Physical stimulation load intensity detection and evaluation method and system based on dynamic modeling

Abstract

The invention discloses a physical stimulation load intensity detection and evaluation method and system based on dynamic modeling, and relates to the technical field of biomedicine. In order to solve the problems that data accuracy is difficult to guarantee and an effective data synchronization and error monitoring mechanism is lacked due to noise interference in a data acquisition process; the evaluation model is not accurate enough and cannot comprehensively and accurately reflect the relationship between the biological signal and the physical stimulation load intensity, so that the evaluation result has a large error; according to the method, the biological signals are processed and subjected to feature extraction, the linear regression equation is constructed by combining the time domain and frequency domain features of the biological signals, and the load intensity parameters are extracted for multi-index evaluation, so that the relationship between the biological signals and the physical stimulation load intensity can be accurately reflected, the actual physical stimulation load intensity value is corrected, and the evaluation accuracy is improved. It is ensured that the evaluation result can truly reflect the influence of physical stimulation on an experimenter, and reliable technical support is provided for related research and application.

Description

Technical Field

[0001] The present invention relates to the field of biomedical technology, and particularly to a method and system for detecting and evaluating the intensity of physical stimulation load based on dynamic modeling. Background Art

[0002] In many fields such as biomedicine and sports science, accurately detecting the intensity of physical stimulation load is crucial for understanding the physiological responses and health conditions of experimental subjects or patients. For example, in rehabilitation therapy, it is necessary to adjust the treatment intensity according to the patient's response to physical stimulation (such as electrical stimulation, heat stimulation, etc.); in sports training, it is also necessary to accurately grasp the load of athletes under different physical stimulations in order to formulate a reasonable training plan.

[0003] However, the existing methods for detecting the intensity of physical stimulation load have many deficiencies. During the data acquisition process, it is easily affected by noise interference, resulting in difficulty in ensuring data accuracy, and lacking effective data synchronization and error monitoring mechanisms; moreover, the evaluation model is not precise enough to comprehensively and accurately reflect the relationship between biological signals and the intensity of physical stimulation load, making the evaluation results have large errors. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system for detecting and evaluating the intensity of physical stimulation load based on dynamic modeling. By collecting and processing biological signals, constructing an evaluation model to detect the intensity of physical stimulation load, and performing multi-index evaluation, it can accurately reflect the relationship between biological signals and the intensity of physical stimulation load, providing a precise evaluation means for the detection of the intensity of physical stimulation load, so as to solve the problems raised in the above background art.

[0005] To achieve the above purpose, the present invention provides the following technical solutions:

[0006] A method for detecting and evaluating the intensity of physical stimulation load based on dynamic modeling, including:

[0007] Real-time collect the biological signals of the experimental subject during the physical stimulation process, obtain the original data reflecting the physiological state of the body, and perform preprocessing and feature extraction on the biological signals to determine the change law of the biological signals in different time periods, and generate a change feature vector of the biological signals;

[0008] Establish a mapping relationship between the change feature vector of the biological signal characteristics and the load intensity, construct a physical stimulation load intensity evaluation model, and evaluate the real-time collected biological signals based on the physical stimulation load intensity evaluation model to obtain the evaluation result of the physical stimulation load intensity.

[0009] Furthermore, collecting the biological signals of an individual during the physical stimulation process further includes:

[0010] Install the corresponding biosignal sensor at the signal acquisition site of the experimenter, and during the physical stimulation process, monitor and record the actual biosignals of each biosignal sensor in real time;

[0011] And perform the first-time marking on the actual biosignals according to the acquisition time, and perform the second-time marking on the preset biosignal parameters. Among them, the preset biosignal parameters include the expected heart rate range, electromyogram activity level, and skin conductance change;

[0012] Align the first-time marking and the second-time marking according to the time characteristics to obtain the alignment result;

[0013] Based on the alignment result, compare the difference between the actual biosignal and the preset biosignal parameters at each time point, and perform statistical evaluation to obtain the alignment verification value of the alignment result;

[0014] Count the number of time points in the alignment result whose alignment verification value is greater than the preset alignment verification value. When the statistical result is greater than the preset number, determine that the alignment result is the target alignment result; otherwise, perform the alignment operation again until the alignment verification condition is met;

[0015] Based on the target alignment result, synchronize the actual biosignal and the preset biosignal parameters, calculate the data error value between the actual data and the preset data, and determine the data error range.

[0016] Furthermore, determining the data error range also includes:

[0017] Based on the biosignal parameter type of the actual data corresponding to the data error value, determine the preset threshold corresponding to this type;

[0018] When the data error value is greater than or equal to the corresponding preset threshold, pause the physical stimulation load intensity detection process, determine the alarm level according to the size of the data error range, and issue the corresponding alarm according to the alarm level.

[0019] Furthermore, preprocess and extract features from the biosignals, specifically including:

[0020] Perform filtering and denoising processing on the biosignals collected in real time, and perform clustering processing on the biosignals based on the biosignal type to obtain each biosignal data group under the load intensity;

[0021] Determine the target value of each biosignal data group, and draw a biosignal change curve based on the target value;

[0022] Furthermore, determine the total change value from the start to the end of the physical stimulation according to the biosignal change curve. At the same time, divide the biosignal change curve into several curve segments according to the time from the start to the end of the physical stimulation;

[0023] Perform time-domain feature analysis on each segmented curve segment, calculate the time-domain feature parameters of each curve segment, perform Fourier transform on each curve segment, and determine the frequency-domain feature parameters of each biological signal change curve segment;

[0024] Based on the time-domain feature parameters and frequency-domain feature parameters, determine the variation law of the biological signal in different time periods, and generate a change feature vector of the biological signal.

