Remote monitoring method and system for electromyographic signals

Abstract

The invention discloses an electromyographic signal remote monitoring method and system, and relates to the technical field of electromyographic signal monitoring, and the method comprises the following steps: 1, continuously collecting original electromyographic signals of upper eyelid skin of a current patient in the eye opening and closing process through a flexible microelectrode array attached to the surface of the upper eyelid skin of the current patient, and outputting the original electromyographic signals as electromyographic data, the method comprises the following steps: step 1, acquiring basic information data of a current patient from a medical information terminal, and acquiring physiological data of the current patient at the same time, step 2, calling myoelectricity data, lesion states and corresponding physiological data of historical patients from a hospital database, and performing analysis in combination with historical similar patient data. Therefore, the current lesion state of the patient is more accurately judged, the accuracy and reliability of the lesion analysis process are greatly improved, a doctor can more deeply know the condition of the patient and formulate a corresponding treatment scheme, and the treatment effect on the patient is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electromyographic signal monitoring, in particular to an electromyographic signal remote monitoring method and system. BACKGROUND

[0002] Myasthenia gravis, as an autoimmune disease caused by neuromuscular junction transmission dysfunction, seriously affects the quality of life of patients, and its typical symptoms include ptosis, diplopia, limb weakness, etc., and the symptoms are volatile, which greatly disturbs the daily activities and physical health of patients. When analyzing the myasthenia gravis lesion of the upper eyelid skin of a patient, the traditional analysis process often relies on short-range invasive examination, such as repetitive nerve electrical stimulation, single-fiber electromyography, etc. First, the symptoms of myasthenia gravis patients are volatile, and these examinations are short in time, and the examination period may just be the time when the patient's symptoms are relieved, causing false negatives of the examination results. Secondly, traditional examinations are not sensitive enough for ocular myasthenia gravis patients, and most myasthenia gravis patients have ocular muscle involvement. The positive rate of traditional examinations for muscles such as trapezius and deltoid muscles is high, and some ocular myasthenia gravis patients usually have no involvement of these muscles. Thirdly, invasive examination is complex and often requires professional medical equipment and technical personnel, thereby increasing the medical cost and diagnosis time, and these examinations will cause obvious discomfort to the patient, and even cause some complications, resulting in high resistance of the patient to the examination, especially for some patients with severe illness, combined with heart disease or weak body, who are difficult to withstand the stimulation of high-intensity current of repetitive electrical stimulation and repeated invasive examination. At the same time, when detecting the myasthenia gravis of the upper eyelid of a patient, due to individual differences of the patient, these difference factors will affect the detection of electromyographic signal and lesion state, and the traditional analysis process often fails to fully consider these individual differences and comprehensive influencing factors, and only makes judgments based on immediate and limited data, which is easy to cause deviation of the analysis results, and thus fails to judge the real state and development trend of the lesion, resulting in misdiagnosis or missed diagnosis and delay of the treatment opportunity of the patient.

[0003] Therefore, in view of the above, the present application provides an electromyographic signal remote monitoring method and system to make up for and improve the shortcomings of the prior art. SUMMARY

[0004] To solve the above technical problems, the present application provides an electromyographic signal remote monitoring method and system to solve the corresponding technical problems in the background art.

[0005] To achieve the above purposes, the technical scheme adopted by the present application is as follows: an electromyographic signal remote monitoring method, comprising the following steps: Step one, continuously collecting original electromyography signals of the upper eyelid skin of the current patient during the blinking process of the current patient through a flexible microelectrode array attached to the skin surface of the upper eyelid of the current patient, and outputting as electromyography data, obtaining basic information data of the current patient from a medical information terminal, and simultaneously obtaining physiological data of the current patient; Step two, retrieving electromyography data, lesion states and corresponding physiological data of historical patients from a hospital database, integrating the basic information data, physiological data and electromyography data of the current patient into a basic data set, and simultaneously integrating the electromyography data, lesion states and corresponding physiological data of the historical patients into a reference data set; Step three, extracting features from the electromyography data of the current patient to obtain a fatigue index and a frequency offset, comparing the fatigue index with a standard fatigue index range and comparing the frequency offset with a standard offset threshold according to a preset standard fatigue index range and standard offset threshold, constructing a support vector machine model based on machine learning according to the comparison results in combination with the basic data set of the current patient and the reference data set of the historical patients, and determining the lesion state of the current patient; Step four, obtaining lesion data of historical patients from a hospital database, screening out influencing factors associated with the lesion states of the historical patients using a Pearson correlation coefficient analysis method, and forming an associated factor list, constructing an influencing factor analysis model, collecting data of corresponding influencing factors of the current patient and substituting them into the influencing factor analysis model, calculating regression coefficients of the influencing factors to obtain quantized values of the influencing factors, and sending the lesion state of the current patient and the quantized values of the influencing factors to a physician terminal.

