noise subspace; and determining the respiratory rate and heart rate through the peak value of the cost function.

[0008] In the above embodiments, the step of determining the target's motion state in real time based on the echo signal includes: mixing and filtering the echo signal and the transmitted signal of the millimeter radar wave to obtain an intermediate frequency signal; performing Fourier transform and spectrum analysis on the intermediate frequency signal to obtain the target's distance data and velocity data; and determining the target's motion state based on the distance data and the velocity data.

[0009] A second aspect of the present invention provides a system for constructing a COPD prediction model based on millimeter-wave radar, comprising: a determination module for determining the motion state of a target in real time based on the echo signal of a millimeter-wave radar used for step testing; an adjustment module for adjusting the filtering method of the echo signal according to the motion state of the target; an extraction module for extracting the target's vital signs data from the filtered echo signal using variational mode decomposition and spatial spectrum estimation methods, the vital signs data including velocity, acceleration, respiratory rate, and heart rate; and a construction module for constructing a COPD dataset based on the vital signs data of multiple targets; and constructing a COPD prediction model based on the COPD dataset using a random forest method.

[0010] A third aspect of the present invention provides an electronic device comprising: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method for constructing a prediction model for COPD based on millimeter-wave radar provided in the first aspect of the present invention.

[0011] In a fourth aspect, the present invention provides a computer-readable medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for constructing a prediction model for COPD based on millimeter-wave radar provided in the first aspect of the present invention.

[0012] The beneficial effects of this invention are:

[0013] This disclosure enhances the safety of testing by reducing the complexity of collecting chronic obstructive pulmonary disease (COPD) data. Simultaneously, millimeter-wave radar information is input and transmitted back to the voice module for broadcast, providing timely feedback on the testing progress. After completing all exercise cycles, the test subject rests for two minutes at a designated location, observing changes in heart rate and respiratory rate. All information collected during the testing process is input into a pre-trained random forest classification model, which can identify the results in a short time. Furthermore, the test can be conducted unattended, allowing patients to self-test while ensuring their safety. After extensive validation testing, the model achieved a 92% accuracy rate in identifying COPD patients. The test results are reported to the testing platform, allowing test subjects to conduct further examinations based on the results. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the basic process of constructing a prediction model for COPD based on millimeter-wave radar in some embodiments of the present invention.

[0015] Figure 2 This is one of the flowcharts illustrating the random forest method in some embodiments of the present invention;

[0016] Figure 3 This is a second flowchart illustrating the random forest method in some embodiments of the present invention;

[0017] Figure 4 This is a schematic diagram of the structure of a system for constructing a prediction model for COPD based on millimeter-wave radar in some embodiments of the present invention;

[0018] Figure 5 This is a schematic diagram of the structure of an electronic device in some embodiments of the present invention. Detailed Implementation

[0019] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0020] refer to Figure 1In a first aspect, the present invention provides a method for constructing a COPD prediction model based on millimeter-wave radar, comprising: S100. determining the motion state of the target in real time based on the echo signal of the millimeter-wave radar used for step testing of the target; S200. adjusting the filtering method of the echo signal according to the motion state of the target; S300. extracting the target's vital signs data from the filtered echo signal using variational mode decomposition and spatial spectrum estimation methods, wherein the vital signs data includes velocity, acceleration, respiratory rate, and heart rate; S400. constructing a COPD dataset based on the vital signs data of multiple targets; and constructing a COPD prediction model using a random forest method based on the COPD dataset.

[0021] The basic procedure of the step test is illustrated as follows: The tester stands behind the starting line. After the radar detects the target, the voice module issues an instruction: "Start the test. Each time you go up and down the stairs counts as one activity cycle. You need to complete ten activity cycles. Please keep walking during the test and watch your step." The tester then begins climbing the stairs. The radar detects the target in motion mode, monitoring its trajectory and calculating its speed and acceleration information, which is then uploaded to the random forest model as input. After completing an odd number of activity cycles, the tester receives a voice prompt: "You have completed 1 (3, 5, 7, 9) activity cycle. Keep going!" After every even number of activity cycles, the tester rests for 30 seconds at a designated location and receives a voice prompt: "You have completed 2 (4, 6, 8) activity cycles. Please stand in the red area and rest for 30 seconds." "You have completed 10 activity cycles. Please stand in the red area and rest for two minutes." During the rest period, the radar detects the target in stationary mode, testing its heart rate and respiration, which are then uploaded to the random forest model as input. This data is also fed back to the voice module, which determines whether the tester can continue the test based on a pre-set threshold. If they can... The voice prompt reads, "You can continue the test." If not, it reads, "Please pause the test!" This is considered an abnormal situation and uploaded to the testing platform. Based on the actual situation, the thresholds are set to 60 breaths / minute and 160 heart rate / minute. The test is repeated until ten exercise cycles are completed. During this time, the voice module randomly announces, "Good, keep going." "You did very well." "Very good! Keep going a little longer." After the test, the voice module issues the instruction, "You have completed the test, please leave the testing area." The collected test data is categorized to determine if the test subject is a COPD patient, and the results are uploaded to the testing platform. The test subject then undergoes further examination based on the test results.

