A quantitative assessment method for breathing training

Abstract

This invention discloses a quantitative assessment method for breathing training. This method collects physiological parameters related to heart rate, respiration, and brain function from a human physiological data acquisition platform. The features are then subjected to dimensionality reduction, dimensionless processing, and weighted summation to obtain the quantitative assessment result of breathing training. The specific steps are as follows: First, physiological features related to heart rate, respiration, and brain function are collected through the physiological parameter acquisition platform. Then, the classification weights of each dimension of features are calculated using random forest and cross-validation, and redundant feature dimensions are removed. Next, the dimensions of features in the feature subset are removed using a dimensionless method. Finally, the feature data are weighted and summed to obtain the quantitative assessment result of breathing training. Quantitative assessment of breathing training can help people develop more scientific breathing training strategies and methods.

Description

Technical Field

[0001] This invention relates to the field of signal processing technology, and in particular to a method for quantitative assessment of breathing training. Background Technology

[0002] Respiration is a chemical process in which cells break down organic matter through mitochondria to produce energy, and it is closely related to human physiological health. Insufficient breathing leads to insufficient oxygen levels in the body, causing nutrient deficiency in cells and potentially resulting in anxiety, depression, and other problems. Breathing training, through controlled breathing exercises, helps restore and enhance the body's physiological and psychological functions. Breathing training is also a widely used intervention to help individuals cope with fatigue and stress. Furthermore, in the process of learning wind instruments, breathing training can improve a performer's technique and endurance, and significantly improve their pitch and tone. However, improper breathing training methods and intensity can also cause harm. Therefore, it is necessary to assess breathing training to guide the formulation and implementation of strategies.

[0003] Currently, the mainstream assessment method for breathing training involves evaluating vital capacity and respiratory muscle-related indicators. This method requires specialized medical equipment for respiratory muscle measurement, making it relatively complex. Furthermore, current breathing training indicators primarily focus on respiratory functions, lacking comprehensive, multi-dimensional information. Additionally, the interpretation of current breathing training indicators is complex, making it difficult to intuitively obtain information about the intensity of breathing training from the test data. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies, such as complex data acquisition and difficulty in interpreting indicators. It constructs a human physiological parameter acquisition platform to collect human physiological signals. Random forests and cross-validation are used to reduce the dimensionality of features and obtain the importance weights of feature subsets. Dimensionlessness is used to remove the signal dimensions. The importance weights of the feature subsets are used to perform a weighted summation of each signal to obtain a quantitative score for breathing training. This achieves low-cost data acquisition and objective, perceptible assessment results. By reflecting human physiological state through respiration, vital capacity, and brain function characteristics, the accuracy of breathing training assessment is enhanced. By reducing the dimensionality of high-dimensional human physiological features and extracting importance weights, the reduced features are compressed to the same dimension, obtaining a feature subset highly correlated with breathing training and removing the feature dimension, thus realizing a quantitative score for breathing training and effectively achieving a quantitative assessment of breathing training.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A quantitative assessment method for breathing training involves collecting multi-dimensional physiological parameter features of the human body through a human physiological parameter acquisition platform, reducing the dimensionality of the collected features and removing redundant feature dimensions, removing the dimensions of each feature in the feature subset, and obtaining an accurate quantitative score for breathing training. The assessment method includes the following steps:

[0007] S1. Collect multi-dimensional human physiological parameter characteristics through a human physiological parameter acquisition platform;

[0008] S2. Obtain feature subsets and their weights through random forest and cross-validation;

[0009] S3. Remove the dimensions of each feature in the feature subset by dimensionless transformation;

[0010] S4. Obtain the quantitative assessment results of breathing training by weighted summation of feature weights.

[0011] Furthermore, the human physiological parameter acquisition platform in step S1 consists of a control module, a heart rate signal acquisition module, a respiratory signal acquisition module, and a brain function signal acquisition module. The control module guides the testing process, processes and analyzes the acquired raw data, and scores it. The heart rate signal acquisition module acquires raw heart rate signal data. The respiratory signal acquisition module acquires raw data related to lung capacity signal through the lung capacity acquisition unit and raw data related to expiratory airflow signal through the expiratory airflow acquisition unit. The raw data of lung capacity and expiratory airflow signal are combined to form complete raw respiratory signal data. The brain function signal acquisition module acquires raw brain function signals.