[0006] As preferred, the electromyography data includes movement amplitude and movement frequency, the basic information data includes age, gender, occupation and medical history record, the physiological data includes sleep duration, stress level and eye use degree, and the lesion state includes normal state and abnormal state.

[0007] As preferred, the specific process of comparing the fatigue index with the standard fatigue index range and comparing the frequency offset with the standard offset threshold is as follows: S101, obtaining electromyography data of the current patient and processing to obtain pure electromyography signals , calculating integral electromyography value iEMG to quantify the contraction strength of the muscle, and the formula is: ; Wherein, and are the start time and end time of a complete blinking cycle, respectively; linearly converting the integral electromyography value to convert the integral electromyography value to a fatigue index F, and the formula is: ; wherein a and b are both constants; S102, performing fast Fourier transform on the pure electromyogram signal to convert the pure electromyogram signal from time domain to frequency domain to obtain a frequency domain signal wherein f is frequency; determining a main frequency band range according to the energy distribution of the frequency domain signal and calculating the main frequency band energy The formula is: ; calculating the total energy of the entire frequency band The formula is: ; wherein, is the highest frequency of the signal; calculating the main frequency band energy proportion The formula is: ; According to the preset standard main frequency band energy proportion, the calculated main frequency band energy proportion is compared with the standard main frequency band energy proportion to obtain the frequency offset ; S103, according to the preset standard fatigue index range , the calculated fatigue index F is compared with the standard fatigue index range, if , , both indicate that the fatigue index is abnormal; According to the preset standard offset threshold , the calculated frequency offset D is compared with the standard offset threshold, if , it indicates that the frequency offset is abnormal.

[0008] As preferred, the specific process of determining the lesion state of the current patient is as follows: S201, obtaining a basic data set of a current patient and a reference data set of historical patients, selecting a radial basis function as a kernel function, the formula is: wherein, and are input samples, is a kernel parameter; S202, construct a support vector machine model based on machine learning, randomly divide a part of data in the reference data set as a training set, and another part as a test set, obtain the basic information data of the historical patients, and take the basic information data, electromyographic data and corresponding physiological data of the historical patients as input feature vectors ifv, and take the lesion state of the historical patients as labels bq, wherein the normal state label is 0 and the abnormal state label is 1; The training set is used to train the support vector machine model, the parameters of the model are adjusted by the sequential minimal optimization algorithm, the trained model is evaluated by using the test set, the kernel parameters and penalty parameters of the model are adjusted according to the evaluation result, and the training and evaluation process is repeated until the performance of the model reaches the optimal; S203, obtaining the comparison result and converting the comparison result into a numerical form, recording the fatigue index normal as 0 and the fatigue index abnormal as 1, recording the frequency offset normal as 0 and the frequency offset abnormal as 1, and forming a feature vector containing the comparison result; The basic data set of the current patient and the feature vector formed by the comparison result are input into the optimized support vector machine model, the support vector machine model classifies the current patient according to the classification hyperplane, and outputs the lesion state label ds of the current patient, so as to determine the lesion state of the current patient, if ds=0, it is determined that the lesion state of the current patient is normal, and if ds=1, it is determined that the lesion state of the current patient is abnormal.

