Limb ischemia pre-adaptation training method and device

Abstract

The invention relates to the technical field of ischemia training, in particular to a limb ischemia pre-adaptation training method and device.The limb ischemia pre-adaptation training method comprises the steps that feature parameters, training purposes and cuff feature parameters of an individual AOP of a target user are collected; inputting the obtained feature parameters of the individual AOP of the target user, the training purpose and the feature parameters of the cuff into a pressure prediction model, and outputting a pressure value; inputting the obtained blood vessel health coefficient, the limb perimeter, the pressure value and the training purpose into a pressure blocking time prediction model, and outputting pressure blocking time; according to the method, the optimal pressure value under different training scenes can be dynamically predicted, so that the physiological requirements of different training targets are met, and then the blood vessel health coefficient and pressure application blocking time prediction model is introduced, so that the blood vessel health coefficient and pressure application blocking time prediction accuracy is improved. The training safety is further optimized, and tissue damage caused by too large pressure or too long time is avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ischemic training, in particular to a limb ischemic preconditioning training method and device. BACKGROUND

[0002] Ischemic preconditioning training is a physical treatment method that stimulates the body's endogenous protection mechanism through a short and repeated process of limb blood flow blockage-reperfusion, thereby improving the tolerance of tissues to ischemia and hypoxia. Its main forms include local ischemic preconditioning and remote (contralateral) ischemic preconditioning, which can be used for the prevention and rehabilitation of cardiovascular and cerebrovascular diseases, the improvement of athletes' endurance, and the prevention of high altitude reactions.

[0003] Chinese patent with publication number CN110613593A discloses an ischemic preconditioning training method and system based on blood pressure value adjustment. First, the blood pressure value in the state of not performing ischemic preconditioning is obtained, and it is judged whether this blood pressure value belongs to a user who can use an ischemic training instrument. Then, a gas bag, i.e., a blood pressure measurement module, is bound on another limb to obtain the real-time blood pressure value after training. Because this real-time blood pressure value can truly reflect the blood pressure of the user during ischemic preconditioning training, it can also reflect the physical condition of the user during ischemic preconditioning training. By mastering the real-time blood pressure value of the user, the gas pressure of ischemic preconditioning is adjusted using the real-time blood pressure value.

[0004] Chinese patent application with publication number CN119014832A discloses an ischemic preconditioning training method, device, equipment and medium, relating to the technical field of medical and health care instruments. The method includes: obtaining a current blood pressure value and a current respiratory frequency value of a target user who has not performed ischemic preconditioning training operation; if the current blood pressure value and the current respiratory frequency value meet the operation condition of performing ischemic preconditioning training operation, selecting a target training strategy from a preset ischemic preconditioning training strategy of a preset ischemic preconditioning training instrument based on the current pressure tolerance value of the wearing part information of the target user who selects to wear the arm band on the joint part or the limb part of the four limbs, and performing ischemic preconditioning training on the target user in accordance with each strategy condition in the target training strategy.

[0005] As in the above-mentioned application, in the existing technology, during the training process of ischemic preconditioning training, the blood pressure value is generally relied on as the only basis for adjusting the pressure of ischemic preconditioning training, while the differences in individual physiological characteristics, such as limb circumference, subcutaneous fat thickness and cuff width, are ignored. Using only the blood pressure value as the basis may lead to inaccurate calculation of the pressure, for example, for users with thick limbs or thick subcutaneous fat, the actual required pressure may be significantly higher than the predicted value based on the blood pressure value, resulting in incomplete arterial blood flow blockage or excessive pressure causing discomfort. In addition, the different needs for pressure for different training purposes are not distinguished, resulting in a lack of targeted training strategies and affecting the training effect and safety. SUMMARY

[0006] To solve the above problems, the application provides a limb ischemic preconditioning training method.

[0007] The application adopts the following technical scheme, a limb ischemic preconditioning training method, comprising:

[0008] Collecting the characteristic parameters of individual AOP of the target user, the training purpose and the cuff characteristic parameters;

[0009] Inputting the obtained characteristic parameters of individual AOP of the target user, the training purpose and the cuff characteristic parameters into a pressure prediction model, and outputting a pressure value;

[0010] Collecting the blood vessel health coefficient of the target user;

[0011] Inputting the obtained blood vessel health coefficient, limb circumference and pressure value, and training purpose into a pressure blocking time prediction model, and outputting a pressure blocking time;

[0012] Performing a target action with a preset cycle number on the limb of the target user through the ischemic preconditioning training instrument, wherein the target action is ischemic preconditioning training through the ischemic preconditioning training instrument based on the output pressure blocking time and pressure value.

[0013] Further description of the above technical scheme: the limb ischemic preconditioning training method further comprises:

[0014] Real-time acquisition of physiological signal parameters of the target user before and during ischemic preconditioning training;

[0015] Constructing a prediction training risk score model according to the reaction change of the physiological signal parameters during training, and obtaining a training risk score;

[0016] Based on the obtained training risk score and blood vessel health coefficient, dynamically adjusting the next round of training pressure value and pressure blocking time, and generating an optimized pressure value and pressure blocking time.