[0012] The process of using the control module to guide the parameter acquisition process and drive the heart rate signal acquisition module, respiratory signal acquisition module, and brain function signal acquisition module to acquire raw parameters of multi-dimensional human physiological parameters, and then processing and analyzing the raw parameters to obtain the characteristics of multi-dimensional human physiological parameters, is as follows:

[0013] S101. The control module guides the user through the raw heart rate signal acquisition process. The heart rate signal acquisition module acquires facial video through a recording device. The control module processes the facial video using the RPPG algorithm to extract heart rate-related features. The steps are as follows:

[0014] S101.1 Under sufficient lighting conditions, the control module prompts the user to keep their heart rate stable and their body still. At the same time, the heart rate signal acquisition module uses a recording device to capture facial video in the background. After the facial video is captured, the control module completes the subsequent signal processing and analysis.

[0015] S101.2. The acquired face video is segmented into frames to obtain a face video frame sequence, which is beneficial for subsequent face recognition and face feature point detection tasks.

[0016] S101.3 Perform face recognition and face feature point detection on the face video image sequence, and locate the region of interest based on the face feature points;

[0017] S101.4 Extract the color channel values ​​of the pixels in the region of interest as the raw pulse wave signal;

[0018] S101.5. Perform mode decomposition on the obtained raw pulse wave signal to obtain multiple Intrinsic Mode Function (IMF) components. Perform Fast Fourier Transform on each IMF component, calculate the frequency corresponding to the maximum peak in the spectrum, and classify each component into high-frequency components and low-frequency components according to a certain frequency threshold.

[0019] S101.6. The selected high-frequency IMF components are denoised by wavelet threshold filtering to remove noise from the high-frequency signal.

[0020] S101.7 Perform baseline drift removal on the selected low-frequency IMF to remove the low-frequency components obtained through mode decomposition.

[0021] Offset in the signal caused by instrument drift, environmental changes, or other low-frequency factors;

[0022] S101.8. The high-frequency IMF component processed in step S103.6 is superimposed with the low-frequency IMF component processed in step S103.7 to reconstruct the denoised signal.

[0023] S101.9. Bandpass filtering is performed on the denoised signal to enhance the signal-to-noise ratio and obtain the final pulse wave signal of the region of interest.

[0024] S101.10 Extract heart rate-related features from the final pulse wave signal using time-domain features;

[0025] S102. The control module guides the user to complete the respiratory-related parameter acquisition process. The respiratory signal acquisition module drives the vital capacity acquisition unit to acquire the user's raw vital capacity signal and drives the airflow signal acquisition unit to acquire the user's raw expiratory airflow signal. After processing and analysis, the control module obtains vital capacity characteristics and expiratory airflow characteristics, which are combined to form complete respiratory-related features. The specific steps are as follows:

[0026] S102.1 The control module prompts the user to take a deep breath in a relaxed and calm state, and then use the lung capacity acquisition unit to exhale quickly to the limit. The lung capacity acquisition unit collects and records the exhaled airflow signal when the user exhales quickly.

[0027] S102.2 The control module processes and analyzes the expiratory airflow signal during rapid exhalation to obtain the user's vital capacity characteristics;

[0028] S102.3 The control module prompts the user to take a deep breath in a relaxed and calm state, and then use the airflow signal acquisition unit to breathe slowly, steadily, and for a long time until the limit. The airflow signal acquisition unit collects and records the exhaled airflow signal when the user exhales steadily for a long time.

[0029] S102.4 The control module processes and analyzes the expiratory airflow signal during a long period of stable exhalation by the user to obtain the user's expiratory airflow characteristics.

[0030] S102.5 The control module summarizes the characteristics related to vital capacity and expiratory airflow to obtain complete respiratory characteristics;

[0031] S103. The control module guides the user to complete the memory test. The brain function acquisition module displays symbols on the display device, and the user memorizes the symbols by inputting them into the input device. The control module calculates the user's memory accuracy as a brain function characteristic. The specific steps are as follows:

[0032] S103.1 The control module prompts the user to look at the display device when they are in a good mental state. The display device sequentially displays any number of numbers, characters, graphics and other characteristic symbols, and the user memorizes these symbols and their corresponding order.