[0022] In some embodiments of the present invention, step S100, which involves determining the motion state of the target in real time based on the echo signal, includes:

[0023] S101. Mix and filter the echo signal and the transmitted signal of the millimeter radar wave to obtain the intermediate frequency signal;

[0024] Specifically, a millimeter-wave radar based on frequency modulated continuous wave (FMCW) is used, with the antenna acting as a transceiver. The radar module transmits a linearly modulated pulse signal (TX) whose frequency changes linearly with time. The echo signal (RX) is obtained after reflection from the target. The original signal and the echo signal are mixed and filtered by a mixer to obtain an intermediate frequency signal (IF).

[0025] S102. Perform Fourier transform and spectrum analysis on the intermediate frequency signal to obtain the target's distance and velocity data;

[0026] Specifically, the intermediate frequency (IF) signal is the frequency difference between TX and RX, and the initial phase is the phase difference at the current moment. The IF signal is sampled by an analog-to-digital converter (ADC) to obtain the radar's time-domain signal. The raw ADC-sampled signal data is then processed chirp by chirp using multi-pulse incoherent accumulation to form an N×M matrix. Range-dimensional FFT is performed on the M frames of fast-time sampling points to obtain range features. After Fourier transform, the frequency corresponding to the peak value is extracted and is proportional to the distance, thus converting the frequency to distance. Since a single frame is too short, velocity information over a distance is calculated, with each 0.5m movement considered a node. The velocity can be calculated by combining the distance features with the frame time and the number of frames. Finally, acceleration is obtained by time differentiation of the velocity data, with one acceleration calculated every 5m movement.

[0027] S103. Based on the distance data and the speed data, determine the motion state of the target.

[0028] Specifically, the distance feature combined with the speed feature can determine whether the target is stationary or moving. When the speed is 0 and the distance is less than 1m (within the red area), the radar determines that it is in stationary mode and tests the target's heart rate and breathing information. When the distance is greater than 2m (outside the starting line), the radar determines that it is in moving mode and tests the target's speed and acceleration information.

[0029] In step S200 of some embodiments of the present invention, the method for adjusting the filtering of the echo signal according to the motion state of the target includes:

[0030] S201. If the target is stationary, filter the phase and frequency of the echo signal;

[0031] When the subject is stationary, heartbeat and respiration information can be extracted from the intermediate frequency signal. The target's location can be determined from the extracted distance features. However, because the displacement change during chest cavity vibration is very small (1-12 mm for respiration and 0.1-0.5 mm for heartbeat), it cannot be calculated using the intermediate frequency signal information. Since the phase is significantly affected by these displacements, phase processing is performed to extract vital sign information. The specific phase is obtained using the arctangent function. Because large chest cavity movements can lead to an absolute phase difference greater than π between adjacent sampling points, a phase decoupling method of adding or subtracting 2π is used to process discontinuous phases. Then, continuous phase values ​​are subtracted from the phase difference to enhance the target signal and suppress phase drift.

[0032] Specifically, an FIR bandpass filter was used to filter out respiratory and heart rate signals. Since respiratory and heart rates are higher after exercise than usual, and patients' respiratory and heart rates are also higher than non-patients', the passband frequency ranges were set to 0.2~1.2Hz and 1.0~3.0Hz, respectively. The rest time was set to 30s. An adaptive variational mode decomposition algorithm was used to decompose respiratory and heart rate signals. The intrinsic mode components can be regarded as intrinsic mode signals with a center frequency and finite bandwidth. The final intrinsic mode function components were obtained by iterative updates, which can effectively avoid mode aliasing.

[0033] In step S300 of some embodiments of the present invention, extracting the target's vital signs data from the filtered echo signal using variational mode decomposition and spatial spectrum estimation methods includes:

[0034] S301. Extract the target's respiratory and heartbeat signals from the filtered echo signals using variational mode decomposition.