[0025] Based on the time-domain feature parameters and frequency-domain feature parameters, determine the variation law of the biological signal in different time periods, and generate a change feature vector of the biological signal, including:

[0026] Obtain the feature parameter types and the characteristic of the feature parameter values of the time-domain feature parameters and frequency-domain feature parameters, standardize the time-domain feature parameters and frequency-domain feature parameters to obtain standard feature parameters;

[0027] Based on the correlation characteristics between the standard feature parameters, establish a tree model of the standard feature parameters, obtain the node layer number and the node splitting number of each standard feature parameter in the tree model, perform importance evaluation on each standard feature parameter based on the node layer number and the node splitting number to obtain an important evaluation score, and select the standard feature parameters with the important evaluation score greater than the preset evaluation score as the target standard feature parameters;

[0028] Perform transformation and integration on the change trend of the time series of the target standard feature parameters to obtain a dynamic change value sequence;

[0029] Based on the fluctuation amplitude of each element in the dynamic change value sequence and the total correlation degree between each element and other elements, extract the main elements of the dynamic change value sequence to obtain a target dynamic change value sequence;

[0030] Based on the historical feature vectors and the accuracy results of the corresponding physical stimulus load intensity evaluation, perform model training to obtain a feature vector evaluation model, and input the target dynamic change value sequence into the vector evaluation model to obtain a prediction score;

[0031] When the prediction score is greater than the preset score, use the target dynamic change value sequence as the change feature vector of the biological signal;

[0032] Otherwise, change the strategy again to extract the main elements of the dynamic change value sequence to obtain the latest dynamic change value sequence until the prediction score is greater than the preset score, and use the latest dynamic change value sequence as the change feature vector of the biological signal.

[0033] Furthermore, construct a physical stimulus load intensity evaluation model, including:

[0034] Collect the eigenvectors of the changes in the biological signals of at least one subject under different physical stimulus intensities. Meanwhile, obtain the corresponding actual physical stimulus load intensity values;

[0035] Calculate the time-series characteristic curve graph according to the quantization calculation index and the eigenvectors of the changes in the biological signals of each subject, and construct the linear regression equation between the eigenvectors of the changes in the biological signals of each subject and the physical stimulus time series;

[0036] Among them, the quantization calculation index includes various statistical characteristics of the biological signals in the time domain and frequency domain;

[0037] Solve the linear regression equation between the eigenvectors of the changes in the biological signals of each subject and the physical stimulus time series, and determine the change characteristics of the eigenvectors of the changes in the biological signals of each subject according to the solution results.

[0038] Furthermore, constructing the physical stimulus load intensity evaluation model further includes:

[0039] Extract the load intensity parameters corresponding to multiple physical stimulus load intensity evaluation indexes according to the change characteristics of the eigenvectors of the changes in the biological signals of each subject;

[0040] Conduct a pass evaluation on the load intensity parameters corresponding to each physical stimulus load intensity evaluation index, and determine the evaluation score of the load intensity parameters corresponding to this evaluation index according to the evaluation results and the comprehensive evaluation weight of this physical stimulus load intensity evaluation index;

[0041] Add up the evaluation scores of multiple physical stimulus load intensity evaluation indexes of each subject to obtain the calculated performance evaluation total score of the physical stimulus load intensity corresponding to this subject;

[0042] Based on the calculated performance evaluation total score of the physical stimulus load intensity of each subject, determine whether the physical stimulus load intensity evaluation model corresponding to this subject meets the effective model screening conditions;

[0043] If the conditions are met, it is considered that the physical stimulus load intensity evaluation model can accurately reflect the relationship between the biological signal and the physical stimulus load intensity and can be used for subsequent evaluations; otherwise, a prompt to adjust the experimental conditions or reconstruct the model is issued.

[0044] Furthermore, obtaining the corresponding actual physical stimulus load intensity values further includes:

[0045] Based on the operating parameters of the physical stimulus device of the subject during the physical stimulus process, determine the parameter values under the physical stimulus operating stability, biological signal response efficiency, and physical stimulus safety guarantee function indexes;

[0046] Determine whether the operation stability of the physical stimulus is greater than a preset stability threshold. If so, no stability correction is performed. Otherwise, based on the stability difference between the operation stability and the preset stability, a first correction is made to the actual physical stimulus load intensity value;

[0047] After determining that the stability of the physical stimulus meets the requirements, determine whether the parameter value under the physical stimulus safety guarantee index is within the preset physical stimulus safety guarantee index value range. When the parameter value exceeds the preset range, based on the difference between the parameter value and the preset physical stimulus safety guarantee index value range, a second correction is made to the actual physical stimulus load intensity value;

[0048] After determining that the safety guarantee index of the physical stimulus meets the requirements, determine whether the biological signal response efficacy is greater than the preset biological signal response efficacy. If the biological signal response efficacy is less than or equal to the preset value, based on the efficacy difference between the biological signal response efficacy and the preset biological signal response efficacy, a third correction is made to the actual physical stimulus load intensity value;

[0049] Determine whether the difference between the load intensity value after the third correction and the actual physical stimulus load intensity value before the correction is within the preset correction range, and output the load intensity value after the third correction with the difference within the preset correction range. Otherwise, adjust the load intensity value after the third correction until the difference between the adjusted load intensity value and the actual physical stimulus load intensity value before the correction is within the preset correction range.

[0050] Furthermore, a feature vector evaluation model is obtained by training a model based on historical feature vectors and their corresponding accuracy results of physical stimulus load intensity evaluation, including:

[0051] Process the same historical feature vector based on different vector processing methods to obtain corresponding multiple prediction evaluation results. Based on the difference between the physical stimulus load intensity evaluation and the prediction evaluation results, select the target feature processing vector;

[0052] Based on the correlation between elements in the target feature processing vector and the vector similarity between target feature processing vectors, calculate the scoring value of the target feature processing vector;

[0053] Select the preferred feature processing vectors with scoring values greater than the preset scoring value from the target feature processing vectors;

[0054] Based on the preferred feature processing vectors, calculate the importance degree of each type of vector element;

[0055] Based on the importance degree, perform weight marking on the preferred feature processing vectors, and perform model training based on the weight-marked preferred feature processing vectors to obtain the feature vector evaluation model.