[0033] S103.2 After a certain period of time, these symbols disappear;

[0034] S103.3 The user inputs the memorized symbols and their corresponding order through the input device within a certain time range;

[0035] S103.4 Repeat steps S103.1-S103.3 for a certain number of rounds;

[0036] S103.5 The control module calculates the accuracy of the user's input characters and uses the accuracy as a characteristic of the user's brain function.

[0037] S104. After the control module completes the processing of the raw data related to heart rate, respiration and brain function, it analyzes and obtains the features of each dimension, and obtains and displays the quantitative score of breathing training through subsequent processing.

[0038] Furthermore, the process of obtaining the feature subset and its weights in step S2 is as follows:

[0039] S201. In step S1, evaluation features of various physiological parameters were obtained. Feature subsets were constructed by randomly sampling these features. Random sampling was performed multiple times to obtain feature subsets with different numbers of features.

[0040] S202. Evaluate and score each feature subset. Evaluation metrics can include common feature evaluation metrics such as precision and accuracy. Precision represents the proportion of positive samples that are actually positive. Recall represents the proportion of samples that are correctly predicted as positive out of all samples that are actually positive. For scoring, F1-weighted is the most commonly used feature subset scoring metric. F1-weighted comprehensively considers factors such as class frequency, and the final score is proportional to the importance of the feature subset. After completion, obtain the correspondence between the number of features in different feature subsets and their scores.

[0041] S203. Based on the correspondence between the number of features and the score in step S202, calculate the number of features when the score is the highest.

[0042] S204. Process the features by removing different features and calculating scores. Calculate the importance weight of each feature based on the changes in scores. Remove the feature with the lowest weight. Iterate multiple times until the number of features is the same as the number of features with the highest scores in step S203. This reduces the influence of irrelevant features, prevents the curse of dimensionality, and finally obtains the feature subset with the strongest correlation.

[0043] S205. By removing different features and calculating scores, the importance weight of each feature in the feature subset obtained in step S204 is calculated based on the changes in scores.

[0044] Furthermore, step S3 involves removing the dimensions of each feature in the feature subset through dimensionless transformation, as follows:

[0045] S301. For different feature dimensions in different feature subsets in step S2, analyze the corresponding feature distribution characteristics, transform the feature direction, and offset the feature value through the benchmark value to transform each feature into a value range that is positive and positively correlated with the human physiological state. For example, the heart rate value is negatively correlated with the human physiological state. The transformed feature is obtained by taking its negative value and adding an offset value.

[0046] S302. By using the relationship between the difference between the feature values ​​and the feature extrema, the range of values ​​of all features is compressed so that the range of values ​​of all features is between [0,1], thus obtaining dimensionless features that are comparable in the same dimension.

[0047] Furthermore, the process of obtaining the quantitative evaluation result of breathing training through weighted summation of feature weights in step S4 is as follows:

[0048] S401. Compare the importance weight percentage of each parameter feature with the corresponding parameter feature dimension obtained in step S205.

[0049] Multiply them to obtain the weighted values ​​of each parameter;

[0050] S402. Add the weighted values ​​of each parameter to obtain a quantitative assessment score for breathing training. The score ranges from [0,1].

[0051] S403, the human physiological parameter acquisition platform displays the quantitative assessment score of breathing training.

[0052] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0053] (1) This invention collects raw data related to human heart rate, respiration and brain function through a low-cost and easy-to-use heart rate, exhalation and brain function physiological parameter acquisition module, which simplifies the acquisition process of relevant parameters in the respiratory training assessment process and reduces the need for complex medical instruments.

[0054] (2) This invention simplifies the process of collecting multi-dimensional physiological parameters through the guidance of the control module, while improving the accuracy of test results. With the guidance of the control module, users can easily and independently complete the entire testing process without additional guidance or instruction, reducing the complexity of the entire system and solving the problem of complex physiological parameter collection.

[0055] (3) By combining cross-validation and feature selection, this invention calculates the importance of multi-dimensional physiological parameter features. Based on the number of the optimal feature subsets obtained by cross-validation, the least important physiological parameter features are removed one by one by feature elimination, thereby removing redundant feature data dimensions and improving computational efficiency and accuracy.

