A deep brain stimulation system based on adaptive adjustment

Abstract

The application discloses a deep brain nerve stimulation system based on adaptive adjustment, and relates to the technical field of deep brain nerve regulation, and comprises the following steps: collecting a local field potential signal of a target region in the deep brain of a target patient, extracting a beta-band power spectral density and a gamma-band phase synchronization index as a neural activity characteristic parameter; determining an individual baseline value and a preset threshold value based on historical data, outputting a stimulation adjustment trigger signal when the beta-band power spectral density exceeds the individual baseline value and the gamma-band phase synchronization index is lower than the preset threshold value; calculating an optimal stimulation parameter combination by using a gradient descent optimization algorithm and performing neural regulation in response to the trigger signal; and monitoring a signal change after regulation, calculating a relative change rate and updating the threshold value. Through a double-parameter joint judgment mechanism and a threshold value updating strategy, the application realizes individualized adaptive adjustment of stimulation parameters, and improves stimulation accuracy and treatment effect.

Description

Technical Field

[0001] This application relates to the field of deep brain neuromodulation technology, and in particular to a deep brain neurostimulation system based on adaptive regulation. Background Technology

[0002] Deep brain stimulation (DBS) is an important treatment for movement disorders such as Parkinson's disease and essential tremor. By implanting electrodes into specific deep brain nuclei and applying high-frequency electrical stimulation, it can effectively improve motor symptoms in target patients. With a deeper understanding of the pathophysiological mechanisms, studies have revealed abnormal electrophysiological characteristics in the basal ganglia circuits of target Parkinson's disease patients, including enhanced β-band (13-30Hz) oscillations and reduced γ-band (30-100Hz) synchronicity. Based on these findings, closed-loop DBS systems are evolving from traditional open-loop fixed-parameter stimulation to adaptive stimulation based on biomarker feedback, dynamically adjusting stimulation parameters by real-time monitoring of local field potential (LFP) signals.

[0003] However, existing adaptive DBS technology still faces several technical bottlenecks. First, most systems rely solely on a single biomarker (such as beta-band power) as a feedback signal, ignoring the complex interactions between multiple frequency bands under pathological conditions, leading to an incomplete and inaccurate assessment of the pathological state. Second, existing systems generally employ fixed-threshold triggering strategies, which cannot adapt to individual differences in target patients and dynamic changes during disease progression, easily resulting in overstimulation or understimulation. Furthermore, the optimization of stimulation parameters typically uses preset parameter combinations or simple linear adjustments, lacking an intelligent optimization mechanism based on real-time feedback, making truly precise control difficult. In addition, the lack of an effective learning mechanism prevents continuous optimization of the control strategy based on treatment effects, limiting the improvement of long-term treatment outcomes. Summary of the Invention

[0004] In view of the aforementioned problems, this application is hereby filed.

[0005] Therefore, this application provides an adaptive deep brain stimulation system that can solve the problems mentioned in the background art.

[0006] To solve the above-mentioned technical problems, this application provides the following technical solution:

[0007] In a first aspect, this application provides an adaptive deep brain stimulation system, comprising: acquiring local field potential signals of a target region in the deep brain of a target patient in real time via implanted electrodes, and extracting the β-band power spectral density and γ-band phase synchronization index of the local field potential signals as neural activity characteristic parameters; determining an individual baseline value based on historical data of the β-band power spectral density, determining a preset threshold based on the normal range of the γ-band phase synchronization index, and outputting a stimulation modulation trigger signal when the β-band power spectral density exceeds the individual baseline value and the γ-band phase synchronization index is lower than the preset threshold; responding to the stimulation modulation trigger signal, calculating an optimal combination of stimulation parameters in real time using a gradient descent optimization algorithm, and outputting the optimal combination of stimulation parameters to a deep brain stimulator to perform neural modulation; wherein the optimal combination of stimulation parameters includes frequency, pulse width, and amplitude; monitoring changes in the local field potential signal after performing the neural modulation, calculating the relative rate of change of the neural activity characteristic parameters before and after modulation, and updating the individual baseline value and the preset threshold according to the relative rate of change.