[0009] As preferred, the specific process of screening the influence factors associated with the lesion state of the historical patients is as follows: S301, obtaining the lesion data of the historical patients, integrating the basic information data, electromyographic data, lesion state and corresponding physiological data of the historical patients into a comprehensive data set based on the basic information data, electromyographic data, lesion state and corresponding physiological data of the historical patients, and combining the basic information data, physiological data and electromyographic data of the current patient, and performing one-hot encoding processing on the classification variables in the comprehensive data set and standardizing the continuous variables; S302, using Pearson correlation coefficient analysis method, taking each data of the historical patients as sample, calculating the Pearson correlation coefficient r between each influence factor and the lesion state of the historical patients, and the formula is: ; Wherein, is the influence factor of the i th sample; is the average value of the influence factor value; is the lesion state value of the i th sample; is the average value of the lesion state value; n is the sample number; Based on the calculated Pearson correlation coefficient and sample size, the t-statistic is calculated using the following formula: Where n-2=df, df is the degree of freedom. Using the properties of the t-distribution, based on the calculated t-statistic and degrees of freedom, the corresponding two-sided p-value can be obtained by consulting the t-distribution table. S303. According to the preset correlation coefficient threshold ry, if If the influencing factors corresponding to the Pearson correlation coefficient are initially retained, and based on the preset significance level px, if If the correlation is significant, random interference is excluded, and only factors that meet both of the above conditions are retained to screen out factors that are associated with the disease status of historical patients and form a list of related factors.

[0010] As a preferred method, the specific process for outputting the quantitative values ​​of each influencing factor is as follows: S401. Select a multiple linear regression model and construct an influencing factor analysis model. The formula is as follows: ; Where h represents the disease state; These are the identified related influencing factors; The intercept; These are the regression coefficients; This is the error term; Using historical patient data as the training set, regression coefficients are estimated using the least squares method to minimize the error between the predicted value and the actual disease state, thus obtaining a well-trained influencing factor analysis model. S402. Obtain a list of related factors and collect data on each influencing factor corresponding to the current patient. Substitute these data into the trained influencing factor analysis model and calculate the regression coefficient of each influencing factor. The absolute value of the regression coefficient indicates the degree of influence of the influencing factor on the disease state. The larger the absolute value, the greater the influence. The sign of the regression coefficient indicates the direction of the relationship between the influencing factor and the disease state. A positive sign indicates that the increase of the influencing factor will aggravate the disease, and a negative sign indicates that the increase of the influencing factor will alleviate the disease. S403. Normalize the regression coefficients and map the results to... The interval is used to obtain the quantitative values ​​of each influencing factor, and the current lesion status of the patient and the quantitative values ​​of each influencing factor are compiled and sent to the physician.

[0011] A remote electromyography (EMG) signal monitoring system, comprising a data acquisition unit, a data integration unit, a lesion analysis unit, and a quantification output unit; The data acquisition unit is used to continuously collect the original electromyographic signals of the upper eyelid skin during the blinking process of the current patient through a flexible microelectrode array attached to the surface of the upper eyelid skin of the current patient, and output it as electromyographic data. It also obtains the basic information data of the current patient from the medical information terminal, and at the same time obtains the physiological data of the current patient and sends it to the data integration unit. The data integration unit is used to retrieve historical patients' electromyography data, lesion status, and corresponding physiological data from the hospital database, integrate the current patient's basic information data, physiological data, and electromyography data into a basic dataset, and integrate historical patients' electromyography data, lesion status, and corresponding physiological data into a reference dataset and send it to the lesion analysis unit. The lesion analysis unit is used to extract features from the current patient's electromyography data to obtain fatigue index and frequency offset. Based on the preset standard fatigue index range and standard offset threshold, the fatigue index is compared with the standard fatigue index range and the frequency offset is compared with the standard offset threshold. Based on the comparison results, combined with the current patient's basic dataset and the historical patient's reference dataset, a machine learning-based support vector machine model is constructed to determine the current patient's lesion status and send it to the quantization output unit. The quantitative output unit is used to obtain historical patient lesion data from the hospital database, use Pearson correlation coefficient analysis to screen out influencing factors that are associated with the lesion status of historical patients, form a list of associated factors, construct an influencing factor analysis model, collect data of each influencing factor corresponding to the current patient and substitute them into the influencing factor analysis model, calculate the regression coefficient of each influencing factor to obtain the quantitative value of each influencing factor, and organize and send the current patient's lesion status and the quantitative values ​​of each influencing factor to the physician.

