MST individualized electrical stimulation dose quantification method and device

Abstract

The invention discloses an MST individualized electrical stimulation dose quantification method comprising the following steps: obtaining individual multi-dimensional physiological data of a target patient, and carrying out feature extraction on the individual multi-dimensional physiological data to obtain individual multi-dimensional feature data; constructing an electrical stimulation dose quantification model based on the acquired sample multi-dimensional physiological data and the corresponding electrical stimulation dose parameters subjected to clinical verification; and predicting the individual multi-dimensional feature data based on the electrical stimulation dose quantification model to obtain the electrical stimulation dose of the target patient. According to the invention, the degree of consideration on individual differences of patients is improved, so that the treatment scheme can better fit the actual situation of each patient, and the treatment effect is further obviously improved; in addition, the dose is predicted through the constructed electrical stimulation dose quantification model, subjectivity and uncertainty of human experience judgment are avoided, and potential rules between physiological features and the electrical stimulation dose can be captured more accurately.

Description

Technical Field

[0001] This invention relates to the field of neuroscience technology, and in particular to a method and apparatus for quantifying individualized electrical stimulation doses (MST). Background Technology

[0002] In the field of modern medicine, MST (Motor Cortex Stimulation) is an important treatment method, and the precise determination of its individualized electrical stimulation dose plays a key role in the treatment effect.

[0003] Currently, determining the dosage of electrical stimulation (EMS) relies heavily on the physician's personal experience. This approach is highly subjective and uncertain. Different physicians, due to differences in their educational background, professional experience, and clinical practice, have varying criteria for judging the dosage. For example, when treating Parkinson's disease patients with MST, some physicians may, based on their limited experience with successful cases, tend to use higher stimulation intensity and longer stimulation time to improve the patient's motor symptoms; while others may choose a relatively lower stimulation dose due to concerns about adverse reactions. This difference not only leads to significant inconsistencies in treatment outcomes between different medical institutions and patients but also easily triggers a series of problems. For instance, excessively high EMS doses may cause unnecessary suffering for patients, leading to severe muscle spasms, increased pain, and even potential nerve tissue damage; excessively low doses may fail to achieve the desired therapeutic effect, causing patients to miss the optimal treatment window, resulting in ineffective symptom relief and delayed rehabilitation. Summary of the Invention

[0004] The purpose of this invention is to solve the problems in the prior art, and to propose a method and apparatus for quantifying individualized electrical stimulation doses using MST.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: The method for quantifying individualized electrical stimulation doses using MST includes the following steps: Acquire individual multidimensional physiological data of the target patient, and extract features from the individual multidimensional physiological data to obtain individual multidimensional feature data; Based on the obtained multidimensional physiological data of the samples and their corresponding clinically validated electrical stimulation dose parameters, an electrical stimulation dose quantification model was constructed. Based on the electrical stimulation dose quantification model, the individual's multidimensional feature data are predicted to obtain the electrical stimulation dose for the target patient.

[0006] According to the MST individualized electrical stimulation dose quantification method provided by the present invention, the individual multidimensional physiological data includes electroencephalogram (EEG) information, electromyogram (EMG) data, patient physiological metabolic data, and patient physical indicator data. The step of extracting features from the individual multidimensional physiological data to obtain individual multidimensional feature data includes: Frequency band power features are extracted from the electroencephalogram information to obtain the first feature data; The electromyography data is subjected to activity intensity features, frequency features, and potential features to obtain the second feature data; The patient's physiological metabolic data were subjected to extraction of blood glucose metabolism characteristics, oxygen metabolism characteristics, and enzyme metabolism characteristics to obtain third characteristic data. The patient's physical index data were subjected to basic physical parameter features, detailed medical history features, and cardiovascular index features to obtain fourth feature data; The first feature data, the second feature data, the third feature data, and the fourth feature data are weighted and fused to obtain individual multidimensional feature data.

[0007] According to the MST individualized electrical stimulation dose quantification method provided by the present invention, the extraction of activity intensity features, frequency features, and potential features from the electromyography data to obtain second feature data includes: Based on the root mean square value and integral electromyography value of the electromyography signal in the electromyography data, muscle activity intensity characteristic data are obtained. Power spectrum analysis is performed on the electromyography signals in the electromyography data to determine the main frequency components in the electromyography signals and obtain muscle frequency characteristic data. The electromyography (EMG) signals in the EMG data are decomposed and the MUAP parameters are extracted to obtain muscle action potential characteristic data. The muscle activity intensity characteristic data, the muscle frequency characteristic data, and the muscle action potential characteristic data are combined to obtain the second characteristic data.