[0056] The present invention provides another technical solution, a physical stimulation load intensity detection and evaluation system based on dynamic modeling, including:

[0057] A biological signal acquisition module, configured to collect the biological signals of the experimenter during the physical stimulation in real time, extract time-domain and frequency-domain characteristic parameters from the preprocessed biological signals, and generate a change characteristic vector of the biological signals;

[0058] A load intensity correction module, configured to determine the operation stability of the physical stimulation, the biological signal response efficiency, and the physical stimulation safety guarantee index parameters according to the operation parameters of the physical stimulation device, and correct the actual physical stimulation load intensity value;

[0059] A model construction and evaluation module, configured to construct a linear regression equation between the change characteristic vector of the biological signals and the physical stimulation time series based on the change characteristic vector of the biological signals and the corrected actual physical stimulation load intensity value, determine the physical stimulation load intensity evaluation model, and evaluate the biological signals obtained in real time.

[0060] Compared with the prior art, the beneficial effects of the present invention are:

[0061] By performing time marking, alignment verification, data synchronization, and error margin calculation on the biological signal acquisition process, the accuracy of the collected data can be effectively ensured; at the same time, when the data error value exceeds the preset threshold, the detection is paused in time and an alarm is issued, ensuring the safety of the experimental process and the reliability of the data; determining the change characteristics of the change characteristic vector of the biological signals provides data support for establishing a dynamic relationship model with the load intensity, and then extracting the load intensity parameters for evaluation. By comprehensively evaluating multiple indicators and calculating the total score to screen the effective model, the relationship between the biological signals and the physical stimulation load intensity can be accurately reflected, providing a precise evaluation method for the detection of the physical stimulation load intensity. The actual physical stimulation load intensity value is corrected three times based on the operation parameters of the physical stimulation device, and verified and adjusted, making the load intensity value more in line with the actual situation, and improving the scientificity and rationality of the evaluation results. Description of the Drawings

[0062] Figure 1 It is a flowchart of the physical stimulation load intensity detection and evaluation method based on dynamic modeling of the present invention. Detailed Embodiments

[0063] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0064] To solve the technical problems that during the data acquisition process, it is vulnerable to noise interference, resulting in difficulty in ensuring data accuracy, and there is a lack of effective data synchronization and error monitoring mechanisms; moreover, the evaluation model is not precise enough to comprehensively and accurately reflect the relationship between biological signals and the intensity of physical stimulus load, leading to large errors in the evaluation results, please refer to Figure 1 , the following technical solutions are provided in this embodiment:

[0065] A method for detecting and evaluating the intensity of physical stimulus load based on dynamic modeling, including:

[0066] Real-time collect the biological signals of the experimenter during the physical stimulus process, obtain the raw data reflecting the physiological state of the body, and perform preprocessing and feature extraction on the biological signals to determine the change law of the biological signals in different time periods, and generate a change feature vector of the biological signals;

[0067] Establish a mapping relationship between the change feature vector of the biological signal characteristics and the load intensity, construct a physical stimulus load intensity evaluation model, and evaluate the real-time collected biological signals based on the physical stimulus load intensity evaluation model to obtain the evaluation result of the physical stimulus load intensity.

[0068] In this embodiment, the experimenter refers to the subject participating in the physical stimulus load intensity detection process. This subject can be a human or an experimental animal, with normal physiological functions and response abilities, and can generate detectable physiological parameter changes under physical stimuli for the evaluation of the load intensity.

[0069] In this embodiment, collecting the biological signals of an individual during the physical stimulus process further includes:

[0070] Install the corresponding biological signal sensors at the signal collection sites of the experimenter. For example, the ECG sensor is attached to the chest, the EMG sensor is attached to the muscle surface, and the skin conductance sensor is clipped on the finger or palm, etc. During the physical stimulus process, the actual biological signals of each biological signal sensor are monitored and recorded in real time, including heart rate, heart rate variability, muscle activity, skin conductance change, respiratory rate, etc.;

[0071] And perform the first time marking on the actual biological signals according to the collection time, and perform the second time marking on the preset biological signal parameters, where the preset biological signal parameters include the expected heart rate range, electromyogram activity level, and skin conductance change;

[0072] Align the first time marking and the second time marking according to the time characteristics to obtain the alignment result;

[0073] Based on the alignment result, compare the differences between the actual biological signals and the preset biological signal parameters at each time point, and conduct statistical evaluation to obtain the alignment verification value of the alignment result;

[0074] Count the number of time points in the alignment result where the alignment verification value (e.g., signal matching degree at the time point) is greater than the preset alignment verification value. When the statistical result is greater than the preset number, determine that the alignment result is the target alignment result; otherwise, perform the alignment operation again until the alignment verification condition is met;

[0075] Based on the target alignment result, synchronize the actual biological signals and the preset biological signal parameters, calculate the data error value between the actual data and the preset data, and determine the data error margin (such as heart rate error margin, electromyogram activity error margin, skin conductance error margin, etc.);

[0076] Based on the type of biological signal parameter of the actual data corresponding to the data error value, determine the preset threshold corresponding to this type;

[0077] When the data error value is greater than or equal to the corresponding preset threshold, pause the physical stimulus load intensity detection process, determine the alarm level according to the size of the data error margin (heart rate error margin, electromyogram activity error margin, skin conductance error margin, etc.), and issue the corresponding alarm according to the alarm level to notify the experimental operator or the subject to take corresponding measures.