[0012] Compared with existing technologies, the beneficial effects of this invention are as follows: It continuously collects raw electromyographic (EMG) signals from the upper eyelid skin during the blinking process of the current patient using a flexible microelectrode array and outputs them as EMG data. Simultaneously, it acquires the current patient's basic information and physiological data, retrieves EMG data, lesion states, and corresponding physiological data from historical patients, and integrates them into a reference dataset. It then integrates the current patient's basic information, physiological data, and EMG data into a basic dataset. Feature extraction is performed on the current patient's EMG data to obtain the fatigue index and frequency offset. Combining the basic and reference datasets, the current patient's lesion state is determined. Historical patient lesion data is acquired, and influencing factors correlated with historical patient lesion states are screened out. Data on each influencing factor corresponding to the current patient is collected, and the quantitative values ​​of each influencing factor are obtained. The current patient's lesion state and the quantitative values ​​of each influencing factor are then compiled and sent to the physician's end. This allows for analysis in conjunction with historical data from similar patients, thereby more accurately determining the current patient's lesion state. This significantly improves the accuracy and reliability of the lesion analysis process, enabling physicians to gain a deeper understanding of the patient's condition, formulate corresponding treatment plans, and improve treatment outcomes. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of the overall structure of a preferred embodiment of the present invention; Figure 2 This is a schematic diagram of the overall structure of the remote monitoring system shown in this invention. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] Embodiments of the present invention: Please refer to Figure 1 As shown, the remote monitoring method for electromyography signals includes the following steps: Step 1: The raw electromyographic signals of the upper eyelid skin during the blinking process of the current patient are continuously collected by a flexible microelectrode array attached to the surface of the upper eyelid skin of the current patient, and output as electromyographic data. The basic information data of the current patient is obtained from the medical information terminal, and the physiological data of the current patient is also obtained. Step 2: Retrieve historical patients' electromyography (EMG) data, lesion status, and corresponding physiological data from the hospital database. Integrate the current patient's basic information data, physiological data, and EMG data into a basic dataset, and simultaneously integrate historical patients' EMG data, lesion status, and corresponding physiological data into a reference dataset. Step 3: Extract features from the current patient's electromyography data to obtain the fatigue index and frequency offset. Based on the preset standard fatigue index range and standard offset threshold, compare the fatigue index with the standard fatigue index range and compare the frequency offset with the standard offset threshold. Based on the comparison results, combine the current patient's basic dataset and the historical patient's reference dataset to construct a machine learning-based support vector machine model to determine the current patient's disease status. Step 4: Obtain historical patient lesion data from the hospital database, use Pearson correlation coefficient analysis to screen out influencing factors that are associated with the lesion status of historical patients, form a list of associated factors, construct an influencing factor analysis model, collect data of each influencing factor corresponding to the current patient and substitute them into the influencing factor analysis model, calculate the regression coefficient of each influencing factor to obtain the quantitative value of each influencing factor, and compile and send the current patient's lesion status and the quantitative values ​​of each influencing factor to the physician.

[0016] Electromyographic data includes amplitude and frequency of movement; basic information data includes age, gender, occupation, and medical history; physiological data includes sleep duration, stress level, and eye strain; and pathological states include normal and abnormal states.

[0017] The specific process of comparing the fatigue index with the standard fatigue index range and comparing the frequency offset with the standard offset threshold is as follows: S101. Acquire and process the current patient's electromyography (EMG) data to obtain a pure EMG signal. The integrated electromyography (iEMG) value is calculated to quantify the muscle contraction intensity. The formula is as follows: ; in, and These represent the start and end times of a complete blink cycle; A linear transformation is performed on the integrated electromyography (EMG) values ​​to convert them into a fatigue index F, using the following formula: ; Where a and b are both constants; S102. Perform a Fast Fourier Transform on the pure electromyography (EMG) signal to convert it from the time domain to the frequency domain, obtaining the frequency domain signal. where f is the frequency; Determine the main frequency band range based on the energy distribution of the frequency domain signal. And the main frequency band energy was calculated. Its formula is: ; Calculate the total energy of the entire frequency band Its formula is: ; in, The highest frequency of the signal; Calculate the energy percentage of the main frequency band Its formula is: ; Based on the preset standard main frequency band energy ratio, the calculated main frequency band energy ratio is compared with the standard main frequency band energy ratio. By comparison, the frequency offset is obtained. ; S103. According to the preset standard fatigue index range The calculated fatigue index F is compared with the standard fatigue index range. If... , All of these indicate an abnormal fatigue index; According to the preset standard offset threshold The calculated frequency offset D is compared with the standard offset threshold. If... This indicates an abnormal frequency offset.