[0056] (4) This invention removes the dimension of features by dimensionless transformation and obtains the quantitative score of breathing training by weighted summation of the importance weights of each feature. The quantitative score can help people establish an accurate and intuitive understanding of the intensity of breathing training and help people make more scientific decisions. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is a flowchart of a quantitative assessment method for breathing training in an embodiment of the present invention;

[0059] Figure 2 This is an overall structural diagram of the human physiological signal acquisition platform in an embodiment of the present invention;

[0060] Figure 3This is a green channel waveform diagram of the original pulse wave signal in an embodiment of the present invention;

[0061] Figure 4 This is a schematic diagram of the result of peak detection of the filtered signal in an embodiment of the present invention;

[0062] Figure 5 This is a schematic diagram of the change signal of exhaled air pressure relative to atmospheric pressure when the human body blows air steadily in an embodiment of the present invention;

[0063] Figure 6 This is the result of truncating the portion of the air pressure signal with a value of 0 in the time domain in the embodiments of the present invention;

[0064] Figure 7 This is a waveform diagram of the air pressure signal after removing the unstable parts at the beginning and end stages in an embodiment of the present invention;

[0065] Figure 8 This is a waveform diagram of the airflow velocity signal in an embodiment of the present invention;

[0066] Figure 9 This is a schematic diagram illustrating the correspondence between the number of features in different feature subsets and the f1-weighted scores in embodiments of the present invention. Detailed Implementation

[0067] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative effort are within the scope of protection of the present application.

[0068] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application can be combined with other embodiments.

[0069] Example

[0070] This embodiment discloses a quantitative assessment method for breathing training, such as... Figure 1 As shown, the specific steps are as follows:

[0071] S1. Collect multi-dimensional physiological parameters of the human body through the human physiological parameter acquisition platform. The human physiological parameter acquisition platform consists of a control module combined with raw signal acquisition modules for heart rate, respiration, and brain function. It guides the user to collect and analyze raw data to obtain multi-dimensional physiological parameter characteristics of the human body. The specific steps are as follows:

[0072] S101, the control module is a computer program that guides the entire testing process. It also processes and analyzes the raw data collected by the heart rate, respiration, and brain function signal acquisition modules to obtain corresponding features. The heart rate signal acquisition module captures facial video via a computer camera; the control module then analyzes this data to obtain heart rate-related features. In the respiration signal acquisition module, the lung capacity acquisition unit collects air pressure signals during full exhalation using a pressure sensor, while the expiratory airflow acquisition unit collects air pressure signals during steady exhalation using a pressure sensor. The control module calculates airflow velocity signals using the formula relating airflow and air pressure, thereby obtaining lung capacity and airflow-related features. The brain function signal acquisition module collects raw signals of human memory ability through a memory UI testing program; the control module analyzes the memory accuracy to obtain human brain function features. After obtaining multi-dimensional physiological characteristics, the control module performs comprehensive processing and analysis to obtain a quantitative score for the breathing training. Figure 2 This is a structural diagram of the human physiological signal acquisition platform;

[0073] S102. The control module prompts the user to look directly at the computer screen. Once the user is ready, the test process will begin.

[0074] S103, the screen displays a 2-minute countdown, while the heart rate signal acquisition module drives the computer's webcam in the background.

[0075] Capture user face video signals at 1920×1080 resolution and 30FPS;

[0076] S104. After the screen countdown ends, the control module records the face video signal and obtains human heart rate-related features through the RPPG algorithm. The specific steps are as follows:

[0077] S104.1. The acquired face video is divided into frames according to FPS to obtain a face video frame sequence of 30 frames per second;

[0078] S104.2. Using the machine learning face recognition library MediaPipe, face video images are individually subjected to face recognition and face feature point detection. After obtaining 68 facial key points, the region of interest is located based on the face feature points.