[0008] Preferably, the step of extracting the β-band power spectral density and γ-band phase synchronization index of the local field potential signal as neural activity characteristic parameters includes: preprocessing the local field potential signal; performing frequency domain analysis on the preprocessed local field potential signal to extract the power characteristics of the β-band and obtain the β-band power spectral density; and performing phase analysis on the preprocessed local field potential signal to extract the phase consistency characteristics of the γ-band and obtain the γ-band phase synchronization index.

[0009] Preferably, determining the individual baseline value based on historical data of the β-band power spectral density and determining the preset threshold based on the normal range of the γ-band phase synchronization index includes: collecting the β-band power spectral density during the symptom relief period of the target patient to form a first dataset; collecting the γ-band phase synchronization index during the symptom relief period of the target patient to form a second dataset, and using the data range during the symptom relief period as the normal range; if the distribution width of the first dataset is greater than the distribution width of the second dataset, then the high quantile of the first dataset is selected as the individual baseline value, and the low quantile of the second dataset is selected as the preset threshold; otherwise, the individual baseline value is determined according to the central tendency measure of the first dataset, and the preset threshold is determined according to the dispersion tendency measure of the second dataset.

[0010] Preferably, the step of calculating the optimal stimulus parameter combination in real time using the gradient descent optimization algorithm includes: constructing an objective function; the objective function includes a difference measure term between the β-band power spectral density and the individual baseline value and a difference measure term between the γ-band phase synchronization index and the preset threshold; setting a stimulus parameter search space and establishing a constraint relationship between the stimulus frequency, pulse width, and amplitude; if the gradient vector magnitude of the objective function is less than a preset convergence criterion, the current stimulus parameter combination is output as the optimal stimulus parameter combination; otherwise, gradient update continues.

[0011] Preferably, the calculation of the relative change rate of the neural activity characteristic parameters before and after modulation includes: obtaining a representative value of the β-band power spectral density within a preset time period before neural modulation as a first characteristic value; obtaining a representative value of the γ-band phase synchronization index within a preset time period before neural modulation as a second characteristic value; obtaining a representative value of the β-band power spectral density within a preset time period after neural modulation as a third characteristic value; obtaining a representative value of the γ-band phase synchronization index within a preset time period after neural modulation as a fourth characteristic value; calculating the relative change rate of the β-band power spectral density based on the first characteristic value and the third characteristic value; and calculating the relative change rate of the γ-band phase synchronization index based on the second characteristic value and the fourth characteristic value.

[0012] Preferably, updating the individual baseline value and the preset threshold based on the relative change rate includes: determining an update function based on a combination of the relative change rate of the β-band power spectral density and the relative change rate of the γ-band phase synchronization index; calculating a new individual baseline value based on the update function and the current individual baseline value; and calculating a new preset threshold based on the update function and the current preset threshold. Wherein, when the relative change rate of the β-band power spectral density is negative and the relative change rate of the γ-band phase synchronization index is positive, the update function makes the new individual baseline value greater than the current value and the new preset threshold less than the current value.

[0013] Preferably, in the difference measurement terms, the coefficient of the difference measurement term between the β-band power spectral density and the individual baseline value is greater than the coefficient of the difference measurement term between the γ-band phase synchronization index and the preset threshold.