[0008] According to the MST individualized electrical stimulation dose quantification method provided by the present invention, the extraction of blood glucose metabolism characteristics, oxygen metabolism characteristics, and enzyme metabolism characteristics from the patient's physiological metabolic data to obtain third characteristic data includes: Blood glucose data were extracted from the patient's physiological metabolic data to obtain blood glucose metabolism characteristic data; Oxygen metabolism indexes were measured in the blood data from the patient's physiological metabolic data to obtain oxygen metabolism characteristic data. Enzyme activity was measured in the patient's physiological metabolic data to obtain enzyme metabolic characteristic data; The blood glucose metabolism characteristic data, oxygen metabolism characteristic data, and enzyme metabolism characteristic data are combined to obtain the third characteristic data.

[0009] According to the MST individualized electrical stimulation dose quantification method provided by the present invention, the extraction of basic physical parameter characteristics, detailed medical history characteristics, and cardiovascular index characteristics from the patient's physical index data to obtain fourth feature data includes: The height, weight, and body surface area of ​​the patient's physical indicators are analyzed and extracted to obtain basic physical parameter feature data. The vascular function commands in the patient's physical index data are measured and extracted to obtain cardiovascular index feature data; Historical medical record data were extracted from the patient's physical indicator data to obtain detailed medical history feature data; The basic physical parameter feature data, the cardiovascular indicator feature data, and the detailed medical history feature data are combined to obtain the fourth feature data.

[0010] According to the MST individualized electrical stimulation dose quantification method provided by the present invention, the model architecture of the electrical stimulation dose quantification model adopts a neural network architecture based on an attention mechanism, and the prediction results of the model are interpreted by feature attribution.

[0011] According to the MST individualized electrical stimulation dose quantification method provided by the present invention, the loss function of the electrical stimulation dose quantification model is: ; ; ; ; Where N represents the number of samples, Represents the parameters of the model. This indicates the actual electrical stimulation dose. M represents the electrical stimulation dose predicted by the model; M represents the number of patient physical indicator samples. This represents the j-th physiological indicator of the i-th sample. Let represent the mean of the physiological indicators of the i-th sample. This represents the mean of the predicted electrical stimulation dose; K represents the number of types of neurological diseases. This represents the number of samples for the k-th disease. This represents the sample index set for the k-th disease. This represents the weight coefficient for the k-th disease; , and Both represent hyperparameters that weigh the importance of the losses of each component.

[0016] The MST individualized electrical stimulation dose quantization device is used in the above-mentioned MST individualized electrical stimulation dose quantization method; the MST individualized electrical stimulation dose quantization device includes: Acquisition unit: used to acquire individual multidimensional physiological data of the target patient, and to extract features from the individual multidimensional physiological data to obtain individual multidimensional feature data; Model building unit: used to build an electrical stimulation dose quantification model based on the acquired multidimensional physiological data of the samples and their corresponding clinically validated electrical stimulation dose parameters; Processing unit: used to predict the individual's multidimensional feature data based on the electrical stimulation dose quantification model to obtain the electrical stimulation dose for the target patient.

[0017] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described MST individualized electrical stimulation dose quantification method.

[0018] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described MST individualized electrical stimulation dose quantification method.

[0019] Compared with the prior art, the present invention has the following advantages: The MST individualized electrical stimulation dose quantification method and device provided by this invention can deeply explore the unique physiological characteristics of patients by acquiring multidimensional physiological data of the target patient. Compared with traditional methods that determine the electrical stimulation dose based on only a single physiological indicator, it improves the consideration of individual patient differences, making the treatment plan more suitable for each patient's actual situation, thereby significantly improving the treatment effect. In addition, by constructing a well-developed electrical stimulation dose quantification model to predict the dose, it avoids the subjectivity and uncertainty of human experience judgment, and can more accurately capture the potential pattern between physiological characteristics and electrical stimulation dose, thereby providing precise electrical stimulation dose recommendations for each patient and improving the scientificity and reliability of the instructions. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1This is a schematic diagram of the MST individualized electrical stimulation dose quantification method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the MST individualized electrical stimulation dose quantification device provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the electronic device proposed in this invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0023] The following is combined with Figure 1 - Figure 3 The present invention describes the MST individualized electrical stimulation dose quantification method and apparatus.