[0079] S104.3 Extract the green pixel color channel values ​​of the region of interest and calculate their average value as the signal value. Figure 3 The waveform of the green channel of the original pulse wave signal;

[0080] S104.4 Perform empirical mode decomposition on the preprocessed pulse wave signal to obtain 5 intrinsic mode function components. Perform fast Fourier transform on each IMF component to calculate the frequency corresponding to the maximum peak value. Use a certain frequency threshold as a classification to divide the IMF component into 4 high-frequency components and 1 fundamental frequency component.

[0081] S104.5. Wavelet denoising is performed on the four selected high-frequency IMF components. The wavelet basis function in the wavelet threshold denoising is the "db9" wavelet, the wavelet decomposition level is 3, the threshold function is the global hard threshold function, and the threshold is 0.1 times the maximum value of the signal.

[0082] S104.6. For the selected low-frequency IMF, baseline fitting is performed using a quadratic polynomial regression model, and baseline drift is removed to obtain the drift-free signal.

[0083] S104.7. The four high-frequency IMF components processed in step S104.5 are linearly summed with the one low-frequency IMF component processed in step S104.6 to obtain a clean, denoised signal.

[0084] S104.8. The reconstructed pulse wave signal is processed using a Butterworth bandpass filter, with the filter order set to 5 and the filter bandpass frequency set to 0.6~3.1Hz.

[0085] S104.9. By detecting the peaks in the bandpass filtered signal using a peak detection algorithm, and obtaining the peak values, the heart rate can be calculated by counting the number of peaks within one minute. The instantaneous heart rate and related physiological parameters such as heart rate variability can be calculated by analyzing the time intervals between adjacent peaks. Figure 4 This diagram illustrates the results of peak detection for the filtered signal.

[0086] S105. The control module guides the user to collect the user's breathing-related signals through the breathing signal acquisition module. The specific steps are as follows:

[0087] S105.1 After the control module drives the air pressure sensor in the lung capacity acquisition unit to start collecting data for a few seconds, it guides the user to relax the body and take a deep breath, and then blow air quickly into the air pressure sensor. After blowing to the limit, stop. After stopping for a few seconds, the control module stops recording the signal value of the air pressure sensor.

[0088] S105.2 The control module calculates the change in air pressure in the air pressure signal relative to atmospheric pressure;

[0089] S105.3, The formula for calculating the equation of state for an ideal gas is:

[0090] Where P represents pressure and V represents gas volume. Indicates the number of moles of gas. Let T represent the ideal gas constant, and T represent the temperature.

[0091] When blowing air, the formulas for calculating the gas pressure and the airflow velocity are as follows:

[0092] in Indicates gas pressure. This indicates the density of a gas (unit: kg / m³). This indicates the gas flow rate (unit: m / s).

[0093] The control module obtains the exhaled airflow signal when the user blows air with all their might through the formula relating airflow and air pressure. Then, by integrating the airflow velocity over time and multiplying it by the exhalation cross-section, the user's vital capacity characteristics are calculated.

[0094] S105.4 After the control module drives the air pressure sensor in the exhalation airflow signal acquisition unit to start collecting data for a few seconds, it guides the user to relax the body and take a deep breath, and then blow air smoothly and slowly into the air pressure sensor. After blowing to the limit, stop. After stopping for a few seconds, the control module stops recording the signal value of the air pressure sensor.

[0095] S105.5 The control module calculates the change in air pressure relative to atmospheric pressure in the air pressure signal. Figure 5 A schematic diagram showing the change in expiratory air pressure relative to atmospheric pressure during steady exhalation of a human body;

[0096] S105.6 Calculate the time-domain values ​​of the barometric pressure signal change where the signal value is 0 at the beginning and end. Truncate the signal in the time domain to remove the signal periods at the beginning and end where there is no exhalation. Figure 6 The result is the barometric pressure signal after truncating the portion of the signal with a value of 0 in the time domain.

[0097] S105.7. Truncate 5% of the signal in the time domain at the beginning and end of the acquisition phases, removing the unstable portions at the start and end of the acquisition, and retaining the stable portions of the signal. Figure 7 The waveform of the air pressure signal after removing the unstable parts at the beginning and end;

[0098] S105.8 The control module obtains the airflow velocity signal when the user blows air slowly and steadily by measuring the relationship between airflow and air pressure. Figure 8 This is a waveform diagram of the airflow velocity signal;

[0099] S105.9 The control module analyzes the airflow velocity signal when the user blows air slowly and steadily and obtains airflow-related characteristics;

[0100] S106. The control module guides the user to conduct a memory test through the memory UI test program. The specific process is as follows:

[0101] S106.1 The memory UI test program displays a 20-digit random number on the computer screen, while a countdown timer is displayed on the screen for 20 seconds, allowing the user to memorize the number.