[0017] Performing a target action with a preset cycle number on the limb of the target user through the ischemic preconditioning training instrument based on the optimized pressure value and pressure blocking time.

[0018] Further description of the above technical scheme: the characteristic parameters of individual AOP include resting systolic blood pressure, limb circumference, and subcutaneous fat thickness, wherein the limb circumference is the limb circumference of the part for ischemic pressure training;

[0019] The training purpose includes blood flow restriction training and heart-brain preconditioning training;

[0020] The cuff characteristic parameter is a cuff width.

[0021] As a further description of the above technical solution: the training method of the pressure prediction model comprises:

[0022] A pre-constructed pressure prediction dataset is pre-processed, including data cleaning and data standardization processing, the pressure prediction dataset includes the characteristic parameters of the individual AOP of the p groups of target users, the training purposes and the cuff characteristic parameters, and the characteristic parameters of the individual AOP of the p groups of target users, the training purposes and the cuff characteristic parameters corresponding to the pressure values, p is a positive integer greater than 0, the pressure prediction dataset is divided into a pressure prediction data training set and a pressure prediction data validation set, wherein the pressure prediction data training set is used for parameter learning of the pressure prediction model, and the pressure prediction data validation set is used for real-time evaluation of the generalization ability of the pressure prediction model;

[0023] In the pressure prediction model training process, a deep neural network structure based on a multilayer perceptron is adopted, the characteristic parameters of the individual AOP of the target user, the training purposes and the cuff characteristic parameters are converted into high-dimensional feature vectors as inputs, the nonlinear features in the data are extracted through several hidden layers, and finally the probability distribution of the pressure prediction result is generated by using a softmax activation function in the output layer, and the pressure prediction value corresponding to the maximum probability is output as the final prediction result; the training process aims to minimize the cross-entropy loss function, and an early stopping strategy is introduced to monitor the performance of the pressure prediction data validation set, when the prediction accuracy on the pressure prediction data validation set reaches a preset threshold, it is considered that the pressure prediction model has converged, and the training of the pressure prediction model is stopped.

[0024] As a further description of the above technical solution: the method for obtaining the blood vessel health coefficient of the target user comprises:

[0025] The carotid-femoral pulse wave velocity is measured by an arterial stiffness instrument, and the augmentation index is automatically calculated by a pulse wave analyzer;

[0026] The standard deviation is calculated after 5min of electrocardiogram monitoring, and the heart rate variability is obtained;

[0027] The mean and standard deviation of the pulse wave velocity, the augmentation index and the heart rate variability are calculated, and the original data are converted into Z-score standardized values;

[0028] The pulse wave velocity, the augmentation index and the heart rate variability are calculated by a formula, and the blood vessel health coefficient of the target user is generated.

[0029] As a further description of the above technical solution: the training method of the pressure prediction model comprises:

[0030] Pre-construct a time prediction dataset, preprocess the collected time prediction dataset, including data cleaning and data standardization processing, the time prediction dataset includes M groups of target users' vascular health coefficients, limb circumferences and pressure values and M groups of target users' vascular health coefficients, limb circumferences and pressure values corresponding to the pressure blocking time, M is a positive integer greater than 0, and the time prediction dataset is divided into a time prediction data training set and a time prediction data validation set, wherein the time prediction data training set is used for parameter learning of the pressure blocking time prediction model, and the time prediction data validation set is used for real-time evaluation of the generalization ability of the pressure blocking time prediction model.

[0031] In the training process of the pressure blocking time prediction model, a random forest regression model is used for training. In the training process, the model will construct multiple decision trees according to the input feature data and the corresponding pressure blocking time label. Each decision tree will randomly select part of the features and samples for training when constructing. By calculating the information gain index, the best split feature and split point are selected, and the samples are gradually divided into different sub-nodes until the preset stopping condition is met. The grid search or random search method is used to adjust the hyperparameters of the random forest regression model to find the optimal hyperparameter combination and improve the performance of the model.

[0032] The trained model is evaluated using the time prediction data validation set. When the prediction accuracy on the time prediction data validation set reaches the preset threshold, it is considered that the pressure blocking time prediction model has converged, and the training of the pressure blocking time prediction model is stopped.

[0033] Further description of the above technical solutions: the method for adjusting the next round of pressure blocking time based on the training risk score and the vascular health coefficient includes:

[0034] Based on the current round of pressure blocking time, the negative inhibition of the training risk score and the positive enhancement of the vascular health coefficient are dynamically adjusted, wherein the positive enhancement of the vascular health coefficient is to gradually enhance the pressure value through an exponential decay characteristic, and the individual with a better vascular health coefficient can withstand higher training intensity, wherein the negative inhibition of the training risk score is to automatically reduce the pressure through a hyperbolic function saturation characteristic to avoid excessive physiological load due to risk accumulation, thereby obtaining the optimized pressure blocking time.

[0035] Further description of the above technical solutions: the method for adjusting the next round of pressure blocking time based on the training risk score and the vascular health coefficient includes:

[0036] Based on the predicted pressure blocking time, the positive adjustment of the vascular health coefficient and the negative inhibition of the training risk score are dynamically adjusted to generate the optimized pressure blocking time.