[0078] In this embodiment, by considering the biological signals at multiple time points, it is possible to comprehensively capture the changes in different time periods, compare the time stamps of the actual biological signals with the time stamps of the expected parameters, find the matching degree between the actual signal and the expected signal. Once the target alignment result is determined, by comparison, calculate the difference between the actual data and the expected data, that is, the data error value. According to the size of the data error value, determine the data error margin, and issue the corresponding alarm according to the determined alarm level to notify the experimental operator or the subject to take necessary measures to ensure the safety of the experiment and the accuracy of the data.

[0079] In this embodiment, preprocess and extract features from the biological signals, specifically including:

[0080] Perform filtering and denoising processing on the real-time collected biological signals to remove high-frequency noise and low-frequency drift, remove random noise and non-linear noise in the signals, and improve the signal quality; and perform clustering processing on the biological signals based on the biological signal type to obtain each biological signal data group under the load intensity;

[0081] Determine the target values of each group of biosignal data, including at least one of the mean value, median value, and peak value, and draw a biosignal change curve based on the target values, such as a heart rate change curve, an electromyogram activity change curve, a skin conductance change curve, etc.;

[0082] Determine the total change value from the start to the end of the physical stimulus according to the biosignal change curve. At the same time, divide the biosignal change curve into several curve segments according to the time from the start to the end of the physical stimulus;

[0083] Perform time-domain feature analysis on each segmented curve segment, calculate the time-domain feature parameters of each curve segment, perform Fourier transform on each curve segment, and determine the frequency-domain feature parameters of each biosignal change curve segment;

[0084] Based on the time-domain feature parameters and frequency-domain feature parameters, determine the change law of the biosignal in different time periods, and generate a change feature vector of the biosignal.

[0085] In this embodiment, by drawing the biosignal change curve, the change trend of the biosignal over time during the physical stimulus is visually displayed, providing a visual basis for further analyzing the physical stimulus load intensity, accurately extracting the biosignal features, and generating a change feature vector. When performing operations such as clustering processing of the biosignal, drawing the change curve, and analyzing the time-domain and frequency-domain feature parameters, these processes are the basic steps of dynamic modeling, improving the accuracy and effectiveness of physical stimulus load intensity detection, and helping to identify physiological response patterns.

[0086] In one embodiment, based on the time-domain feature parameters and frequency-domain feature parameters, determining the change law of the biosignal in different time periods and generating a change feature vector of the biosignal includes:

[0087] Obtain the feature parameter types and feature parameter value characteristics of the time-domain feature parameters and frequency-domain feature parameters, standardize the time-domain feature parameters and frequency-domain feature parameters to obtain standard feature parameters;

[0088] Based on the correlation characteristics between the standard feature parameters, establish a tree model of the standard feature parameters, obtain the node layer number and node split number of each standard feature parameter in the tree model, perform importance evaluation on each standard feature parameter based on the node layer number and node split number to obtain an important evaluation score, and select the standard feature parameters with an important evaluation score greater than the preset evaluation score as the target standard feature parameters;

[0089] Perform change trend transformation and integration on the time series of the target standard feature parameters to obtain a dynamic change value series;

[0090] Performing principal element extraction on the dynamically changing value sequence based on the fluctuation amplitude of each element in the dynamically changing value sequence and the total correlation degree between each element and other elements to obtain a target dynamically changing value sequence;

[0091] Training a feature vector evaluation model based on the historical feature vector and its corresponding accuracy result of evaluating the physical stimulus load intensity, and inputting the target dynamically changing value sequence into the vector evaluation model to obtain a prediction score;

[0092] When the prediction score is greater than a preset score, using the target dynamically changing value sequence as the change feature vector of the biological signal;

[0093] Otherwise, changing the strategy again to perform principal element extraction on the dynamically changing value sequence to obtain the latest dynamically changing value sequence until the prediction score is greater than the preset score, and using the latest dynamically changing value sequence as the change feature vector of the biological signal.

[0094] In this embodiment, standardizing the time-domain feature parameters and frequency-domain feature parameters, for example, unifying the numerical units, and obtaining the time-series difference of the feature vector as the time-series change feature, etc.

[0095] In this embodiment, the time-domain feature parameters include mean, variance, standard deviation, etc.

[0096] In this embodiment, the frequency-domain feature parameters include spectral center, spectral bandwidth, etc.

[0097] In this embodiment, the higher the number of node layers, the more the number of node splits, and the higher the corresponding important evaluation score.

[0098] In this embodiment, performing a change trend transformation of the time series on the target standard feature parameters as a difference calculation to obtain the change feature in the time dimension.

[0099] In this embodiment, the greater the fluctuation amplitude and the greater the total correlation degree, the greater the possibility of being the main element.