[0102] S106.2 After the countdown ends, the numbers disappear, and an input box is displayed. At the same time, the countdown restarts for 20 seconds, during which the user copies the numbers from memory.

[0103] S106.3 When the countdown ends, user input ceases and no further input is possible, thus ending one round of the experiment;

[0104] S106.4 Experimental steps S106.1-S106.3 are repeated 3 times. The memory UI test program records the random numbers displayed in each round and the corresponding input numbers of the user.

[0105] S106.5 The control module calculates the average accuracy of the user's memory in three rounds of experiments as an evaluation feature of brain function indicators.

[0106] S2. Through random forest and cross-validation, redundant data dimensions are removed to obtain feature subsets and their weights. The specific steps are as follows:

[0107] S201. In step S1, evaluation features of various physiological parameters were obtained. Feature subsets were constructed by randomly sampling these features. Random sampling was performed multiple times to obtain feature subsets with different numbers of features.

[0108] S202. Evaluate and score each feature subset. Evaluation metrics include common feature evaluation metrics such as precision and accuracy. Scoring is done using f1-weighted as the feature subset scoring metric. After completion, obtain the correspondence between the number of features and the scores in different feature subsets. Figure 9 The correlation between the number of features in different feature subsets and the f1-weighted scores is shown;

[0109] S203. Based on the correspondence between the number of features and the score in step S202, it is calculated that the highest score is achieved when there are 8 features, and the score is also relatively high when there are 4-5 features.

[0110] S204. The features are processed by removing different features and calculating scores. The importance weight of each feature is calculated from the changes in scores. The feature with the lowest weight is removed. This process is repeated multiple times until the number of features is 8. At this point, the feature parameter set consists of 8 parameters: average airflow velocity, airflow skewness, root mean square velocity, average RR interval, maximum vital capacity, airflow velocity coefficient of variation, heart rate per minute, and memory accuracy. Considering that the scores when the number of features is 5 are closer to those when the number of features is 8, and that the airflow velocity-related data in the feature parameter set may have strong cross-correlation, the average airflow velocity, average RR interval, maximum vital capacity, heart rate per minute, and memory accuracy were ultimately selected as the evaluation indicators.

[0111] S205. By removing different features and calculating scores, the importance weights of each feature in the feature subset from step S204 are calculated from the changes in scores as follows:

[0112] Table 1. Weighting of the Importance of Each Feature

[0113]

[0114] S3. Remove the dimensions of each feature in the feature subset by dimensionless transformation:

[0115] S301. For different feature dimensions in different feature subsets in step S2, the corresponding dimensionless transformation methods shown in the table below are used for processing:

[0116] Table 2. Dimensionless Table

[0117]

[0118] After dimensionless processing, the value range of each dimension feature data is compressed to the [0,1] interval, and all of them are positively correlated with breathing training;

[0119] S4. Obtain the quantitative evaluation results of breathing training by weighted summation of feature weights:

[0120] S401. Multiply each parameter feature by the importance weight percentage of the corresponding parameter feature dimension to obtain the weighted value of each parameter.

[0121] S402. Add the weighted values ​​of each parameter together to obtain a quantitative assessment score for breathing training.

[0122] S403 The control module displays the quantitative assessment score of the breathing training. Through this assessment score, users can accurately understand the intensity of the current breathing training, confirm whether the breathing training is reasonable, and formulate a more scientific breathing training execution strategy.