[0014] Secondly, this application also provides an adaptive deep brain stimulation system, comprising: a signal acquisition module, used to acquire local field potential signals of a target region in the deep brain of a target patient in real time through implanted electrodes, and extract the β-band power spectral density and γ-band phase synchronization index of the local field potential signals as neural activity characteristic parameters; a threshold judgment module, used to determine an individual baseline value based on historical data of the β-band power spectral density, determine a preset threshold based on the normal range of the γ-band phase synchronization index, and output a stimulation modulation trigger signal when the β-band power spectral density exceeds the individual baseline value and the γ-band phase synchronization index is lower than the preset threshold; a parameter optimization module, used to respond to the stimulation modulation trigger signal, calculate the optimal combination of stimulation parameters in real time through a gradient descent optimization algorithm, and output the optimal combination of stimulation parameters to a deep brain stimulator to perform neural modulation; and a feedback update module, used to monitor the changes in the local field potential signal after performing the neural modulation, calculate the relative rate of change of the neural activity characteristic parameters before and after modulation, and update the individual baseline value and the preset threshold according to the relative rate of change.

[0015] Thirdly, this application also provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to perform the following steps: real-time acquisition of local field potential signals in a target region of the deep brain of a target patient via implanted electrodes, and extraction of the β-band power spectral density and γ-band phase synchronization index of the local field potential signals as neural activity characteristic parameters; determination of an individual baseline value based on historical data of the β-band power spectral density, determination of a preset threshold based on the normal range of the γ-band phase synchronization index, and output of a stimulation modulation trigger signal when the β-band power spectral density exceeds the individual baseline value and the γ-band phase synchronization index is lower than the preset threshold; response to the stimulation modulation trigger signal, real-time calculation of the optimal stimulation parameter combination using a gradient descent optimization algorithm, and output of the optimal stimulation parameter combination to a deep brain stimulator for neural modulation; wherein the optimal stimulation parameter combination includes frequency, pulse width, and amplitude; monitoring of changes in the local field potential signal after neural modulation, calculation of the relative rate of change of the neural activity characteristic parameters before and after modulation, and updating the individual baseline value and the preset threshold according to the relative rate of change.

[0016] Fourthly, this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, performs the following steps: real-time acquisition of local field potential signals in a target region of the deep brain of a target patient via an implanted electrode, and extraction of the β-band power spectral density and γ-band phase synchronization index of the local field potential signal as neural activity characteristic parameters; determination of an individual baseline value based on historical data of the β-band power spectral density, determination of a preset threshold based on the normal range of the γ-band phase synchronization index, and output of a stimulation modulation trigger signal when the β-band power spectral density exceeds the individual baseline value and the γ-band phase synchronization index is lower than the preset threshold; responding to the stimulation modulation trigger signal, real-time calculation of the optimal stimulation parameter combination using a gradient descent optimization algorithm, and outputting the optimal stimulation parameter combination to a deep brain stimulator to perform neural modulation; wherein the optimal stimulation parameter combination includes frequency, pulse width, and amplitude; monitoring changes in the local field potential signal after performing the neural modulation, calculating the relative rate of change of the neural activity characteristic parameters before and after modulation, and updating the individual baseline value and the preset threshold according to the relative rate of change.

[0017] Implementing this application offers the following advantages: This application provides an adaptive deep brain stimulation system. By constructing a dual-parameter joint judgment mechanism based on β-band power spectral density and γ-band phase synchronization index, it overcomes the limitations of existing single-indicator judgments and can more comprehensively and accurately reflect the pathological state. Simultaneously, this application employs a threshold update strategy based on treatment effect feedback, enabling the judgment threshold to adaptively adjust according to the target patient's state and disease progression, solving the technical problem that fixed thresholds cannot adapt to individual differences. This application uses a gradient descent optimization algorithm to achieve real-time optimization of stimulation parameters, which, compared to existing preset parameter or linear adjustment methods, can find the optimal solution in a multi-dimensional parameter space. The entire closed-loop control system achieves a transformation from passive response to active adaptation through continuous learning and optimization. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is an overall flowchart of a deep brain stimulation system based on adaptive adjustment, which is the subject of this application.