[0024] Figure 1 This is a schematic flowchart of the MST individualized electrical stimulation dose quantification method provided by the present invention. Figure 1 As shown, the method includes: Step 101: Obtain individual multidimensional physiological data of the target patient and extract features from the individual multidimensional physiological data to obtain individual multidimensional feature data.

[0025] Specifically, individual multidimensional physiological data includes electroencephalogram (EEG) information, electromyography (EMG) data, patient physiological metabolic data, and patient physical indicator data. When acquiring individual multidimensional physiological data, sensors are primarily used to collect patient physiological data. After the sensors complete the data collection, the collected data is transmitted in real time to a data integration center. At the data integration center, the collected data undergoes preprocessing, including noise removal, signal enhancement, and data standardization, ultimately yielding preprocessed data. This preprocessed physiological data improves feature extraction efficiency and avoids interference from data impurities.

[0026] Furthermore, after obtaining the various types of preprocessed data, the specific process of obtaining individual multidimensional feature data is described in steps 1011-1015.

[0027] Step 102: Based on the obtained multidimensional physiological data of the samples and their corresponding clinically validated electrical stimulation dose parameters, construct an electrical stimulation dose quantification model.

[0028] Specifically, the multidimensional sample data also includes EEG information, EMG data, patient physiological metabolic data, and patient physical indicator data. Furthermore, the electrical stimulation dose quantification model employs an attention-based neural network architecture and uses feature attribution to interpret the model's predictions. This introduces a multi-head attention mechanism, enabling the model to simultaneously focus on different parts of the input individual multidimensional feature data and capture the complex relationships between features. For example, for EEG features and cardiovascular function features, the model can learn the correlation weights between them through the attention mechanism, thereby more accurately determining the electrical stimulation dose. The model architecture includes multiple hidden layers, each using different activation functions (such as ReLU, LeakyReLU, etc.) to increase the model's non-linear expressive power, and finally outputs the predicted electrical stimulation dose through a fully connected layer.

[0029] Furthermore, in the process of training the pre-built model using multidimensional physiological data of the samples and their corresponding electrical stimulation dose parameters, the multidimensional physiological data of the samples are also preprocessed first, including data cleaning, data standardization, feature selection and dimensionality reduction. In the process of feature selection and dimensionality reduction, feature selection algorithms (such as correlation analysis, recursive feature elimination, etc.) are used to screen out physiological features that are highly correlated with electrical stimulation dose, remove redundant or irrelevant features, and reduce data dimensionality in order to reduce the computational complexity of the model and the risk of overfitting.

[0030] Step 103: Based on the electrical stimulation dose quantification model, predict the individual multidimensional feature data to obtain the electrical stimulation dose for the target patient.

[0031] Specifically, during the treatment of the target patient, the patient's physiological data is first collected in real time and fused with previously extracted individual multidimensional feature data. This data is then input into the electrical stimulation dose quantification model to achieve factual prediction and dynamic adjustment of the electrical stimulation dose. For example, the patient's electroencephalogram (EEG), electromyogram (EMG), and cardiovascular function data are collected at regular intervals (e.g., 5 minutes). These data are combined with previous historical data, and the individual multidimensional feature data is updated using a sliding window method. The model then quickly predicts whether the current electrical stimulation dose needs to be adjusted, as well as the direction and magnitude of the adjustment, based on the updated feature data.

[0032] This invention relates to the field of neuroscience and proposes a method for quantifying individualized electrical stimulation (MST) dosage. The proposed MST method can deeply explore the unique physiological characteristics of patients. Compared to traditional methods that determine the electrical stimulation dosage based on only a single physiological indicator, it improves the consideration of individual patient differences, allowing treatment plans to be more tailored to each patient's specific situation, thereby significantly improving treatment efficacy. Furthermore, by constructing a well-developed electrical stimulation dosage quantification model to predict the dosage, it avoids the subjectivity and uncertainty of human experience-based judgment, and can more accurately capture the potential patterns between physiological characteristics and electrical stimulation dosage. This provides precise electrical stimulation dosage recommendations for each patient, improving the scientific rigor and reliability of the instructions.