[0018] The specific process for determining the current disease state of a patient is as follows: S201. Obtain the current patient's baseline dataset and the historical patient's reference dataset, and select the radial basis function as the kernel function. The formula is as follows: ,in, and For the input sample, For kernel parameters; S202. Construct a support vector machine model based on machine learning. Randomly divide a portion of the data from the reference dataset as the training set and another portion as the test set. Obtain basic information data of historical patients and use the basic information data, electromyography data, and corresponding physiological data of historical patients as the input feature vector ifv. Use the disease state of historical patients as the label bq, where the label of normal state is 0 and the label of abnormal state is 1. The support vector machine model is trained using the training set, and the model parameters are adjusted using the sequence minimization algorithm. The trained model is evaluated using the test set, and the kernel parameters and penalty parameters are adjusted based on the evaluation results. The training and evaluation process is repeated until the model's performance reaches its optimal level. S203. Obtain the comparison results and convert them into numerical form. Record the normal fatigue index as 0 and the abnormal fatigue index as 1. Record the normal frequency offset as 0 and the abnormal frequency offset as 1 to form a feature vector containing the comparison results. The basic dataset of the current patient and the feature vector formed by the comparison results are input into the optimized support vector machine model. The support vector machine model classifies the current patient according to the classification hyperplane and outputs the lesion status label ds of the current patient to determine the lesion status of the current patient. If ds=0, the lesion status of the current patient is determined to be normal. If ds=1, the lesion status of the current patient is determined to be abnormal.

[0019] The specific process for screening influencing factors associated with the patient's historical disease status is as follows: S301. Obtain the lesion data of historical patients. Based on the basic information data, electromyography data, lesion status and corresponding physiological data of historical patients, and combine them with the basic information data, physiological data and electromyography data of current patients, integrate them into a comprehensive dataset. Perform one-hot encoding on the categorical variables in the comprehensive dataset and standardize the continuous variables. S302. Using the Pearson correlation coefficient analysis method, and taking historical patient data as a sample, calculate the Pearson correlation coefficient r between each influencing factor and the disease status of historical patients. The formula is as follows: ; in, For the i-th sample, the influencing factors are: This represents the average value of the influencing factors. Let be the lesion status value of the i-th sample; This represents the average value of the disease state values; n is the number of samples; Based on the calculated Pearson correlation coefficient and sample size, the t-statistic is calculated using the following formula: Where n-2=df, df is the degree of freedom. Using the properties of the t-distribution, based on the calculated t-statistic and degrees of freedom, the corresponding two-sided p-value can be obtained by consulting the t-distribution table. S303. According to the preset correlation coefficient threshold ry, if If the influencing factors corresponding to the Pearson correlation coefficient are initially retained, and based on the preset significance level px, if If the correlation is significant, random interference is excluded, and only factors that meet both of the above conditions are retained to screen out factors that are associated with the disease status of historical patients and form a list of related factors.

[0020] The specific process for outputting the quantitative values ​​of each influencing factor is as follows: S401. Select a multiple linear regression model and construct an influencing factor analysis model. The formula is as follows: ; Where h represents the disease state; These are the identified related influencing factors; The intercept; These are the regression coefficients; This is the error term; Using historical patient data as the training set, regression coefficients are estimated using the least squares method to minimize the error between the predicted value and the actual disease state, thus obtaining a well-trained influencing factor analysis model. S402. Obtain a list of related factors and collect data on each influencing factor corresponding to the current patient. Substitute these data into the trained influencing factor analysis model and calculate the regression coefficient of each influencing factor. The absolute value of the regression coefficient indicates the degree of influence of the influencing factor on the disease state. The larger the absolute value, the greater the influence. The sign of the regression coefficient indicates the direction of the relationship between the influencing factor and the disease state. A positive sign indicates that the increase of the influencing factor will aggravate the disease, and a negative sign indicates that the increase of the influencing factor will alleviate the disease. S403. Normalize the regression coefficients and map the results to... The interval is used to obtain the quantitative values ​​of each influencing factor, and the current lesion status of the patient and the quantitative values ​​of each influencing factor are compiled and sent to the physician.