[0020] Figure 2This is a schematic diagram of the application environment of an adaptive adjustment-based deep brain stimulation system involved in this application;

[0021] Figure 3 This is a schematic diagram of the overall structure of a deep brain stimulation system based on adaptive adjustment, which is involved in this application.

[0022] Figure 4 This is a computer device diagram of a deep brain stimulation system based on adaptive adjustment, which relates to this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0024] Deep brain stimulation (DBS) technology has broad clinical application value in the treatment of neurological diseases. As an important means of treating movement disorders such as Parkinson's disease, essential tremor, and dystonia, DBS can effectively improve motor symptoms and enhance the quality of life of target patients by applying electrical stimulation to specific brain nuclei. With the development of implantable medical device technology and a deeper understanding of the pathophysiological mechanisms of diseases, DBS systems are evolving from traditional open-loop stimulation to closed-loop adaptive stimulation.

[0025] In related technologies, deep brain stimulation systems primarily employ a fixed-parameter open-loop stimulation mode. Doctors manually adjust the stimulation parameters post-surgery based on the target patient's symptoms, maintaining a fixed combination of stimulation frequency, pulse width, and amplitude for long-term use. However, the pathological state of the target patient is dynamic, with symptom severity fluctuating throughout the day, and disease progression causing changes in optimal stimulation parameters. Fixed-parameter stimulation cannot adapt to these dynamic changes, easily leading to insufficient or excessive stimulation, affecting treatment efficacy and increasing the risk of side effects.

[0026] To address the problem that stimulation parameters in related technologies cannot be dynamically adjusted according to the real-time neural state of the target patient, an adaptive deep brain stimulation system is proposed. This system acquires local field potential signals in real time, extracts the β-band power spectral density and γ-band phase synchronization index as neural activity characteristic parameters, determines the stimulation triggering timing based on a dual-parameter joint judgment mechanism, calculates the optimal combination of stimulation parameters through a gradient descent optimization algorithm, and dynamically updates the judgment threshold based on treatment effect feedback, thereby achieving individualized adaptive adjustment of stimulation parameters.

[0027] The adaptive deep brain stimulation system provided in this application embodiment can be applied to, for example... Figure 2The clinical application environment is shown. In clinical application, the target patient 100 undergoes surgery to implant a deep brain stimulation system, including a stimulation electrode 102 implanted in the brain, a pulse generator 104 implanted subcutaneously, and an extension wire 106 connecting the two. An external programmer 108 interacts with the pulse generator 104 wirelessly, and medical staff can monitor the target patient's neural signals and adjust the stimulation parameters through the programmer 108.

[0028] In actual treatment, the stimulating electrode 102 is implanted into the target nucleus deep within the patient's brain (such as the subthalamic nucleus or the medial part of the globus pallidus) to continuously acquire local field potential signals in this region. The processing unit built into the pulse generator 104 analyzes the acquired signals in real time, extracting electrophysiological characteristics in the β band (13-30Hz) and γ band (30-100Hz). When a pathological neural oscillation pattern is detected, the stimulation parameters are automatically adjusted and electrical stimulation is applied to inhibit abnormal neural activity. During treatment, changes in the target patient's neural activity are continuously monitored, and the stimulation strategy is dynamically optimized based on the treatment effect to achieve individualized adaptive treatment.

[0029] The stimulation electrode 102 typically uses a four- or eight-contact electrode with a diameter of 1.27 mm, and each contact can be independently recorded or stimulated. The pulse generator 104 is generally implanted in a subcutaneous pouch below the clavicle, and has a built-in rechargeable battery that supports wireless charging. The programmer 108 can be a dedicated medical device or a tablet computer with dedicated software installed.

[0030] In one exemplary embodiment, such as Figure 1 As shown, an adaptive deep brain stimulation system is provided, which can be applied to... Figure 2 The following steps, 202 to 208, are used as an example to illustrate the clinical application environment.