[0033] In one embodiment, steps 1011-1015 are as follows: Step 1011: Extract frequency band power features from the EEG information to obtain the first feature data.

[0034] Specifically, using Fast Fourier Transform (FFT) or other spectral analysis techniques, the EEG signals in the EEG information are converted from the time domain to the frequency domain, and the power spectral density (PSD) of different frequency bands (such as delta waves 0.5–4 Hz, theta waves 4–8 Hz, alpha waves 8–13 Hz, beta waves 13–30 Hz, and gamma waves 30–100 Hz) is calculated. The power distribution of these frequency bands reflects the characteristics of neural electrical activity in the brain under different states of consciousness and functional activities. By extracting the power characteristics of each frequency band, the basic electrophysiological state of the patient's brain can be understood, as well as the differences compared with the normal population. These differences may be related to neurological diseases or functional disorders, thus providing a reference for determining the electrical stimulation dose.

[0035] Step 1012: Extract activity intensity features, frequency features, and potential features from the electromyography data to obtain the second feature data.

[0036] Specifically, muscle activity intensity characteristics are obtained based on the root mean square and integrated electromyographic (EMG) values ​​of the EMG signals in the EMG data. For discrete EMG signals... The formula for calculating its root mean square value is: The root mean square (RMS) value reflects the average power level of the electromyography (EMG) signal over a period of time and is closely related to the muscle contraction intensity. The formula for calculating the integral EMG value is: Integral electromyography (EMG) values ​​are also used to measure the total amount of muscle activity over a certain period of time. They are also quite sensitive to changes in muscle activity intensity. Compared to the root mean square (RMS) value, the integrated EMG value focuses more on reflecting the cumulative effect of muscle activity. By calculating the RMS and integrated EMG values, two sets of data that accurately reflect the characteristics of muscle activity intensity were obtained.

[0037] Power spectrum analysis was performed on the electromyography (EMG) signals in the EMG data to determine the main frequency components in the EMG signals and obtain muscle frequency characteristic data.

[0038] Specifically, the power spectrum of the electromyography (EMG) signal is calculated using Fast Fourier Transform (FFT) for discrete EMG signals. First, zero-padding is performed to make its length reach the specified value. (where b is an integer) to improve the computational efficiency and frequency resolution of the FFT. Then, the zero-padded signal is subjected to an FFT: Where b = 0, 1, ..., W⁻¹, and h is the imaginary unit. Calculate... Obtain the power spectral density of the signal This represents the energy distribution of the signal across different frequency components. Then, the power spectral density... The analysis is then performed to determine the main frequency components. Typically, an energy threshold or frequency range is set to filter out frequency bands that contribute significant energy. For example, during normal muscle activity, the power spectrum of electromyography (EMG) signals is mainly concentrated within a certain frequency range (e.g., 10–500 Hz). By observing the peak positions and amplitude distribution of the power spectrum, the main frequency components of this muscle activity can be determined.

[0039] The electromyography (EMG) signals in the EMG data are decomposed, and the MUAP parameters are extracted to obtain muscle action potential characteristic data.

[0040] Specifically, Empirical Mode Decomposition (EMD) or wavelet decomposition methods are used to decompose electromyography (EMG) signals. Taking EMD as an example, it decomposes complex EMG signals into a series of intrinsic mode functions (IMFs) with different frequencies and amplitudes. Parameters of the motor unit action potential (MUAP) are extracted from the decomposed IMFs, including the MUAP's amplitude, duration, phase, and area. For example, the MUAP's amplitude can be determined by finding its maximum value on its waveform; the duration can be calculated by measuring the time span of the MUAP from start to finish; the phase can be determined by its relative time relationship with a reference signal (such as a trigger signal or other synchronization signal); and the area can be obtained by integrating the MUAP waveform. MUAP parameters reflect the electrophysiological characteristics of individual motor units and are of great value for assessing muscle innervation, diagnosing muscle diseases, and evaluating the activation effect of electrical stimulation on motor units.

[0041] The second feature data is obtained by combining the muscle activity intensity characteristic data, muscle frequency characteristic data, and muscle action potential characteristic data.