[0037] A limb ischemic preconditioning training device applied to a preset ischemic preconditioning training instrument, comprising:

[0038] A data acquisition module acquires characteristic parameters of individual AOP of a target user, training purposes and cuff characteristic parameters, and acquires a blood vessel health coefficient of the target user;

[0039] A pressure prediction module inputs the acquired characteristic parameters of individual AOP of the target user, training purposes and cuff characteristic parameters into a pressure prediction model, and outputs a pressure value;

[0040] A time prediction module inputs the acquired blood vessel health coefficient, limb circumference and pressure value, and training purposes into a pressure blocking time prediction model, and outputs a pressure blocking time;

[0041] A pressure training module executes a target action of a preset cycle number on the limb of the target user through the ischemic preconditioning training instrument, and the target action is ischemic preconditioning training through the ischemic preconditioning training instrument based on the output pressure blocking time and pressure value;

[0042] A dynamic parameter adjustment module acquires physiological signal parameters before and during ischemic preconditioning training of the target user in real time, constructs a prediction training risk score model according to changes in physiological signal parameters during training, acquires a training risk score, dynamically adjusts a next round of training pressure value and pressure blocking time based on the acquired training risk score and blood vessel health coefficient, generates an optimized pressure value and pressure blocking time, and executes a target action of a preset cycle number on the limb of the target user through the ischemic preconditioning training instrument based on the optimized pressure value and pressure blocking time.

[0043] An ischemic preconditioning training instrument comprises a cuff capable of being inflated and pressurized and a control box, and the control box can be arranged on the outer wall of the cuff or arranged separately from the cuff;

[0044] The control box is provided with an inflation pump, the inflation pump is communicated with a gas guide pipe, the end of the gas guide pipe is communicated with the cuff, and the inflation pump inflates the cuff through the gas guide pipe to achieve the purpose of pressure training on the limbs of the patient.

[0045] Beneficial effects:

[0046] In the technical scheme, the limb ischemic preconditioning training method provided by the application integrates individual AOP characteristic parameters (resting systolic blood pressure, limb circumference, cuff width, subcutaneous fat thickness) and training purposes (blood flow restriction training or heart and brain preconditioning training), constructs a pressure prediction model based on a deep neural network, significantly improves the accuracy and individualization of the applied pressure, extracts nonlinear features by combining a multilayer perceptron, can dynamically predict the optimal pressure value in different training scenarios, thereby meeting the physiological needs of different training goals, and further optimizes the training safety by introducing a blood vessel health coefficient and an applied pressure blockage time prediction model, thereby avoiding tissue damage caused by excessive pressure or excessive time.

[0047] Further, by collecting physiological signal parameters of the target user before and during ischemic adaptation training, a comprehensive training risk score Rf is constructed to represent the physiological load degree of the current training scheme on the individual, and the physiological load risk term, heart rate overload risk term and blood oxygen risk term are comprehensively analyzed, thereby guiding the automatic adjustment of the subsequent pressure application strategy, and realizing the dual protection of safety and adaptability. BRIEF DESCRIPTION OF DRAWINGS

[0048] The application will be further explained below in combination with the drawings and embodiments:

[0049] Figure 1 The limb ischemic preconditioning training method provided for the embodiment 1 of the application Figure 1 ;

[0050] Figure 2 The limb ischemic preconditioning training method provided for the embodiment 2 of the application Figure 2 ;

[0051] Figure 3 The module connection schematic diagram of the limb ischemic preconditioning training device provided for the embodiment 3 of the application

[0052] Figure 4 The structure schematic diagram of the ischemic preconditioning training instrument provided for the embodiment 4 of the application.

[0053] The drawings show that: 1, cuff; 2, control box. DETAILED DESCRIPTION

[0054] In order to make the technical means, creative features, purposes and effects realized by the application easy to understand, the application will be further described below in combination with specific drawings. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0055] Embodiment 1

[0056] Please refer to Figure 1The embodiment of the present application provides a technical scheme: a limb ischemic preconditioning training method, comprising:

[0057] Collecting characteristic parameters of individual AOP of a target user, a training purpose and a cuff characteristic parameter, wherein the characteristic parameters of individual AOP include resting systolic blood pressure, limb circumference and subcutaneous fat thickness, wherein the limb circumference is the limb circumference of a part of the body for ischemic pressure training;

[0058] It should be noted that the individual AOP is the arterial occlusion pressure of the individual, the individual AOP is positively correlated with the resting systolic blood pressure of the individual, and a person with a higher resting systolic blood pressure needs a higher cuff pressure to completely block the arterial blood flow; the limb cross-sectional circumference is one of the AOP prediction factors, and the larger the circumference, the higher the blocking pressure required (for example, a person with a thicker thigh or upper arm); a person with a thicker subcutaneous fat layer also needs a higher cuff pressure to completely block the arterial blood flow due to pressure dispersion.