[0035] Furthermore, in step S301, extracting the target's respiratory and heartbeat signals from the filtered echo signal using variational mode decomposition includes:

[0036] S3011. Decompose the filtered echo signal into multiple components, and establish a constrained variational model based on the multiple components;

[0037] Specifically, the original signal Decompose into k components and establish a constrained variational model:

[0038] ,

[0039] in, The function is the unit impulse function. To solve the constrained optimization problem, the constrained variational problem is transformed into an unconstrained variational problem. Leveraging the advantages of the quadratic penalty term and the Lagrange multiplier method, an augmented Lagrange function is introduced:

[0040] S3012. By using a quadratic penalty term and Lagrange multipliers, the constrained variational model is transformed into an unconstrained variational model;

[0041] Specifically, ;in, It is a punishment factor. These are Lagrange multipliers; through iterative algorithms using alternating direction multipliers, for all... Update functional This continues until the convergence condition is met.

[0042] S3013. Based on the quadratic penalty term and Lagrange multipliers, the unconstrained variational model is iterated multiple times until the preset convergence condition is reached, and the optimal solution of the unconstrained variational model is obtained; S3014. Based on the optimal solution, the respiratory signal and the heartbeat signal are separated.

[0043] Specifically, the convergence condition is:

[0044] ,

[0045] ,

[0046] Where the superscript n represents the iteration number, For component relative error, For the absolute error of the components, when both equations are satisfied, the saddle point of the augmented Lagrange model is obtained, the optimal solution of the model is calculated, and the solution is determined based on the components. Identify the corresponding signals to separate respiratory and heartbeat signals.

[0047] S302. Extract the target's respiratory rate and heart rate from the filtered echo signal using a spatial spectrum estimation method.

[0048] It is understandable that the VMD algorithm determines the frequency and amplitude of each mode function through continuous optimization, thus achieving effective signal decomposition. It decomposes a real signal into a series of mode functions. Each The spectrum is concentrated at its center frequency. Nearby, the algorithm iteratively updates, including the modal signals and center frequency, so that all modal functions... The goal is to minimize the error between the sum of modal bandwidths and the original signal, while maintaining the smoothness of the modal signals. VMD requires the modal functions to satisfy two conditions: the sum of modal bandwidths must be minimized, and the sum of modal bandwidths must equal the original signal. VMD transforms the objective function into a constrained Lagrangian function and optimizes it using variational methods.

[0049] Furthermore, the extraction of the target's respiratory rate and heart rate from the filtered echo signal using the spatial spectrum estimation method includes:

[0050] S3021. Construct the covariance matrix of the filtered echo signal, and perform eigenvalue decomposition on the covariance matrix to obtain the signal subspace and noise subspace;

[0051] Specifically, the input signal S(t) is truncated into a signal X(t) of length N*M, thus constructing an N*M data matrix.

[0052] ,

[0053] ，

[0054] ,

[0055] Where A is the array direction matrix and N(t) is the noise matrix; the covariance matrix R is obtained. ,in, Let be the signal covariance matrix. Here is the noise covariance matrix;

[0056] S3022. Construct a cost function based on the signal subspace and noise subspace;

[0057] Specifically, the signal subspace US and noise subspace UN are obtained by eigenvalue decomposition of the covariance matrix, and the angle [0, 2π] is divided into Nω angle points to obtain the angle search vector A. ):

[0058]

[0059] Constructing the cost function for: ,in, for The conjugate matrix, for The conjugate matrix;

[0060] S3023. Determine the respiratory rate and heart rate using the peak value of the cost function.

[0061] Specifically, by utilizing the orthogonality between the signal space and the noise space, we can obtain The maximum value is obtained by finding the frequency corresponding to the peak angle through the peak value of the cost function, namely the respiratory rate and the heart rate.

[0062] It is understandable that eigenvalue decomposition is performed using the covariance matrix of the received data. Based on the signal autocorrelation, it is decomposed into feature signals with strong autocorrelation and feature signals with weak autocorrelation. The strong autocorrelation signal is regarded as the main feature signal, while the weak autocorrelation signal is regarded as noise signal. The orthogonality of the separated signal subspace and noise subspace is used to determine the angle, construct the spatial domain, and perform a global search for spectral peaks, thereby realizing the parameter estimation of the signal.

[0063] refer to Figure 2 and Figure 3 In step S400 of some embodiments of the present invention, a COPD disease dataset is constructed based on the vital sign data of multiple targets; and a COPD disease prediction model is constructed based on the COPD disease dataset using the random forest method.