[0042] Specifically, the calculated muscle activity intensity characteristic data (root mean square value and integral electromyography value), muscle frequency characteristic data (main frequency components and their energy distribution), and muscle action potential characteristic data (amplitude, duration, phase, area, etc. of MUAP) are arranged and combined in a certain order to form a feature vector, which serves as the second feature data.

[0043] Step 1013: Extract blood glucose metabolism characteristics, oxygen metabolism characteristics, and enzyme metabolism characteristics from the patient's physiological metabolic data to obtain the third characteristic data.

[0044] Specifically, this includes extracting blood glucose data from the patient's physiological metabolic data to obtain blood glucose metabolism characteristic data.

[0045] Specifically, regarding blood glucose data, in addition to monitoring fasting and postprandial blood glucose levels, indicators such as blood glucose fluctuation amplitude and coefficient of variation can be calculated to assess blood glucose stability. Abnormal blood glucose metabolism is closely related to nervous system function. Hyperglycemia may lead to oxidative stress damage to nerve cells and slowed nerve conduction velocity, while hypoglycemia may cause insufficient energy supply to the brain and neurological dysfunction. By extracting these blood glucose metabolism characteristics, we can understand the impact of a patient's blood glucose metabolism on the nervous system. This allows for adjustments to the electrical stimulation dose and timing during electrical stimulation therapy based on blood glucose levels, avoiding adverse effects on treatment efficacy due to blood glucose fluctuations.

[0046] Oxygen metabolism indicators were measured in blood data from the patient's physiological metabolic data to obtain oxygen metabolism characteristic data.

[0047] Specifically, oxygen metabolism indicators such as blood oxygen saturation (SpO2), arterial partial pressure of oxygen (PaO2), and partial pressure of carbon dioxide (PaCO2) are measured, along with parameters such as oxygen uptake rate and oxygen delivery, to assess tissue oxygen supply and utilization. Because the brain has a very high oxygen demand, abnormal oxygen metabolism can affect brain neural function and electrophysiological activity. By monitoring oxygen metabolism characteristics, oxygen metabolism problems in patients can be detected promptly, and corresponding measures (such as oxygen therapy) can be taken to improve oxygen supply. Furthermore, the impact of oxygen metabolism status on the effectiveness of electrical stimulation therapy can be considered during treatment, and the stimulation dose can be adjusted accordingly.

[0048] Enzyme activity was measured in the patient's physiological metabolic data to obtain enzyme metabolic characteristic data.

[0049] Specifically, the activity of enzymes related to physiological metabolic processes is measured, such as lactate dehydrogenase (LDH), pyruvate kinase (PK), and hexokinase (HK). Changes in these enzyme activities reflect the rate and efficiency of metabolic pathways and are closely related to the body's metabolic functional status. By detecting the characteristics of metabolic enzyme activity, it is possible to understand whether a patient's metabolic function is normal and the potential impact of electrical stimulation therapy on metabolic enzyme activity.

[0050] The third characteristic data is obtained by combining the blood glucose metabolism characteristic data, oxygen metabolism characteristic data, and enzyme metabolism characteristic data.

[0051] Specifically, the extracted glucose metabolism feature data (including statistical features of blood glucose levels, fluctuation features, curve features, etc.), oxygen metabolism feature data (oxygen partial pressure and oxygen saturation features, oxygen uptake and oxygen utilization features, lactate and pyruvate metabolism features, etc.), and enzyme metabolism feature data (activations of key enzymes in glucose metabolism, mitochondrial respiratory chain enzyme activities, oxidoreductase activities, etc.) are arranged and combined in a certain order into a feature vector, which serves as the third feature data.

[0052] Step 1014: Extract basic physical parameter features, detailed medical history features, and cardiovascular indicator features from the patient's physical indicator data to obtain the fourth feature data.

[0053] Specifically, this includes analyzing and extracting height, weight, and body surface area from the patient's physical indicators to obtain basic physical parameter characteristic data.

[0054] Specifically, Body Mass Index (BMI) is a commonly used indicator to measure body fat and health status. In electrical stimulation therapy, BMI may affect the conduction and distribution of electrical stimulation. Height and weight can be obtained through measurement, while body surface area (BSA) is obtained using the DuBois formula: Body surface area is important in studies involving body metabolism, drug distribution, and the distribution of electrical stimulation energy within the body. It can help determine the appropriate electrical stimulation dose for individual patients more accurately, because patients with different body surface areas may respond differently to the same electrical stimulation intensity. Patients with larger body surface areas may require relatively higher electrical stimulation energy to achieve the same therapeutic effect.