[0059] The training purpose includes blood flow restriction training and heart-brain preconditioning training;

[0060] It should be noted that the blood flow restriction training and the heart-brain preconditioning training require different pressure application pressures, and generally, 40%-80% AOP is selected for blood flow restriction training, and 100% AOP is selected for short-term complete occlusion for heart-brain preconditioning training;

[0061] The cuff characteristic parameter is a cuff width;

[0062] It should be noted that a narrow cuff (5 cm) requires a higher pressure than a wide cuff (10-12 cm) to achieve the same AOP; a wide cuff reduces the required pressure due to a large contact surface;

[0063] The acquired characteristic parameters of individual AOP of the target user, the training purpose and the cuff characteristic parameter are input into a pressure prediction model, and a pressure application pressure value is output;

[0064] The training method of the pressure prediction model comprises:

[0065] A pressure prediction data set is pre-constructed, and the collected pressure prediction data set is pre-processed, including data cleaning and data standardization processing, the pressure prediction data set includes p groups of characteristic parameters of individual AOP of target users, training purposes and cuff characteristic parameters and p groups of characteristic parameters of individual AOP of target users, training purposes and cuff characteristic parameters corresponding to the pressure application pressure value, p is a positive integer greater than 0, and the pressure prediction data set is divided into a pressure prediction data training set and a pressure prediction data verification set, wherein the pressure prediction data training set is used for parameter learning of the pressure prediction model, and the pressure prediction data verification set is used for real-time evaluation of the generalization ability of the pressure prediction model;

[0066] In the pressure prediction model training process, a deep neural network structure based on a multilayer perceptron is adopted, the feature parameters of the individual AOP of the target user, the training purpose and the cuff feature parameters are converted into high-dimensional feature vectors as inputs, the nonlinear features in the data are extracted through several hidden layers, and finally the probability distribution of the pressure prediction result is generated in the output layer using the softmax activation function, and the pressure value prediction result corresponding to the maximum probability is output as the final prediction result; the training process aims to minimize the cross-entropy loss function, and an early stopping strategy is introduced to monitor the performance of the pressure prediction data validation set. When the prediction accuracy on the pressure prediction data validation set reaches the preset threshold, it is considered that the pressure prediction model has converged, and the training of the pressure prediction model is stopped.

[0067] It should be noted that the data cleaning includes checking whether there are missing values, abnormal values, etc. in the collected data. For missing values, mean filling, median filling or model-based filling method can be used for processing; for abnormal values, 3σ principle or machine learning-based anomaly detection method can be used for identification and processing; data standardization processing includes standardization processing and encoding processing of the input features of the collected feature parameters of the individual AOP of the target user, the training purpose and the cuff feature parameters, such as using Z-score standardization to convert the data into a distribution with a mean of 0 and a standard deviation of 1.

[0068] The blood vessel health coefficient of the target user is collected, and the obtained blood vessel health coefficient, limb circumference and pressure value, and training purpose are input into the pressure blocking time prediction model to output the pressure blocking time.

[0069] The training method of the pressure blocking time prediction model comprises:

[0070] A pre-constructed time prediction data set is preprocessed, including data cleaning and data standardization processing. The time prediction data set includes M groups of blood vessel health coefficients, limb circumferences and pressure values of target users, and the pressure blocking time corresponding to the blood vessel health coefficients, limb circumferences and pressure values of the M groups of target users. M is a positive integer greater than 0. The time prediction data set is divided into a time prediction data training set and a time prediction data validation set, wherein the time prediction data training set is used for parameter learning of the pressure blocking time prediction model, and the time prediction data validation set is used for real-time evaluation of the generalization ability of the pressure blocking time prediction model.

[0071] In the training process of the pressure blocking time prediction model, a random forest regression model is used for training. In the training process, the model will construct multiple decision trees according to the input feature data and the corresponding pressure blocking time label. Each decision tree will randomly select part of the features and samples for training when constructing. By calculating the information gain (or Gini impurity) index, the best split feature and split point are selected, and the samples are gradually divided into different sub-nodes until the preset stopping condition (such as reaching the maximum depth or the number of sub-node samples is less than the minimum sample number) is met. The grid search or random search method is used to adjust the hyperparameters (such as the number of decision trees, the maximum depth of the tree, and the minimum sample number) of the random forest regression model to find the optimal combination of hyperparameters and improve the performance of the model.

[0072] The trained model is evaluated using the time prediction data validation set. When the prediction accuracy on the time prediction data validation set reaches the preset threshold, it is considered that the pressure blocking time prediction model has converged, and the training of the pressure blocking time prediction model is stopped.

[0073] It should be noted that there is a complex nonlinear relationship between the blood vessel health coefficient, the limb circumference, the pressure value, the training purpose, and the pressure blocking time. The random forest regression model can well capture these complex relationships, and by integrating the results of multiple decision trees, the stability and generalization ability of the model are improved, which is suitable for such multi-feature regression prediction problems.