[0100] The beneficial effects of the above design solution are as follows: By obtaining the characteristic parameter types and characteristic parameter value characteristics of the time-domain characteristic parameters and frequency-domain characteristic parameters, standardizing the time-domain characteristic parameters and frequency-domain characteristic parameters to obtain standard characteristic parameters, eliminating the differences caused by different parameter representation forms, establishing a tree model of the standard characteristic parameters based on the correlation characteristics between the standard characteristic parameters, obtaining the node levels and node splitting numbers of each standard characteristic parameter in the tree model, evaluating the importance of each standard characteristic parameter based on the node levels and node splitting numbers to obtain an important evaluation score, selecting the standard characteristic parameters with an important evaluation score greater than the preset evaluation score as the target standard characteristic parameters to avoid redundancy during the integration of characteristic parameters, transforming and integrating the change trends of the target standard characteristic parameters in a time series to obtain a dynamic change value sequence, extracting the main elements of the dynamic change value sequence based on the fluctuation amplitude of each element in the dynamic change value sequence and the total correlation degree between each element and other elements to obtain a target dynamic change value sequence, achieving dimensionality reduction to ensure the conciseness of the sequence and facilitating subsequent analysis, training a feature vector evaluation model based on the historical feature vectors and their corresponding accurate results of the physical stimulus load intensity evaluation, and inputting the target dynamic change value sequence into the vector evaluation model to obtain a prediction score; when the prediction score is greater than the preset score, using the target dynamic change value sequence as the change feature vector of the biological signal; otherwise, changing the strategy again to extract the main elements of the dynamic change value sequence to obtain the latest dynamic change value sequence until the prediction score is greater than the preset score, and using the latest dynamic change value sequence as the change feature vector of the biological signal, which can reflect the dynamic changes of the biological signal in real time, realizing the accurate detection and evaluation of the physical stimulus load intensity based on dynamic modeling, and verifying and optimizing the obtained change sequence based on the evaluation model to improve the accuracy of the final obtained change feature vector for evaluating the physical stimulus load intensity.

[0101] In one embodiment, training a feature vector evaluation model based on the historical feature vectors and their corresponding accurate results of the physical stimulus load intensity evaluation includes:

[0102] Processing the same historical feature vector based on different vector processing methods to obtain corresponding multiple prediction evaluation results, and selecting a target feature processing vector based on the difference between the physical stimulus load intensity evaluation and the prediction evaluation results;

[0103] Calculating a scoring value of the target feature processing vector based on the correlation degree between the elements in the target feature processing vector and the vector similarity between the target feature processing vectors;

[0104]

[0105] where K ωRepresents the scoring value of the ω-th target feature processing vector, and m represents the number of target feature processing vectors. Represents the vector similarity between the ω-th target feature processing vector and the -th other target feature processing vectors, and n represents the number of elements in the target feature processing vector, and σ ωij Represents the correlation between the i-th element and the j-th other element in the ω-th target feature processing vector;

[0106] Select preferred feature processing vectors with scoring values greater than a preset scoring value from the target feature processing vectors;

[0107] Based on the preferred feature processing vectors, calculate the importance of each type of vector element;

[0108]

[0109] Among them, H γ Represents the importance of the γ-th type of vector element, T represents the number of preferred feature processing vectors, and K ρ Represents the scoring value of the ρ-th preferred feature processing vector, and Z γρ Represents the main proportion of the γ-th type of vector element in the ρ-th preferred feature processing vector, and F γρ Represents the relevant proportion of the γ-th type of vector element in the ρ-th preferred feature processing vector, and δ F Represents the relevant proportion weight;

[0110] Based on the importance, perform weight marking on the preferred feature processing vectors, and based on the weight-marked preferred feature processing vectors, perform model training to obtain a feature vector evaluation model.

[0111] In this embodiment, the correlation, similarity, and importance are all standardized, and the values are all between (0, 1).

[0112] In this embodiment, an element refers to one of the parameters in the feature vector.

[0113] In this embodiment, the main proportion of the γ-th type of vector element in the ρ-th preferred feature processing vector is the proportion of the number of the γ-th type of vector element in the ρ-th preferred feature processing vector, and the relevant proportion of the γ-th type of vector element in the ρ-th preferred feature processing vector is the proportion of the number of elements related to the γ-th type of vector element in the ρ-th preferred feature processing vector.

[0114] In this embodiment, the relevant proportion weight is designed in advance according to requirements.

[0115] The beneficial effects of the above design solution are as follows: By processing the same historical feature vector based on different vector processing methods, multiple corresponding prediction and evaluation results are obtained. Based on the difference between the physical stimulus load intensity evaluation and the prediction and evaluation results, the target feature processing vector is selected, thereby providing a vector basis for learning appropriate vectors. Based on the correlation between the elements in the target feature processing vector and the vector similarity between the target feature processing vectors, the scoring value of the target feature processing vector is calculated to obtain the importance of each target feature processing vector for model training. The preferred feature processing vectors with scoring values greater than the preset scoring value are selected from the target feature processing vectors; based on the preferred feature processing vectors, the importance degree of each type of vector element is calculated to obtain the importance of each type of vector element for model training. Based on the importance degree, weight marking is performed on the preferred feature processing vectors, and model training is performed based on the weight-marked preferred feature processing vectors to obtain a feature vector evaluation model. By marking the training parameters, the accuracy of the obtained feature vector evaluation model is ensured, providing a basis for the evaluation of feature vectors.

[0116] In this embodiment, constructing a physical stimulus load intensity evaluation model includes:

[0117] Collect the change feature vectors of the biological signals of no less than one experimenter under different physical stimulus intensities. At the same time, obtain the corresponding actual physical stimulus load intensity values;

[0118] According to the quantization calculation index and the biological signal change feature vector of each experimenter, calculate the time series feature curve graph, and construct the linear regression equation between the biological signal change feature vector of each experimenter and the physical stimulus time series;

[0119] Among them, the quantization calculation index includes various statistical features of the biological signal in the time domain and frequency domain, such as mean value, variance, frequency peak value, etc.;

[0120] Solve the linear regression equation between the biological signal change feature vector of each experimenter and the physical stimulus time series, and determine the change characteristics of the biological signal change feature vector of each experimenter according to the solution results, such as parameters such as the slope and intercept of the change, which can reflect the change trend and amplitude of the biological signal under different physical stimulus times;

[0121] According to the change characteristics of the biological signal change feature vector of each experimenter, extract the load intensity parameters corresponding to multiple physical stimulus load intensity evaluation indicators (heart rate change rate, electromyogram activity intensity change, etc.);