[0021] Please refer to Figure 2 As shown, the electromyography signal remote monitoring system includes a data acquisition unit, a data integration unit, a lesion analysis unit, and a quantitative output unit. The data acquisition unit is used to continuously collect the raw electromyographic signals of the upper eyelid skin during the blinking process of the current patient through a flexible microelectrode array attached to the surface of the upper eyelid skin of the current patient, and output it as electromyographic data. It also obtains the basic information data of the current patient from the medical information terminal, and at the same time obtains the physiological data of the current patient and sends it to the data integration unit. The data integration unit is used to retrieve historical patients' electromyography data, lesion status, and corresponding physiological data from the hospital database, integrate the current patient's basic information data, physiological data, and electromyography data into a basic dataset, and integrate historical patients' electromyography data, lesion status, and corresponding physiological data into a reference dataset and send it to the lesion analysis unit. The lesion analysis unit is used to extract features from the current patient's electromyography data to obtain the fatigue index and frequency offset. Based on the preset standard fatigue index range and standard offset threshold, the fatigue index is compared with the standard fatigue index range and the frequency offset is compared with the standard offset threshold. Based on the comparison results, combined with the current patient's basic dataset and the historical patient's reference dataset, a machine learning-based support vector machine model is constructed to determine the current patient's lesion status and send it to the quantization output unit. The quantitative output unit is used to obtain historical patient lesion data from the hospital database, use Pearson correlation coefficient analysis to screen out influencing factors that are associated with the lesion status of historical patients, form a list of associated factors, construct an influencing factor analysis model, collect data of each influencing factor corresponding to the current patient and substitute them into the influencing factor analysis model, calculate the regression coefficient of each influencing factor to obtain the quantitative value of each influencing factor, and then organize and send the current patient's lesion status and the quantitative values ​​of each influencing factor to the physician.

[0022] This system continuously acquires raw electromyographic (EMG) signals from the upper eyelid skin during blinking using a flexible microelectrode array, outputting the data as EMG data. Simultaneously, it acquires the patient's basic information and physiological data, retrieves EMG data, disease states, and corresponding physiological data from historical patients, and integrates them into a reference dataset. The system then integrates the current patient's basic information, physiological data, and EMG data into a basic dataset. Feature extraction is performed on the current patient's EMG data to obtain fatigue index and frequency offset. Combining the basic and reference datasets, the current patient's disease state is determined. Historical patient disease data is retrieved, and influencing factors correlated with historical patient disease states are identified. Data on each influencing factor corresponding to the current patient are collected, and quantitative values ​​for each factor are obtained. The current patient's disease state and the quantitative values ​​of each influencing factor are then compiled and sent to the physician. This allows for analysis in conjunction with historical data from similar patients, leading to a more accurate assessment of the current patient's disease state. This significantly improves the accuracy and reliability of the disease analysis process, enabling physicians to gain a deeper understanding of the patient's condition, develop appropriate treatment plans, and improve treatment outcomes.

[0023] The size of the interval and threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value.

[0024] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation. In the two embodiments provided in this application, it should be understood that the disclosed apparatus and system can be implemented in other ways; for example, the apparatus embodiments described above are merely illustrative, and the division of modules is merely a logical functional division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed; furthermore, the coupling or direct coupling or communication connection between the shown or discussed mutuals can be through some interfaces, and the indirect coupling or communication connection between the apparatus or modules can be electrical, mechanical or other forms. The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for remote monitoring of electromyographic signals, characterized in that, Includes the following steps: Step 1: The raw electromyographic signals of the upper eyelid skin during the blinking process of the current patient are continuously collected by a flexible microelectrode array attached to the surface of the upper eyelid skin of the current patient, and output as electromyographic data. The basic information data of the current patient is obtained from the medical information terminal, and the physiological data of the current patient is also obtained. Step 2: Retrieve historical patients' electromyography (EMG) data, lesion status, and corresponding physiological data from the hospital database. Integrate the current patient's basic information data, physiological data, and EMG data into a basic dataset, and simultaneously integrate historical patients' EMG data, lesion status, and corresponding physiological data into a reference dataset. Step 3: Extract features from the current patient's electromyography data to obtain the fatigue index and frequency offset. Based on the preset standard fatigue index range and standard offset threshold, compare the fatigue index with the standard fatigue index range and compare the frequency offset with the standard offset threshold. Based on the comparison results, combine the current patient's basic dataset and the historical patient's reference dataset to construct a machine learning-based support vector machine model to determine the current patient's disease status. Step 4: Obtain historical patient lesion data from the hospital database, use Pearson correlation coefficient analysis to screen out influencing factors that are associated with the lesion status of historical patients, form a list of associated factors, construct an influencing factor analysis model, collect data of each influencing factor corresponding to the current patient and substitute them into the influencing factor analysis model, calculate the regression coefficient of each influencing factor to obtain the quantitative value of each influencing factor, and compile and send the current patient's lesion status and the quantitative values ​​of each influencing factor to the physician.