[0074] The method for obtaining the blood vessel health coefficient of the target user comprises:

[0075] The carotid-femoral pulse wave velocity is measured by an arterial stiffness instrument, and the augmentation index is automatically calculated by a pulse wave analyzer;

[0076] The heart rate variability is obtained by calculating the standard deviation after 5 minutes of electrocardiogram monitoring;

[0077] The mean and standard deviation of the pulse wave velocity, augmentation index, and heart rate variability are calculated, and the original data is converted into Z-score standardized values;

[0078] It should be noted that by standardization, the dimensional difference is eliminated, different parameters are mapped to the same order of magnitude, and the rationality and physical meaning of the weighted calculation are ensured;

[0079] The pulse wave velocity, augmentation index, and heart rate variability are calculated by a formula to generate the blood vessel health coefficient of the target user;

[0080] Preferably, the technical formula of the blood vessel health coefficient is:

[0081] In the formula, VHI is a blood vessel health coefficient, PWV is a pulse wave velocity, Alx is an augmentation index, HRV is a heart rate variability value, β1, β2 and β3 are weight factors, β1, β2 and β3 are all greater than 0, preferably, β1 = 0.4, β2 = 0.35, and β3 = 0.25;

[0082] It should be noted that the size of the weight coefficient is a specific numerical value obtained by quantizing each data, which is convenient for subsequent comparison. The size of the weight coefficient depends on the number of comprehensive parameters and the corresponding weight coefficient preliminarily set by the person skilled in the art for each group of comprehensive parameters.

[0083] It should be noted that the pulse wave velocity (PWV) is an index for evaluating arterial stiffness, and the higher the value, the stiffer the artery, the poorer the elasticity, and the poorer the blood vessel health status. The pulse wave velocity (PWV) is negatively correlated with the blood vessel health status. The augmentation index (AIx) is an index for evaluating arterial stiffness, and the higher the value, the stiffer the artery, the stronger the reflected wave, the heavier the burden on the heart, and the poorer the blood vessel health status. AIx is negatively correlated with the blood vessel health status. Heart rate variability (HRV) is an index for evaluating autonomic nervous function, and the higher the value, the stronger the autonomic nervous regulation ability, the better the adaptability of the cardiovascular system, and the better the blood vessel health status. Heart rate variability (HRV) is positively correlated with the blood vessel health status. In summary, the larger the value of the blood vessel health coefficient (VHI), the better the blood vessel health status.

[0084] The target user's limb is subjected to a target action of a preset number of cycles by the ischemic preconditioning training instrument, and the target action is ischemic preconditioning training by the ischemic preconditioning training instrument based on the output pressure blocking time and pressure value.

[0085] In this embodiment, by integrating individual AOP characteristic parameters (such as resting systolic blood pressure, limb circumference, cuff width, and subcutaneous fat thickness) and training purposes (blood flow restriction training or heart and brain preconditioning training), a pressure prediction model based on a deep neural network is constructed, which significantly improves the accuracy and individualization of the pressure. Combined with a multilayer perceptron to extract nonlinear features, the optimal pressure value under different training scenarios can be dynamically predicted, such as 40%-80% AOP for blood flow restriction training and 100% AOP short occlusion for heart and brain preconditioning training, thereby meeting the physiological needs of different training goals. Secondly, the blood vessel health coefficient and the pressure blocking time prediction model are introduced, which further optimizes the training safety and avoids tissue damage caused by excessive pressure or long time. Therefore, this limb ischemic preconditioning training method not only improves the controllability of the training effect, but also reduces the risk brought by individual differences of users, and has significant clinical practical value and technical innovation.

[0086] Embodiment 2

[0087] Please refer toFigure 2 The technical problem to be solved by the embodiment is how to develop a reasonable training strategy based on the pressure blocking time and pressure of the ischemic preconditioning training obtained based on the above content of the embodiment 1, and ensure reasonable pressure adjustment.

[0088] The limb ischemic preconditioning training method further comprises:

[0089] physiological signal parameters of the target user before and during the ischemic preconditioning training are acquired in real time;

[0090] A prediction training risk score model is constructed according to the reaction changes of the physiological signal parameters during the training process, and a training risk score is obtained.

[0091] The expression of the prediction training risk score model is:

[0092]

[0093] In the formula, Rf is the training risk score, ΔHRV is the HRV after training minus the HRV before training, HRV0 is the HRV value before training, HRp is the heart rate peak value during the training process, HRs is the upper limit of the individual safe heart rate (which can be set according to age and physical condition), is the minimum value of blood oxygen, γ1, γ2 and γ3 are weight coefficients, and γ1, γ2 and γ3 are all greater than 0.

[0094] It should be noted that, is a neural load risk term. The greater this term, the more the heart rate variability decreases, and the more obvious the neural fatigue is. It represents the load risk of the sympathetic / parasympathetic nervous system during training.

[0095] is a heart rate overload risk term. The greater this term, the closer the peak heart rate during training is to the dangerous edge, which represents the load risk of the cardiovascular system.

[0096] is a blood oxygen risk term. This term is used to reflect whether the oxygen transport system is in danger of hypoxia due to blocking. If is significantly less than 90%, this term significantly increases, indicating that the risk of tissue hypoxia is rising.

[0097] In the embodiment, by collecting the physiological signal parameters of the target user before and during the ischemic preconditioning training, a comprehensive training risk score Rf is constructed to represent the physiological load degree of the current training scheme on the individual. The neural load risk term, the heart rate overload risk term and the blood oxygen risk term are comprehensively analyzed from three aspects, so as to guide the automatic adjustment of the subsequent pressure strategy, and realize the dual protection of safety and adaptability.