[0122] Perform a pass / fail assessment on the load intensity parameters corresponding to each physical stimulus load intensity assessment index. Based on the assessment results and the comprehensive assessment weight of this physical stimulus load intensity assessment index, determine the assessment score of the load intensity parameter corresponding to this assessment index to reflect the overall load intensity of the experimenter under the current physical stimulus;

[0123] Add up the assessment scores of multiple physical stimulus load intensity assessment indexes of each experimenter to obtain the total calculated performance assessment score of the physical stimulus load intensity corresponding to this experimenter;

[0124] Based on the total calculated performance assessment score of the physical stimulus load intensity of each experimenter, determine whether the physical stimulus load intensity assessment model corresponding to this experimenter meets the effective model screening conditions;

[0125] If the conditions are met, it is considered that this physical stimulus load intensity assessment model can accurately reflect the relationship between the biological signal and the physical stimulus load intensity and can be used for subsequent assessments; otherwise, give a prompt to adjust the experimental conditions or reconstruct the model to ensure the accuracy and reliability of the assessment model.

[0126] In this embodiment, by collecting the biological signal change characteristics of an individual under different stimulus intensities, combining time-domain and frequency-domain statistical characteristics, constructing a personalized linear regression equation, realizing a dynamic and comprehensive assessment of the load intensity, and objectively calculating the assessment score based on indicators such as the heart rate change rate and the myoelectric activity intensity, the accuracy and practicality of the assessment are ensured. Through effective screening, it can accurately reflect the relationship between the biological signal and the physical stimulus load intensity, providing a precise assessment method for the detection of the physical stimulus load intensity.

[0127] In this embodiment, obtaining the corresponding actual physical stimulus load intensity value further includes:

[0128] Based on the physical stimulus device operation parameters of the experimenter during the physical stimulus process, determine the parameter values under the physical stimulus operation stability, biological signal response efficiency, and physical stimulus safety guarantee function indicators;

[0129] Judge whether the operation stability of the physical stimulus is greater than the preset stability threshold. If so, no stability correction is performed. Otherwise, based on the stability difference between the operation stability and the preset stability, perform a first correction on the actual physical stimulus load intensity value. For example, when the stimulus intensity output by the physical stimulus device fluctuates greatly, it indicates unstable operation, and the actual physical stimulus load intensity value can be appropriately reduced;

[0130] After determining that the stability of the physical stimulus meets the requirements, it is judged whether the parameter value under the physical stimulus safety guarantee index is within the preset physical stimulus safety guarantee index value range. When the parameter value exceeds the preset range, a second correction is performed on the actual physical stimulus load intensity value based on the difference between the parameter value and the preset physical stimulus safety guarantee index value range. For example, if the physical stimulus duration exceeds the preset upper limit, the actual physical stimulus load intensity value should be greatly reduced to ensure the safety of the experimenter.

[0131] After determining that the safety guarantee index of the physical stimulus meets the requirements, it is judged whether the biological signal response efficacy is greater than the preset biological signal response efficacy. If the biological signal response efficacy is less than or equal to the preset value, a third correction is performed on the actual physical stimulus load intensity value based on the efficacy difference between the biological signal response efficacy and the preset biological signal response efficacy. If the biological signal of the experimenter responds weakly to the physical stimulus, the actual physical stimulus load intensity value can be appropriately increased to achieve the expected stimulation effect.

[0132] It is judged whether the difference between the load intensity value after the third correction and the actual physical stimulus load intensity value before the correction is within the preset correction range, and the load intensity value after the third correction with the difference within the preset correction range is output. Otherwise, the load intensity value after the third correction is adjusted until the difference between the adjusted load intensity value and the actual physical stimulus load intensity value before the correction is within the preset correction range.

[0133] In this embodiment, the operation stability of the physical stimulus is measured by the fluctuation of parameters such as the stimulus intensity and frequency output by the physical stimulus device; the biological signal response efficiency is evaluated according to the reaction speed, amplitude, and persistence of the experimenter's biological signal to the physical stimulus; the physical stimulus safety guarantee index includes relevant parameters such as the stimulus intensity upper limit, stimulus duration limit, and stimulus interval time to ensure the safety of the experimenter.

[0134] In this embodiment, by real-time collecting and analyzing the biological signals of the experimenter and combining with the operation parameters of the physical stimulus device, the accurate detection and evaluation of the physical stimulus load intensity are realized, improving the accuracy and safety of the experiment, being able to dynamically adapt to the physiological state of the experimenter, optimize the stimulation effect, improve the experimental efficiency, perform three corrections on the actual physical stimulus load intensity value based on the operation parameters of the physical stimulus device, and verify the adjustment, making the load intensity value more in line with the actual situation, being able to dynamically adjust the load intensity according to factors such as the operation stability of the physical stimulus, the biological signal response efficiency, and the safety guarantee index, and improving the scientificity and rationality of the evaluation results.

[0135] To better implement the method for detecting and evaluating the physical stimulus load intensity based on dynamic modeling, the present invention provides a system for detecting and evaluating the physical stimulus load intensity based on dynamic modeling, including:

[0136] A biological signal acquisition module, configured to collect in real time the biological signals of an experimenter during a physical stimulation process, extract time-domain and frequency-domain characteristic parameters from the preprocessed biological signals, and generate a change characteristic vector of the biological signals;

[0137] A load intensity correction module, configured to determine the operation stability of the physical stimulation, the biological signal response efficiency, and the physical stimulation safety guarantee index parameters according to the operation parameters of the physical stimulation device, and correct the actual physical stimulation load intensity value;

[0138] A model construction and evaluation module, configured to construct a linear regression equation between the change characteristic vector of the biological signals and the physical stimulation time series based on the change characteristic vector of the biological signals and the corrected actual physical stimulation load intensity value, determine a physical stimulation load intensity evaluation model, and evaluate the biological signals obtained in real time.