2. The method for remote monitoring of electromyographic signals according to claim 1, characterized in that, The electromyographic data includes amplitude and frequency of movement; the basic information data includes age, gender, occupation, and medical history; the physiological data includes sleep duration, stress level, and degree of eye strain; and the pathological state includes normal and abnormal states.

3. The method for remote monitoring of electromyographic signals according to claim 2, characterized in that, The specific process of comparing the fatigue index with the standard fatigue index range and comparing the frequency offset with the standard offset threshold is as follows: S101. Acquire and process the current patient's electromyography (EMG) data to obtain a pure EMG signal. The integrated electromyography (iEMG) value is calculated to quantify the muscle contraction intensity. The formula is as follows: ； in, and These represent the start and end times of a complete blink cycle; A linear transformation is performed on the integrated electromyography (EMG) values ​​to convert them into a fatigue index F, using the following formula: ； Where a and b are both constants; S102. Perform a Fast Fourier Transform on the pure electromyography (EMG) signal to convert it from the time domain to the frequency domain, obtaining the frequency domain signal. where f is the frequency; Determine the main frequency band range based on the energy distribution of the frequency domain signal. And the main frequency band energy was calculated. Its formula is: ； Calculate the total energy of the entire frequency band Its formula is: ； in, The highest frequency of the signal; Calculate the energy percentage of the main frequency band Its formula is: ； Based on the preset standard main frequency band energy ratio, the calculated main frequency band energy ratio is compared with the standard main frequency band energy ratio. By comparison, the frequency offset is obtained. ; S103. According to the preset standard fatigue index range The calculated fatigue index F is compared with the standard fatigue index range. If... , All of these indicate an abnormal fatigue index; According to the preset standard offset threshold The calculated frequency offset D is compared with the standard offset threshold. If... This indicates an abnormal frequency offset.

4. The method for remote monitoring of electromyographic signals according to claim 3, characterized in that, The specific process for determining the current disease state of a patient is as follows: S201. Obtain the current patient's baseline dataset and the historical patient's reference dataset, and select the radial basis function as the kernel function. The formula is as follows: ,in, and For the input sample, For kernel parameters; S202. Construct a support vector machine model based on machine learning. Randomly divide a portion of the data from the reference dataset as the training set and another portion as the test set. Obtain basic information data of historical patients and use the basic information data, electromyography data, and corresponding physiological data of historical patients as the input feature vector ifv. Use the disease state of historical patients as the label bq, where the label of normal state is 0 and the label of abnormal state is 1. The support vector machine model is trained using the training set, and the model parameters are adjusted using the sequence minimization algorithm. The trained model is evaluated using the test set, and the kernel parameters and penalty parameters are adjusted based on the evaluation results. The training and evaluation process is repeated until the model's performance reaches its optimal level. S203. Obtain the comparison results and convert them into numerical form. Record the normal fatigue index as 0 and the abnormal fatigue index as 1. Record the normal frequency offset as 0 and the abnormal frequency offset as 1 to form a feature vector containing the comparison results. The basic dataset of the current patient and the feature vector formed by the comparison results are input into the optimized support vector machine model. The support vector machine model classifies the current patient according to the classification hyperplane and outputs the lesion status label ds of the current patient to determine the lesion status of the current patient. If ds=0, the lesion status of the current patient is determined to be normal. If ds=1, the lesion status of the current patient is determined to be abnormal.