[0098] Based on the obtained training risk score and blood vessel health coefficient, the next round of training pressure value and pressure blocking time are dynamically adjusted, and an optimized pressure value and pressure blocking time are generated.

[0099] Based on the optimized pressure value and pressure blocking time, the target user's limb is executed by the ischemic preconditioning training instrument for a preset number of cycles of target actions.

[0100] The method for adjusting the next round of pressure value based on the training risk score and the blood vessel health coefficient comprises:

[0101] Based on the current round of pressure value, the negative inhibition of the training risk score, and the positive enhancement of the blood vessel health coefficient, the pressure value is dynamically adjusted, wherein the positive enhancement of the blood vessel health coefficient gradually enhances the pressure value through an exponential decay characteristic, and individuals with better blood vessel health coefficients can withstand higher training intensity, and the negative inhibition of the training risk score automatically reduces the pressure through a saturation characteristic of a hyperbolic function to avoid excessive physiological load due to risk accumulation, thereby obtaining an optimized pressure value.

[0102] A specific calculation example is as follows:

[0103]

[0104] In the formula, Pn+1 is the next round of pressure value, Pn is the current round of pressure value, k1 is an exponential function adjustment parameter, VHI is the blood vessel health coefficient, Rf is the training risk score, k2 is a risk adjustment coefficient, and the influence of Rf on pressure adjustment is controlled, is an exponential decay term, and as VHI increases, the exponential term decreases and the pressure increases; tanh(k2xRf) is a hyperbolic tangent function, representing the pressure reduction trend caused by the increase of risk, which tends to converge to avoid overpressure.

[0105] Specifically, the formula takes into account the blood vessel health coefficient and the current training risk of the individual, and adjusts the individualized pressure strategy based on the current round of pressure value. By combining the exponential function and the hyperbolic function, the training intensity of individuals with poor blood vessel response can be gradually increased, and the pressure can be automatically reduced when the risk score is too high to prevent physiological overload and avoid discomfort or danger caused by excessive adjustment.

[0106] The larger the VHI, the smaller the exponential decay term and the more significant the pressure increase, encouraging individuals with better blood vessel health to gradually increase their training intensity; when Rf increases, the hyperbolic tangent function approaches a saturation value, and the pressure reduction increases, thereby automatically reducing the pressure when the risk is too high to avoid physiological overload;

[0107] The formula combines exponential function and hyperbolic function, which not only ensures that the training intensity of the person with poor blood vessel state is gradually enhanced, but also quickly converges the adjustment range when the risk increases rapidly, and finally achieves a dynamic balance between improving the training effect and ensuring physiological safety.

[0108] The method for adjusting the next round of pressure blocking time based on the training risk score and the blood vessel health coefficient comprises:

[0109] Based on the predicted pressure blocking time, the optimized pressure blocking time is generated through the positive regulation of the blood vessel health coefficient and the negative inhibition of the training risk score (Rf).

[0110] A specific calculation example is as follows:

[0111]

[0112] In the formula, Tn+1 is the next round of pressure blocking time, Tyc is the predicted pressure blocking time, VHI is the blood vessel health coefficient, VHImax is the maximum reference value of the blood vessel health coefficient, that is, the upper limit value of the normal healthy level, Rf is the risk score, Rfmax is the preset risk score threshold, and the ischemic training needs to be stopped when the value is exceeded, θ1 and θ2 are weight coefficients, and θ1 and θ2 are greater than 0.

[0113] θ1 reflects the sensitivity of the blood vessel health to the training time, and θ2 represents the conservative degree of the system to the risk.

[0114] It should be noted that the size of the weight coefficient is a specific value obtained by quantizing each data, which is convenient for subsequent comparison. The size of the weight coefficient depends on how many comprehensive parameters and the corresponding weight coefficients preliminarily set by the person skilled in the art for each group of comprehensive parameters.

[0115] Specifically, in the training pre-stage, the blood vessel reactivity index is established by collecting individual resting physiological parameters, and a baseline is provided for subsequent training; in the training process, the heart rate, blood oxygen saturation and blood flow waveform and other key indicators are monitored in real time to ensure that the training state is in a safe range; the heart rate variability is collected to comprehensively evaluate the training effect and the body load level. Based on the collected data throughout the process, the blood vessel health coefficient and the risk score index are calculated, the next round of pressure and blocking time is dynamically generated, and a closed-loop optimization mechanism is constructed, so as to realize the individual adjustment of the training intensity, improve the training efficiency, effectively avoid the overload risk, and ensure the physiological safety of the user.