[0055] The vascular function indicators in the patient's physical indicators are measured and extracted to obtain the characteristic data of cardiovascular indicators.

[0056] Specifically, systolic blood pressure (SBP), diastolic blood pressure (DBP), and pulse pressure (PP) are measured, and their variation patterns are analyzed. The mean, standard deviation, coefficient of variation, and diurnal rhythm characteristics of blood pressure over a period of time (e.g., 24 hours) are calculated. Simultaneously, various indicators of heart rate variability are calculated, including time-domain indicators (e.g., SDNN – standard deviation of all sinus RR intervals, RMSSD – root mean square of the difference between adjacent RR intervals, pNN50 – percentage of adjacent RR interval differences greater than 50 ms, etc.) and frequency-domain indicators (e.g., LF – low-frequency power, HF – high-frequency power, LF / HF ratio, etc.).

[0057] Detailed medical history data is obtained by extracting historical medical records from the patient's physical indicators.

[0058] Specifically, review the patient's medical history to identify the types of diseases they have previously suffered from (such as hypertension, diabetes, heart disease, and neurological disorders) and the date of the first diagnosis for each disease. Also, record in detail the various treatments the patient received for their previous illnesses (such as medication, surgery, and physical therapy) and their effectiveness.

[0059] The fourth characteristic data is obtained by combining the basic physical parameter characteristic data, cardiovascular indicator characteristic data, and detailed medical history characteristic data.

[0060] Specifically, the extracted basic body parameter feature data (such as BMI, BSA, height-to-weight ratio, etc.), cardiovascular indicator feature data (blood pressure features, HRV index, cardiac output and peripheral vascular resistance, etc.), and detailed medical history feature data (disease type and diagnosis time, treatment history and treatment effect, disease recurrence and complications, etc.) are arranged and combined in a certain order to form a feature vector, which serves as the fourth feature data.

[0061] Furthermore, by integrating various characteristic data such as basic physical parameters, cardiovascular indicators, and detailed medical history, a comprehensive and systematic assessment of the patient's overall health status can be achieved.

[0062] Step 1015: The first feature data, the second feature data, the third feature data, and the fourth feature data are weighted and fused to obtain individual multidimensional feature data.

[0063] Specifically, before weighted fusion, the first feature data (e.g., EEG-related features), the second feature data (EMG-related features), the third feature data (physiological metabolism-related features), and the fourth feature data (body indicator-related features) are first normalized to have the same dimensions and numerical range, facilitating subsequent weighting operations. Examples include Min-Max normalization and Z-Score normalization.

[0064] Furthermore, determining the weighting coefficients is a crucial step in weighted fusion. Taking the Analytic Hierarchy Process (AHP) as an example, a judgment matrix is ​​first constructed to compare the relative importance of each feature data for the quantification of electrical stimulation dosage. By calculating the largest eigenvalue of the judgment matrix and its corresponding eigenvector, the eigenvector is normalized to obtain the weighting coefficients of each feature data. Finally, a weighted multiplication is performed to obtain the final individual multidimensional feature data.

[0065] Step 102, in one embodiment, the loss function of the electrical stimulation dose quantification model is:

[0066] ;in, Represents the parameters of the model. , and These are all hyperparameters that weigh the importance of the losses of each part and can be adjusted according to the actual situation.

[0067]

[0068] Where N represents the sample size. This indicates the actual electrical stimulation dose. This represents the electrical stimulation dose predicted by the model; This represents the basic error loss, ensuring that the model's predictions are generally close to the true values.

[0069]

[0070] Where N represents the sample size, and M represents the number of patient physical indicator samples. This represents the j-th physiological indicator of the i-th sample. Let represent the mean of the physiological indicators of the i-th sample. This represents the mean of the predicted electrical stimulation dose; This means maximizing the correlation loss between the predicted electrical stimulation dose and physiological indicators. By minimizing this loss, the linear relationship between physiological indicators and electrical stimulation dose can be learned, making the prediction results more consistent with the changing trends of physiological indicators. This takes into account the influence of physiological indicators on electrical stimulation dose to a certain extent, and also has a certain robustness to the measurement error of physiological indicators.