5. The method for remote monitoring of electromyographic signals according to claim 4, characterized in that, The specific process for screening influencing factors associated with the patient's historical disease status is as follows: S301. Obtain the lesion data of historical patients. Based on the basic information data, electromyography data, lesion status and corresponding physiological data of historical patients, and combine them with the basic information data, physiological data and electromyography data of current patients, integrate them into a comprehensive dataset. Perform one-hot encoding on the categorical variables in the comprehensive dataset and standardize the continuous variables. S302. Using the Pearson correlation coefficient analysis method, and taking historical patient data as a sample, calculate the Pearson correlation coefficient r between each influencing factor and the disease status of historical patients. The formula is as follows: ； in, For the i-th sample, the influencing factors are: This represents the average value of the influencing factors. Let be the lesion status value of the i-th sample; This represents the average value of the disease state values; n is the number of samples; Based on the calculated Pearson correlation coefficient and sample size, the t-statistic is calculated using the following formula: Where n-2=df, df is the degree of freedom. Using the properties of the t-distribution, based on the calculated t-statistic and degrees of freedom, the corresponding two-sided p-value can be obtained by consulting the t-distribution table. S303. According to the preset correlation coefficient threshold ry, if If the influencing factors corresponding to the Pearson correlation coefficient are initially retained, and based on the preset significance level px, if If the correlation is significant, random interference is excluded, and only factors that meet both of the above conditions are retained to screen out factors that are associated with the disease status of historical patients and form a list of related factors.

6. The method for remote monitoring of electromyographic signals according to claim 5, characterized in that, The specific process for outputting the quantitative values ​​of each influencing factor is as follows: S401. Select a multiple linear regression model and construct an influencing factor analysis model. The formula is as follows: ； Where h represents the disease state; These are the identified related influencing factors; The intercept; These are the regression coefficients; This is the error term; Using historical patient data as the training set, regression coefficients are estimated using the least squares method to minimize the error between the predicted value and the actual disease state, thus obtaining a well-trained influencing factor analysis model. S402. Obtain a list of related factors and collect data on each influencing factor corresponding to the current patient. Substitute these data into the trained influencing factor analysis model and calculate the regression coefficient of each influencing factor. The absolute value of the regression coefficient indicates the degree of influence of the influencing factor on the disease state. The larger the absolute value, the greater the influence. The sign of the regression coefficient indicates the direction of the relationship between the influencing factor and the disease state. A positive sign indicates that the increase of the influencing factor will aggravate the disease, and a negative sign indicates that the increase of the influencing factor will alleviate the disease. S403. Normalize the regression coefficients and map the results to... The interval is used to obtain the quantitative values ​​of each influencing factor, and the current lesion status of the patient and the quantitative values ​​of each influencing factor are compiled and sent to the physician.

7. A remote electromyography (EMG) signal monitoring system, applied to the remote EMG signal monitoring method according to any one of claims 1-6, characterized in that, The remote monitoring system includes a data acquisition unit, a data integration unit, a lesion analysis unit, and a quantitative output unit; The data acquisition unit is used to continuously collect the original electromyographic signals of the upper eyelid skin during the blinking process of the current patient through a flexible microelectrode array attached to the surface of the upper eyelid skin of the current patient, and output it as electromyographic data. It also obtains the basic information data of the current patient from the medical information terminal, and at the same time obtains the physiological data of the current patient and sends it to the data integration unit. The data integration unit is used to retrieve historical patients' electromyography data, lesion status, and corresponding physiological data from the hospital database, integrate the current patient's basic information data, physiological data, and electromyography data into a basic dataset, and integrate historical patients' electromyography data, lesion status, and corresponding physiological data into a reference dataset and send it to the lesion analysis unit. The lesion analysis unit is used to extract features from the current patient's electromyography data to obtain fatigue index and frequency offset. Based on the preset standard fatigue index range and standard offset threshold, the fatigue index is compared with the standard fatigue index range and the frequency offset is compared with the standard offset threshold. Based on the comparison results, combined with the current patient's basic dataset and the historical patient's reference dataset, a machine learning-based support vector machine model is constructed to determine the current patient's lesion status and send it to the quantization output unit. The quantitative output unit is used to obtain historical patient lesion data from the hospital database, use Pearson correlation coefficient analysis to screen out influencing factors that are associated with the lesion status of historical patients, form a list of associated factors, construct an influencing factor analysis model, collect data of each influencing factor corresponding to the current patient and substitute them into the influencing factor analysis model, calculate the regression coefficient of each influencing factor to obtain the quantitative value of each influencing factor, and organize and send the current patient's lesion status and the quantitative values ​​of each influencing factor to the physician.

Images