[0116] Embodiment 3

[0117] Please refer to Figure 3 The embodiment of the present application provides a technical scheme: a limb ischemic preconditioning training device applied to a preset ischemic preconditioning training instrument, comprising:

[0118] a data collection module, configured to collect characteristic parameters of individual AOP of a target user, training purposes and cuff characteristic parameters, and collect a blood vessel health coefficient of the target user;

[0119] a pressure prediction module, configured to input the collected characteristic parameters of individual AOP of the target user, training purposes and cuff characteristic parameters into a pressure prediction model, and output a pressure value;

[0120] a time prediction module, configured to input the collected blood vessel health coefficient, limb circumference and pressure value, and training purposes into a pressure blocking time prediction model, and output a pressure blocking time;

[0121] a pressure training module, configured to execute a target action of a preset cycle number on a limb of the target user by the ischemic preconditioning training instrument, the target action being ischemic preconditioning training performed by the ischemic preconditioning training instrument based on the output pressure blocking time and pressure value;

[0122] a dynamic parameter adjustment module, configured to acquire physiological signal parameters of the target user before and during ischemic preconditioning training in real time, construct a prediction training risk score model according to changes in the physiological signal parameters during training, acquire a training risk score, dynamically adjust a next round of training pressure value and pressure blocking time based on the acquired training risk score and blood vessel health coefficient, generate an optimized pressure value and pressure blocking time, and execute a target action of a preset cycle number on a limb of the target user by the ischemic preconditioning training instrument based on the optimized pressure value and pressure blocking time

[0123] Embodiment 4

[0124] Please refer to Figure 4 The embodiment of the present application provides a technical solution: an ischemic preconditioning training instrument, comprising a cuff 1 capable of being inflated and pressurized and a control box 2, the control box 2 being capable of being arranged on the outer wall of the cuff 1 or being arranged separately from the cuff 1.

[0125] The control box 2 is provided with an inflation pump, the inflation pump being communicated with a gas guide pipe, the end of the gas guide pipe being communicated with the cuff 1, and the inflation pump inflating the cuff through the gas guide pipe to achieve the purpose of performing pressure training on the limbs of the patient (the inflation pump and the gas guide pipe are not shown in the figure).

[0126] The basic principles, main features and advantages of the present application are shown and described above. It should be understood by those skilled in the art that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for limb ischemic preconditioning training, applied to a preset ischemic preconditioning training instrument, characterized in that, The method comprises: collecting individual AOP characteristic parameters, training purposes and cuff characteristic parameters of a target user; inputting the collected individual AOP characteristic parameters, training purposes and cuff characteristic parameters of the target user into a pressure prediction model to output a pressure value; collecting a blood vessel health coefficient of the target user; inputting the collected blood vessel health coefficient, limb circumference and pressure value, and training purpose into a pressure blocking time prediction model to output a pressure blocking time; performing a target action on the limb of the target user through the ischemic preconditioning training instrument for a preset number of cycles, the target action being ischemic preconditioning training based on the output pressure blocking time and pressure value through the ischemic preconditioning training instrument.

2. The method of claim 1, wherein the ischemic preconditioning training is performed for a period of 2 to 4 weeks. The limb ischemic preconditioning training method further comprises: real-time acquisition of physiological signal parameters of the target user before and during ischemic preconditioning training; construction of a training risk score prediction model according to changes in physiological signal parameters during training to obtain a training risk score; dynamic adjustment of the next round of training pressure value and pressure blocking time based on the obtained training risk score and blood vessel health coefficient to generate an optimized pressure value and pressure blocking time; performance of a target action on the limb of the target user through the ischemic preconditioning training instrument for a preset number of cycles based on the optimized pressure value and pressure blocking time.

3. The method of claim 1, wherein the ischemic preconditioning is performed by, The individual AOP characteristic parameters include resting systolic blood pressure, limb circumference and subcutaneous fat thickness, wherein the limb circumference is the circumference of the limb for ischemic pressure training; The training purposes include blood flow restriction training and heart-brain preconditioning training; The cuff characteristic parameters are cuff width.

4. The method of claim 3, wherein the ischemic preconditioning is performed by, The training method of the pressure prediction model comprises: pre-constructing a pressure prediction dataset, pre-processing the collected pressure prediction dataset, including data cleaning and data standardization processing, the pressure prediction dataset comprising p sets of individual AOP characteristic parameters, training purposes and cuff characteristic parameters of target users and p sets of individual AOP characteristic parameters, training purposes and cuff characteristic parameters of target users corresponding to pressure values, p being a positive integer greater than 0, dividing the pressure prediction dataset into a pressure prediction data training set and a pressure prediction data validation set, wherein the pressure prediction data training set is used for parameter learning of the pressure prediction model, and the pressure prediction data validation set is used for real-time evaluation of the generalization ability of the pressure prediction model; In the pressure prediction model training process, a deep neural network structure based on a multilayer perceptron is adopted, the individual AOP characteristic parameters, training purposes and cuff characteristic parameters of the target user are converted into high-dimensional feature vectors as input, non-linear features in the data are extracted through several hidden layers, and finally the probability distribution of the pressure prediction result is generated in the output layer using the softmax activation function, and the pressure prediction result corresponding to the maximum probability is output as the final prediction result; the training process aims to minimize the cross-entropy loss function, and an early stopping strategy is introduced to monitor the performance of the pressure prediction data validation set; when the prediction accuracy on the pressure prediction data validation set reaches a preset threshold, it is considered that the pressure prediction model has converged, and the training of the pressure prediction model is stopped.