[0064] Specifically, a random forest ensemble learning method is employed to improve prediction accuracy by constructing multiple decision trees and averaging the results. The input consists of four features: speed, acceleration, heart rate, and respiratory rate, with the output being two categories: patient and non-patient. Missing and outlier values ​​in the collected data are processed; then, features are normalized to improve model training efficiency and performance; feature importance is assessed, and feature weights are determined (20% each for speed and acceleration, 30% each for heart rate and respiratory rate) to optimize model performance; the dataset is divided into training and testing sets (80% training, 20% testing). The model is trained using the training set, and random forest parameters, such as the number of trees and maximum depth, are adjusted to find the optimal parameter combination and achieve the highest accuracy. The model consists of multiple decision trees, each using a randomly selected subset of features during training, with the final classification result determined by the voting results of each tree. Finally, the trained model is used for classification to screen patients with COPD.

[0065] Specifically, the random forest ensemble learning method mainly consists of three processes: training set generation, decision tree training to form a random forest, and test set testing, i.e., the formation and testing of the random forest.

[0066] The input training dataset consists of 100 non-COPD patients and 100 COPD patients, with features including velocity, acceleration, heart rate, and respiratory rate. The Bootstrap method is used to randomly sample with replacement from the original training set. Each node in each layer represents a specific feature of the input data, while leaf nodes represent the category to which the data belongs. This process begins at the root node, and based on the node's splitting criteria, we partition according to a certain feature of the input object. Through these steps, we eventually reach a specific leaf node, thus obtaining the prediction result of the decision tree.

[0067] The CART algorithm was chosen for splitting. CART stands for Classification and Regression Tree, a binary decision tree. When splitting nodes, the Gini index is minimized to calculate the impurity of the set (P). The Gini index is chosen based on maximizing the purity of each child node. The Gini coefficient is:

[0068] ,

[0069] in, Indicates category The probability in the set, calculating the coefficients after splitting, when divided into... After two categories, the Gini coefficients for this test are obtained:

[0070] ,

[0071] in, The number of samples that meet the conditions. The number of samples that meet the conditions. The total number of sets is given. The feature with the lowest Gini index is selected for splitting; this feature is considered the best choice for the current splitting step. Different splitting methods or combinations for each feature are evaluated to determine the optimal split point for each feature. Then, by comparing the optimal split points of each feature, each decision tree is allowed to grow infinitely to its maximum extent without any pruning. The resulting multiple classification trees form a random forest, ultimately determining the best classification.

[0072] Example 2

[0073] refer to Figure 4 In a second aspect, the present invention provides a system 1 for constructing a COPD prediction model based on millimeter-wave radar, comprising: a determination module 11, used to determine the motion state of a target in real time based on the echo signal of a millimeter-wave radar used for step testing of a target; an adjustment module 12, used to adjust the filtering method of the echo signal according to the motion state of the target; an extraction module 13, used to extract the target's vital signs data from the filtered echo signal using variational mode decomposition and spatial spectrum estimation methods, the vital signs data including velocity, acceleration, respiratory rate, and heart rate; and a construction module 14, used to construct a COPD dataset based on the vital signs data of multiple targets; and to construct a COPD prediction model based on the COPD dataset using a random forest method.

[0074] Furthermore, the extraction module 13 includes: a first extraction unit, used to extract the target's respiratory signal and heartbeat signal from the filtered echo signal using a variational mode decomposition method; and a second extraction unit, used to extract the target's respiratory rate and heartbeat rate from the filtered echo signal using a spatial spectrum estimation method.

[0075] Example 3

[0076] refer to Figure 5 In a third aspect, the present invention provides an electronic device comprising: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method for constructing a prediction model for COPD based on millimeter-wave radar in the first aspect of the present invention.

[0077] Electronic device 500 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 502 or a program loaded from storage device 508 into random access memory (RAM) 503. RAM 503 also stores various programs and data required for the operation of electronic device 500. The processing unit 501, ROM 502, and RAM 503 are interconnected via bus 504. Input / output (I / O) interface 505 is also connected to bus 504.

[0078] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 508 including, for example, hard disks; and communication devices 509. Communication device 509 allows electronic device 500 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 5 An electronic device 500 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 5 Each box shown can represent a device or multiple devices as needed.