[0139] In this embodiment, the biological signal acquisition module collects and preprocesses the biological signals of the experimenter in real time, extracts time-domain and frequency-domain characteristic parameters therefrom, generates a change characteristic vector, accurately captures the change law of the biological signals, and improves the evaluation accuracy and reliability; the load intensity correction module corrects the actual load intensity value according to the operation parameters of the physical stimulation device, comprehensively considering the operation stability, the biological signal response efficiency, and the safety guarantee index parameters, so that the load intensity is more in line with the actual situation, and ensures that the evaluation result can truly reflect the impact of the physical stimulation on the experimenter; the model construction and evaluation module constructs a linear regression equation based on the change characteristic vector and the corrected load intensity value, determines the evaluation model, accurately reflects the relationship between the biological signals and the load intensity, evaluates the real-time biological signals, and provides strong support for timely adjusting the physical stimulation parameters and optimizing the experimental process.

[0140] As described above, the above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A method for detecting and evaluating the intensity of physical stimulus load based on dynamic modeling, characterized in that, Including: Collecting in real time the biological signals of the experimenter during the physical stimulation process, obtaining the original data reflecting the body's physiological state, preprocessing and feature extracting the biological signals, determining the variation law of the biological signals in different time periods, and generating the variation feature vector of the biological signals; Establishing the mapping relationship between the variation feature vector of the biological signal feature and the load intensity, constructing the physical stimulation load intensity evaluation model, and evaluating the biological signals collected in real time based on the physical stimulation load intensity evaluation model to obtain the evaluation result of the physical stimulation load intensity.

2. The physical stimulation load intensity detection and evaluation method based on dynamic modeling according to claim 1, characterized in that Collecting the biological signals of an individual during the physical stimulation process further includes: Installing the corresponding biological signal sensors at the signal collection parts of the experimenter, and during the physical stimulation process, monitoring and recording the actual biological signals of each biological signal sensor in real time; Performing the first time marking on the actual biological signals according to the collection time, and performing the second time marking on the preset biological signal parameters, where the preset biological signal parameters include the expected heart rate range, electromyogram activity level, and skin conductance change; Performing an alignment operation on the first time marking and the second time marking according to the time characteristics to obtain the alignment result; Based on the alignment result, comparing the difference between the actual biological signal and the preset biological signal parameters at each time point, and performing statistical evaluation to obtain the alignment verification value of the alignment result; Counting the number of time points in the alignment result where the alignment verification value is greater than the preset alignment verification value. When the statistical result is greater than the preset number, determining the alignment result as the target alignment result; otherwise, performing the alignment operation again until the alignment verification condition is met; Based on the target alignment result, synchronizing the actual biological signal and the preset biological signal parameters, calculating the data error value between the actual data and the preset data, and determining the data error range.

3. The method for detecting and evaluating the intensity of physical stimulation load based on dynamic modeling according to claim 2, characterized in that, Determining the data error range further includes: Based on the biological signal parameter type of the actual data corresponding to the data error value, determining the preset threshold corresponding to this type; When the data error value is greater than or equal to the corresponding preset threshold, suspending the physical stimulation load intensity detection process, determining the alarm level according to the size of the data error range, and sending the corresponding alarm according to the alarm level.

4. The physical stimulation load intensity detection and evaluation method based on dynamic modeling according to claim 3, characterized in that Preprocessing and feature extracting the biological signals specifically include: Performing filtering and denoising processing on the biological signals collected in real time, and performing clustering processing on the biological signals based on the biological signal type to obtain each biological signal data group under the load intensity; Determining the target value of each biological signal data group, and drawing the biological signal change curve based on the target value; Determining the total change value from the start to the end of the physical stimulation according to the biological signal change curve. At the same time, according to the time from the start to the end of the physical stimulation, dividing the biological signal change curve into several curve segments; Performing time domain feature analysis on each segmented curve segment, calculating the time domain feature parameters of each curve segment, performing Fourier transform on each curve segment, and determining the frequency domain feature parameters of each biological signal change curve segment; Based on the time domain feature parameters and the frequency domain feature parameters, determining the variation law of the biological signals in different time periods, and generating the variation feature vector of the biological signals.

5. The physical stimulation load intensity detection and evaluation method based on dynamic modeling according to claim 4, characterized in that, Based on time-domain characteristic parameters and frequency-domain characteristic parameters, determine the variation law of the biological signal in different time periods, and generate a variation feature vector of the biological signal, including: Obtain the characteristic parameter types and characteristic parameter value characteristics of the time-domain characteristic parameters and frequency-domain characteristic parameters, standardize the time-domain characteristic parameters and frequency-domain characteristic parameters to obtain standard characteristic parameters; Based on the correlation characteristics between the standard characteristic parameters, establish a tree model of the standard characteristic parameters, obtain the node layer number and node split number of each standard characteristic parameter in the tree model, evaluate the importance of each standard characteristic parameter based on the node layer number and node split number to obtain an important evaluation score, and select the standard characteristic parameters with the important evaluation score greater than the preset evaluation score as the target standard characteristic parameters; Perform transformation and integration on the variation trend of the time series of the target standard characteristic parameters to obtain a dynamic variation value sequence; Based on the fluctuation amplitude of each element in the dynamic variation value sequence and the total correlation degree between each element and other elements, extract the main elements of the dynamic variation value sequence to obtain a target dynamic variation value sequence; Train a feature vector evaluation model based on the historical feature vector and the accuracy result of the corresponding physical stimulus load intensity evaluation, and input the target dynamic variation value sequence into the vector evaluation model to obtain a prediction score; When the prediction score is greater than the preset score, use the target dynamic variation value sequence as the variation feature vector of the biological signal; Otherwise, change the strategy again to extract the main elements of the dynamic variation value sequence to obtain the latest dynamic variation value sequence until the prediction score is greater than the preset score, and use the latest dynamic variation value sequence as the variation feature vector of the biological signal.