[0071]

[0072] Where K represents the number of types of neurological diseases, This represents the number of samples for the k-th disease. This represents the sample index set for the k-th disease. The weight coefficient for the k-th disease can be adjusted based on the importance of different diseases or the number of samples. This represents the specific loss for each neurological disease, which is weighted and summed using weighting coefficients. During training, it can pay more attention to the characteristics of different disease types and learn specific electrical stimulation dosage patterns for each disease, thereby improving the treatment effect for different neurological diseases.

[0073] Figure 2 This is a schematic diagram of the structure of the MST individualized electrical stimulation dose quantification device provided by the present invention, as shown below. Figure 2As shown, the device includes: an acquisition unit 10 for acquiring individual multidimensional physiological data of the target patient and extracting features from the individual multidimensional physiological data to obtain individual multidimensional feature data; a model building unit 20 for constructing an electrical stimulation dose quantification model based on the acquired sample multidimensional physiological data and its corresponding clinically validated electrical stimulation dose parameters; and a processing unit 30 for predicting the individual multidimensional feature data based on the electrical stimulation dose quantification model to obtain the electrical stimulation dose of the target patient.

[0074] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 3 As shown, the electronic device may include a processor 310, a communications interface 320, a memory 330, and a communication bus 340. The processor 310, communications interface 320, and memory 330 communicate with each other via the communication bus 340. The processor 310 can call logical instructions in the memory 330 to execute the MST individualized electrical stimulation dose quantification method. This method includes: acquiring individual multidimensional physiological data of the target patient and extracting features from the individual multidimensional physiological data to obtain individual multidimensional feature data; constructing an electrical stimulation dose quantification model based on the acquired sample multidimensional physiological data and its corresponding clinically validated electrical stimulation dose parameters; and predicting the electrical stimulation dose of the target patient based on the electrical stimulation dose quantification model using the individual multidimensional feature data.

[0075] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0076] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, and when the program instructions are executed by the computer, the computer is able to execute the MST individualized electrical stimulation dose quantification method provided by the above methods. The method includes: acquiring individual multidimensional physiological data of the target patient and extracting features from the individual multidimensional physiological data to obtain individual multidimensional feature data; constructing an electrical stimulation dose quantification model based on the acquired sample multidimensional physiological data and its corresponding clinically validated electrical stimulation dose parameters; and predicting the electrical stimulation dose of the target patient based on the electrical stimulation dose quantification model of the individual multidimensional feature data.

[0077] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the aforementioned MST individualized electrical stimulation dose quantification methods. The method includes: acquiring individual multidimensional physiological data of a target patient and extracting features from the individual multidimensional physiological data to obtain individual multidimensional feature data; constructing an electrical stimulation dose quantification model based on the acquired sample multidimensional physiological data and its corresponding clinically validated electrical stimulation dose parameters; and predicting the electrical stimulation dose of the target patient based on the electrical stimulation dose quantification model using the individual multidimensional feature data.

[0078] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0079] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0080] 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 quantifying individualized electrical stimulation dose using MST, characterized in that, Includes the following steps: Acquire individual multidimensional physiological data of the target patient, and extract features from the individual multidimensional physiological data to obtain individual multidimensional feature data; Based on the obtained multidimensional physiological data of the samples and their corresponding clinically validated electrical stimulation dose parameters, an electrical stimulation dose quantification model was constructed. Based on the electrical stimulation dose quantification model, the individual's multidimensional feature data are predicted to obtain the electrical stimulation dose for the target patient.

2. The method for quantifying individualized electrical stimulation (MST) dose according to claim 1, characterized in that, The individual multidimensional physiological data includes electroencephalogram (EEG) information, electromyogram (EMG) data, patient physiological metabolic data, and patient physical indicator data. The process of extracting features from the individual multidimensional physiological data to obtain individual multidimensional feature data includes: Frequency band power features are extracted from the electroencephalogram information to obtain the first feature data; The electromyography data is subjected to activity intensity features, frequency features, and potential features to obtain the second feature data; The patient's physiological metabolic data were subjected to extraction of blood glucose metabolism characteristics, oxygen metabolism characteristics, and enzyme metabolism characteristics to obtain third characteristic data. The patient's physical index data were subjected to basic physical parameter features, detailed medical history features, and cardiovascular index features to obtain fourth feature data; The first feature data, the second feature data, the third feature data, and the fourth feature data are weighted and fused to obtain the individual multidimensional feature data.