[0079] Specifically, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by a processing device 501, it performs the functions defined in the methods of embodiments of this disclosure. It should be noted that the computer-readable medium described in embodiments of this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. Program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0080] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more computer programs, which, when executed by the electronic device, cause the electronic device to:

[0081] Computer program code for performing the operations of embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, C++, and Python—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0082] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0083] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing a predictive model for COPD based on millimeter-wave radar, characterized in that, include: The echo signal of the millimeter-wave radar is used to perform step tests on the target, and the motion state of the target is determined in real time based on the echo signal. Adjust the filtering method of the echo signal according to the motion state of the target; The target's vital signs are extracted from the filtered echo signal using variational mode decomposition (VMD) and spatial spectrum estimation methods. Specifically, the VMD method extracts the target's respiratory and heart rate signals from the filtered echo signal; the spatial spectrum estimation method extracts the target's respiratory and heart rate frequencies from the filtered echo signal. The VMD method for extracting the respiratory and heart rate signals from the filtered echo signal includes: decomposing the filtered echo signal into multiple components and establishing a constrained variational model based on these components; transforming the constrained variational model into an unconstrained variational model using a quadratic penalty term and Lagrange multipliers; and then, based on the quadratic penalty... The unconstrained variational model is iterated multiple times using terms and Lagrange multipliers until a preset convergence condition is met to obtain the optimal solution of the unconstrained variational model. Based on the optimal solution, the respiratory signal and heartbeat signal are separated. The step of extracting the target's respiratory frequency and heartbeat frequency from the filtered echo signal using the spatial spectrum estimation method includes: constructing the covariance matrix of the filtered echo signal and performing eigenvalue decomposition on the covariance matrix to obtain the signal subspace and noise subspace; constructing a cost function based on the signal subspace and noise subspace; and determining the respiratory frequency and heartbeat frequency through the peak value of the cost function. The vital signs data include velocity, acceleration, respiratory frequency, and heartbeat frequency. A COPD dataset was constructed based on vital signs data from multiple targets; a COPD prediction model was then built using the random forest method based on the COPD dataset.

2. The method for constructing a prediction model for COPD based on millimeter-wave radar according to claim 1, characterized in that, The method for adjusting the filtering of the echo signal according to the motion state of the target includes: If the target is stationary, the phase and frequency of the echo signal are filtered.

3. The method for constructing a prediction model for COPD based on millimeter-wave radar according to claim 1, characterized in that, The step of determining the target's motion state in real time based on the echo signal includes: The echo signal and the transmitted signal of the millimeter radar wave are mixed and filtered to obtain the intermediate frequency signal; Fourier transform and spectrum analysis are performed on the intermediate frequency signal to obtain the target's distance and velocity data; Based on the distance data and the velocity data, the motion state of the target is determined.

4. A system for constructing a prediction model for COPD based on millimeter-wave radar according to claim 1, characterized in that, include: The determination module is used to determine the motion state of the target in real time based on the echo signal of the millimeter-wave radar used for step testing of the target. The adjustment module is used to adjust the filtering method of the echo signal according to the motion state of the target; The extraction module is used to extract target vital signs data from the filtered echo signal using variational mode decomposition (VMD) and spatial spectrum estimation methods: using VMD, it extracts the target's respiratory and heartbeat signals from the filtered echo signal; using spatial spectrum estimation, it extracts the target's respiratory and heartbeat frequencies from the filtered echo signal. Specifically, extracting the target's respiratory and heartbeat signals from the filtered echo signal using VMD includes: decomposing the filtered echo signal into multiple components and establishing a constrained variational model based on the multiple components; transforming the constrained variational model into an unconstrained variational model using a quadratic penalty term and Lagrange multipliers; and based on... The unconstrained variational model is iterated multiple times using a penalty term and Lagrange multipliers until a preset convergence condition is met, yielding the optimal solution of the unconstrained variational model. Based on the optimal solution, the respiratory signal and heartbeat signal are separated. The extraction of the target's respiratory rate and heartbeat rate from the filtered echo signal using a spatial spectrum estimation method includes: constructing a covariance matrix for the filtered echo signal and performing eigenvalue decomposition on the covariance matrix to obtain a signal subspace and a noise subspace; constructing a cost function based on the signal subspace and noise subspace; and determining the respiratory rate and heartbeat rate through the peak value of the cost function. The vital signs data include velocity, acceleration, respiratory rate, and heartbeat rate. The module is used to build a COPD dataset based on vital signs data from multiple targets; and to build a COPD prediction model based on the COPD dataset using the random forest method.

5. An electronic device, comprising: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method for constructing a prediction model based on millimeter-wave radar and COPD as described in any one of claims 1 to 3.

6. A computer-readable medium having a computer program stored thereon, wherein, When the computer program is executed by the processor, it implements the method for constructing a prediction model for COPD based on millimeter-wave radar as described in any one of claims 1 to 3.

Images