6. The physical stimulation load intensity detection and evaluation method based on dynamic modeling according to claim 4, characterized in that Construct a physical stimulus load intensity evaluation model, including: Collect the variation feature vectors of the biological signals of not less than one experimenter under different physical stimulus intensities, and at the same time, obtain the corresponding actual physical stimulus load intensity values; Calculate a time series characteristic curve graph according to the quantization calculation index and the biological signal variation feature vector of each experimenter, and construct a linear regression equation between the biological signal variation feature vector of each experimenter and the physical stimulus time series; Among them, the quantization calculation index includes various statistical characteristics of the biological signal in the time domain and frequency domain; Solve the linear regression equation between the biological signal variation feature vector of each experimenter and the physical stimulus time series, and determine the variation characteristics of the biological signal variation feature vector of each experimenter according to the solution result.

7. The physical stimulation load intensity detection and evaluation method based on dynamic modeling according to claim 6, wherein Constructing a physical stimulus load intensity evaluation model also includes: According to the variation characteristics of the biological signal variation feature vector of each experimenter, extract the load intensity parameters corresponding to multiple physical stimulus load intensity evaluation indicators; Perform a qualified evaluation on the load intensity parameters corresponding to each physical stimulus load intensity evaluation indicator, and determine the evaluation score of the load intensity parameter corresponding to this evaluation indicator according to the evaluation result and the comprehensive evaluation weight of this physical stimulus load intensity evaluation indicator; Add up the evaluation scores of multiple physical stimulus load intensity evaluation indicators of each experimenter to obtain the calculation performance evaluation total score value of the physical stimulus load intensity corresponding to this experimenter; Based on the total score of the computational performance evaluation corresponding to the physical stimulus load intensity of each experimenter, determine whether the physical stimulus load intensity evaluation model corresponding to the experimenter meets the effective model screening conditions; If the conditions are met, it is considered that the physical stimulus load intensity evaluation model can accurately reflect the relationship between the biological signal and the physical stimulus load intensity and can be used for subsequent evaluations; otherwise, a prompt to adjust the experimental conditions or reconstruct the model is issued.

8. The physical stimulation load intensity detection and evaluation method based on dynamic modeling according to claim 7, characterized in that, Obtaining the corresponding actual physical stimulus load intensity value also includes: Based on the operating parameters of the physical stimulus device during the physical stimulus process of the experimenter, determine the parameter values under the operating stability, biological signal response efficiency, and physical stimulus safety guarantee function index of the physical stimulus; Judge whether the operating stability of the physical stimulus is greater than the preset stability threshold. If so, no stability correction is performed. Otherwise, based on the stability difference between the operating stability and the preset stability, the actual physical stimulus load intensity value is first corrected; After determining that the stability of the physical stimulus meets the requirements, judge whether the parameter value under the physical stimulus safety guarantee index is within the preset physical stimulus safety guarantee index value range. When the parameter value exceeds the preset range, based on the difference between the parameter value and the preset physical stimulus safety guarantee index value range, the actual physical stimulus load intensity value is second corrected; After determining that the safety guarantee index of the physical stimulus meets the requirements, judge whether the biological signal response efficiency is greater than the preset biological signal response efficiency. If the biological signal response efficiency is less than or equal to the preset value, based on the efficiency difference between the biological signal response efficiency and the preset biological signal response efficiency, the actual physical stimulus load intensity value is third corrected; Judge whether the difference between the load intensity value after the third correction and the actual physical stimulus load intensity value before the correction is within the preset correction range, and output the load intensity value after the third correction with the difference within the preset correction range. Otherwise, adjust the load intensity value after the third correction until the difference between the adjusted load intensity value and the actual physical stimulus load intensity value before the correction is within the preset correction range.

9. The physical stimulus load intensity detection and evaluation method based on dynamic modeling according to claim 5, characterized in that, Model training is performed on the basis of historical feature vectors and their corresponding accuracy results of the physical stimulus load intensity evaluation to obtain a feature vector evaluation model, including: Process the same historical feature vector based on different vector processing methods to obtain corresponding multiple prediction evaluation results. Based on the difference between the physical stimulus load intensity evaluation and the prediction evaluation results, select the target feature processing vector; Calculate the scoring value of the target feature processing vector based on the correlation between the elements in the target feature processing vector and the vector similarity between the target feature processing vectors; Select the preferred feature processing vectors with scoring values greater than the preset scoring value from the target feature processing vectors; Based on the preferred feature processing vectors, calculate the importance degree of each type of vector element; Perform weight marking on the preferred feature processing vectors based on the importance degree, and perform model training based on the weight-marked preferred feature processing vectors to obtain a feature vector evaluation model.

10. A physical stimulation load intensity detection and evaluation system based on dynamic modeling, which is applied in a physical stimulation load intensity detection and evaluation method based on dynamic modeling as described in claim 8, characterized in that, Including: A bio-signal acquisition module, configured to collect in real time the bio-signals of an experimenter during a physical stimulation process, extract time-domain and frequency-domain characteristic parameters from the preprocessed bio-signals, and generate a change feature vector of the bio-signals; A load intensity correction module, configured to determine the operation stability of the physical stimulation, the bio-signal response efficiency, and the physical stimulation safety guarantee index parameters according to the operation parameters of the physical stimulation device, and correct the actual physical stimulation load intensity value; A model construction and evaluation module, configured to construct a linear regression equation between the change feature vector of the bio-signals and the physical stimulation time series based on the change feature vector of the bio-signals and the corrected actual physical stimulation load intensity value, determine a physical stimulation load intensity evaluation model, and evaluate the bio-signals obtained in real time.

Images