3. The method for quantifying individualized electrical stimulation dose for MST according to claim 2, characterized in that, The second feature data is obtained by extracting activity intensity features, frequency features, and potential features from the electromyography data, including: Based on the root mean square value and integral electromyography value of the electromyography signal in the electromyography data, muscle activity intensity characteristic data are obtained. Power spectrum analysis is performed on the electromyography signals in the electromyography data to determine the main frequency components in the electromyography signals and obtain muscle frequency characteristic data. The electromyography (EMG) signals in the EMG data are decomposed and the MUAP parameters are extracted to obtain muscle action potential characteristic data. The muscle activity intensity characteristic data, the muscle frequency characteristic data, and the muscle action potential characteristic data are combined to obtain the second characteristic data.

4. The method for quantifying individualized electrical stimulation dose for MST according to claim 2, characterized in that, The process of extracting blood glucose metabolism characteristics, oxygen metabolism characteristics, and enzyme metabolism characteristics from the patient's physiological metabolic data yields a third characteristic data, including: Blood glucose data were extracted from the patient's physiological metabolic data to obtain blood glucose metabolism characteristic data; Oxygen metabolism indexes were measured in the blood data from the patient's physiological metabolic data to obtain oxygen metabolism characteristic data. Enzyme activity was measured in the patient's physiological metabolic data to obtain enzyme metabolic characteristic data; The blood glucose metabolism characteristic data, the oxygen metabolism characteristic data, and the enzyme metabolism characteristic data are combined to obtain the third characteristic data.

5. The method for quantifying individualized electrical stimulation dose for MST according to claim 1, characterized in that, The process of extracting basic physical parameter features, detailed medical history features, and cardiovascular indicator features from the patient's physical indicator data yields fourth feature data, including: The height, weight, and body surface area of ​​the patient's physical indicators are analyzed and extracted to obtain basic physical parameter feature data. The vascular function commands in the patient's physical index data are measured and extracted to obtain cardiovascular index feature data; Historical medical record data were extracted from the patient's physical indicator data to obtain detailed medical history feature data; The basic physical parameter feature data, the cardiovascular indicator feature data, and the detailed medical history feature data are combined to obtain the fourth feature data.

6. The method for quantifying individualized electrical stimulation dose for MST according to claim 1, characterized in that, The electrical stimulation dose quantification model adopts an attention-based neural network architecture and uses feature attribution to interpret the model's prediction results.

7. The method for quantifying individualized electrical stimulation dose using MST according to claim 1, characterized in that, The loss function of the electrical stimulation dose quantification model is: ； ； ； ； Where N represents the number of samples, Represents the parameters of the model. This indicates the actual electrical stimulation dose. M represents the electrical stimulation dose predicted by the model; M represents the number of patient physical indicator samples. This represents the j-th physiological indicator of the i-th sample. Let represent the mean of the physiological indicators of the i-th sample. This represents the mean of the predicted electrical stimulation dose; K represents the number of types of neurological diseases. This represents the number of samples for the k-th disease. This represents the sample index set for the k-th disease. This represents the weight coefficient for the k-th disease; , and Both represent hyperparameters that weigh the importance of the losses of each component.

8. The MST individualized electrical stimulation dosage quantification device, characterized in that, Applied to the MST individualized electrical stimulation dose quantification method as described in any one of claims 1 to 7; The MST individualized electrical stimulation dose quantification device includes: Acquisition unit: used to acquire individual multidimensional physiological data of the target patient, and to extract features from the individual multidimensional physiological data to obtain individual multidimensional feature data; Model building unit: used to build an electrical stimulation dose quantification model based on the acquired multidimensional physiological data of the samples and their corresponding clinically validated electrical stimulation dose parameters; Processing unit: used to predict the individual's multidimensional feature data based on the electrical stimulation dose quantification model to obtain the electrical stimulation dose for the target patient.

9. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the MST individualized electrical stimulation dose quantification method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the steps of the MST individualized electrical stimulation dose quantification method as described in any one of claims 1 to 7.

Images