[0123] In summary, addressing the issues of complex and costly physiological parameter testing, this invention proposes for the first time to characterize and calculate breathing training using simple and low-cost heart rate, respiration, and brain function-related features, eliminating the complex testing process for respiratory muscle-related parameters. The human physiological parameter acquisition platform simplifies the physiological parameter testing process, achieving rapid and low-cost data collection. By introducing heart rate-related and brain function-related features, and adding expiratory airflow-related features to the traditional lung capacity feature, the comprehensiveness and scientific rigor of the evaluation indicators are enhanced. Addressing the problem of complex interpretation of breathing training indicators hindering the large-scale promotion of breathing training, this invention uses physiological parameters to quantitatively assess the intensity of user breathing training. The quantification standard objectively and accurately reflects the human physiological state, providing an objective, realistic, and perceptible assessment result, which helps users make more scientific adjustments during breathing training.

[0124] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

[0125] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A method for quantitative assessment of breathing training, characterized in that, The quantitative assessment method for breathing training includes the following steps: S1. Collect multi-dimensional human physiological parameter features through a human physiological parameter acquisition platform. The process is as follows: S101, The control module guides the user to complete the raw heart rate signal acquisition process. The heart rate signal acquisition module acquires facial video through the recording device. The control module processes the facial video and extracts heart rate-related features. S102. The control module guides the user to complete the process of acquiring respiratory-related parameters. The respiratory signal acquisition module drives the vital capacity acquisition unit to acquire the user's raw vital capacity signal and drives the airflow signal acquisition unit to acquire the user's raw expiratory airflow signal. After processing and analysis, the control module obtains vital capacity characteristics and expiratory airflow characteristics, which are combined to form complete respiratory-related characteristics. S103, The control module guides the user to complete the memory test. The brain function acquisition module displays symbols on the display device. The user memorizes the symbols and inputs them into the input device. The control module calculates the user's memory accuracy as a brain function characteristic. S104. After the control module completes the processing of the raw data related to heart rate, respiration and brain function, it analyzes and obtains the features of each dimension, and obtains and displays the quantitative score of breathing training through subsequent processing. S2. Obtain the feature subset and its feature weights through random forest and cross-validation, as follows: S201. Construct feature subsets from the above-mentioned heart rate-related features, respiratory-related features, and brain function features by random sampling. S202. Evaluate and score each feature subset. The final score is proportional to the importance of the feature subset. After completion, obtain the correspondence between the number of features in different feature subsets and the score. S203. Based on the correspondence between the number of features and the score, calculate the number of features when the score is the highest. S204. Process the features by removing different features and calculating scores. Calculate the importance weight of each feature from the changes in scores, remove the features with the lowest weights, iterate multiple times, and finally obtain the feature subset with the strongest correlation. S205. By removing different features and calculating scores, the importance weight of each feature in the feature subset obtained in step S204 is calculated based on the changes in scores. S3. Remove the dimensions of each human physiological parameter feature in the feature subset by dimensionless transformation; S4. The physiological parameters of each human body processed in step S3 are weighted and summed using feature weights to obtain the quantitative evaluation results of breathing training.

2. The quantitative assessment method for breathing training according to claim 1, characterized in that, In step S1, the human physiological parameter acquisition platform consists of a control module, a heart rate signal acquisition module, a respiratory signal acquisition module, and a brain function signal acquisition module. The control module guides the parameter acquisition process and drives the heart rate signal acquisition module, respiratory signal acquisition module, and brain function signal acquisition module to acquire multi-dimensional human physiological parameter raw parameters, and processes and analyzes the raw parameters to obtain multi-dimensional human physiological parameter characteristics.

3. The quantitative assessment method for breathing training according to claim 1, characterized in that, The process of removing the dimensions of each feature in the feature subset through dimensionless transformation in step S3 is as follows: S301. For different feature dimensions, analyze the corresponding value distribution characteristics. By transforming the feature direction and shifting the feature value through the benchmark value, each feature is transformed into a positive value range, which is positively correlated with the human physiological state. S302. Compress the value range of all features so that the value range of all features is between [0,1].

4. The quantitative assessment method for breathing training according to claim 3, characterized in that, The process of obtaining the quantitative score for breathing training in step S4 is as follows: S401. Multiply each human physiological parameter feature by the importance weight of the human physiological parameter feature obtained in step S205 to obtain the weighted value of each human physiological parameter feature. S402. The weighted values ​​of each human physiological parameter are summed to obtain a quantitative assessment score for breathing training. The score ranges from [0,1]. S403, the human physiological parameter acquisition platform displays the quantitative assessment score of breathing training.

Images