5. The method of claim 1, wherein the ischemic preconditioning training is performed for a period of 2 to 4 weeks. The method for obtaining the blood vessel health coefficient of the target user comprises: The carotid-femoral pulse wave velocity is measured by an arterial stiffness instrument, and the augmentation index is automatically calculated by a pulse wave analyzer; The standard deviation is calculated after 5 minutes of electrocardiogram monitoring to obtain heart rate variability; The mean and standard deviation of the pulse wave velocity, augmentation index and heart rate variability are calculated to convert the original data into Z-score standardized values; The pulse wave velocity, augmentation index and heart rate variability are calculated by a formula to generate the blood vessel health coefficient of the target user.

6. The method of claim 1, wherein the ischemic preconditioning training is performed for a period of 2 to 4 weeks. The training method of the pressure blocking time prediction model comprises: A pre-constructed time prediction data set is pre-processed, including data cleaning and data standardization, the time prediction data set comprises M groups of blood vessel health coefficients, limb circumferences and pressure values of the target users, and the blood vessel health coefficients, limb circumferences and pressure values of the M groups of target users correspond to the pressure blocking time, M is a positive integer greater than 0, the time prediction data set is divided into a time prediction data training set and a time prediction data validation set, wherein the time prediction data training set is used for parameter learning of the pressure blocking time prediction model, and the time prediction data validation set is used for real-time evaluation of the generalization ability of the pressure blocking time prediction model; In the training process of the pressure blocking time prediction model, a random forest regression model is used for training, in the training process, the model will construct multiple decision trees according to the input feature data and the corresponding pressure blocking time label, each decision tree will randomly select part of the features and samples for training when constructing, the best split feature and split point are selected by calculating the information gain index, the samples are gradually divided into different sub-nodes, until the preset stop condition is met, the hyperparameters of the random forest regression model are adjusted using grid search or random search, to find the optimal combination of hyperparameters and improve the performance of the model; The trained model is evaluated using the time prediction data validation set, when the prediction accuracy on the time prediction data validation set reaches a preset threshold, it is considered that the pressure blocking time prediction model has converged, and the training of the pressure blocking time prediction model is stopped.

7. The method of claim 2, wherein the ischemic preconditioning is performed by, The method for adjusting the next round of pressure blocking time based on the training risk score and the blood vessel health coefficient comprises: The current round of pressure is dynamically adjusted based on the pressure value combined with the negative inhibition of the training risk score and the positive enhancement of the blood vessel health coefficient, wherein the positive enhancement of the blood vessel health coefficient is to gradually enhance the pressure value through an exponential decay characteristic, and the individual with a better blood vessel health coefficient can withstand higher training intensity, wherein the negative inhibition of the training risk score is to automatically reduce the pressure through the saturation characteristic of a hyperbolic function to avoid excessive physiological load due to risk accumulation, thereby obtaining an optimized pressure blocking time.

8. The method of claim 1, wherein the ischemic preconditioning training is performed for a period of 2 to 4 weeks. The method for adjusting the next round of pressure blocking time based on the training risk score and the blood vessel health coefficient comprises: The predicted pressure blocking time is taken as a reference to dynamically adjust the optimized pressure blocking time through the positive regulation of the blood vessel health coefficient and the negative inhibition of the training risk score.

9. A limb ischemic preconditioning training device applied to a preset ischemic preconditioning training instrument, characterized in that, The method comprises: The data acquisition module acquires the characteristic parameters of the individual AOP of the target user, the training purpose and the cuff characteristic parameters, and acquires the blood vessel health coefficient of the target user; The pressure prediction module inputs the acquired characteristic parameters of the individual AOP of the target user, the training purpose and the cuff characteristic parameters into the pressure prediction model, and outputs the pressure value; The time prediction module inputs the acquired blood vessel health coefficient, limb circumference and pressure value, and training purpose into the pressure blocking time prediction model, and outputs the pressure blocking time; The pressure training module executes the target action of a preset cycle number on the limb of the target user through the ischemic preconditioning training instrument, and the target action is ischemic adaptation training through the ischemic preconditioning training instrument based on the output pressure blocking time and pressure value; The dynamic parameter adjustment module acquires the physiological signal parameters before and during the ischemic adaptation training of the target user in real time, constructs a prediction training risk score model according to the reaction change of the physiological signal parameters during the training process, acquires the training risk score, dynamically adjusts the next round of training pressure value and pressure blocking time based on the acquired training risk score and blood vessel health coefficient, generates the optimized pressure value and pressure blocking time, and executes the target action of a preset cycle number on the limb of the target user through the ischemic preconditioning training instrument based on the optimized pressure value and pressure blocking time.

10. Ischemic preconditioning training apparatus, characterized in that, The control box (2) can be arranged on the outer wall of the cuff (1) or arranged separately from the cuff (1); The control box (2) is provided with an inflation pump, the inflation pump is communicated with an air guide pipe, the end of the air guide pipe is communicated with the cuff (1), and the inflation pump inflates the cuff through the air guide pipe to achieve the purpose of pressure training on the limbs of the patient.

Images