[0031] Step 202: Real-time acquisition of local field potential signals in the target area of ​​the deep brain of the target patient using implanted electrodes, and extraction of the β-band power spectral density and γ-band phase synchronization index of the local field potential signals as neural activity characteristic parameters.

[0032] Local field potential (LFP) signals are the sum of postsynaptic potentials and electrical activity of a neuronal population, reflecting the synchronized activity state of the neuronal population. Implantable electrodes are typically implanted in the subthalamic nucleus (STN) or the medial part of the globus pallidus (GPi), which are the main therapeutic targets for movement disorders such as Parkinson's disease. The raw signals acquired by the electrodes contain a broadband component from 0.1 Hz to 500 Hz, requiring signal processing to extract features of specific frequency bands. The β band is defined as 13-30 Hz; in target patients with Parkinson's disease, the power in this band is abnormally enhanced and associated with symptoms such as bradykinesia and rigidity. The γ band is defined as 30-100 Hz; the phase synchronization of this band reflects the functional connectivity state of the neural network, and decreased synchronization is associated with motor dysfunction.

[0033] Specifically, the extraction of the β-band power spectral density and γ-band phase synchronization index of the local field potential signal as neural activity characteristic parameters includes:

[0034] First, the local field potential signal is preprocessed. Preprocessing refers to the initial processing of the raw acquired signal to remove various noises and interferences, and extract the true neural signals. Specifically, this includes: removing the DC offset of the signal, i.e., subtracting the average value of the signal to make it fluctuate around zero; removing 50Hz or 60Hz power frequency interference and its harmonics, which originate from the power supply and surrounding electrical equipment; and removing motion artifacts and other physiological noise, such as electrocardiogram (ECG) and electromyography (EMG) interference. Preprocessing is typically achieved using digital filtering techniques, such as using a bandpass filter with a passband range of 1-200Hz, which preserves the β and γ frequency bands of interest while filtering out low-frequency drift and high-frequency noise.

[0035] Secondly, frequency domain analysis is performed on the preprocessed local field potential signal to extract the power characteristics of the β-band and obtain the power spectral density of the β-band. Frequency domain analysis converts the time-domain local field potential signal to the frequency domain to analyze the power distribution of the signal at different frequencies. Specifically, the Fast Fourier Transform (FFT) algorithm is used to convert the signal within a time window (e.g., 1 second) into a spectrum. To improve spectral resolution and reduce spectral leakage, the signal is windowed before performing the FFT, such as using a Hamming window. After calculating the power spectrum, the power values ​​in the 13-30Hz range are extracted. The total power can be obtained by integrating within this frequency band, or the average power spectral density of this frequency band can be calculated. The unit of power spectral density is usually μV² / Hz, reflecting the signal power per unit frequency.

[0036] Next, phase analysis is performed on the preprocessed local field potential signal to extract the phase consistency characteristics of the γ-band and obtain the γ-band phase synchronization index. Phase analysis is used to evaluate the phase characteristics and synchronization of neural signals. In specific implementation, the signal is first narrowband filtered to extract the γ-band components from 30-100Hz. Then, Hilbert transform is used to obtain the analytical representation of the signal, from which instantaneous phase information is extracted. Hilbert transform constructs orthogonal components of the original signal to form a complex signal, the amplitude of which is the instantaneous phase. For multi-channel recordings or signals from different time periods in the same channel, the distribution of phase differences is calculated. The phase synchronization index is quantified by calculating the concentration of the phase difference distribution, for example, using the mean vector length or phase locking value, with a value ranging from 0 to 1; the larger the value, the stronger the phase synchronization.

[0037] Through the above processing, two key neural activity characteristic parameters were obtained in real time: beta-band power spectral density and gamma-band phase synchronization index. These two parameters characterize pathological neural activity from different dimensions: beta-band power spectral density reflects the intensity of pathological oscillations and is a major electrophysiological marker of Parkinson's disease; gamma-band phase synchronization index reflects the functional connectivity and information transmission efficiency of neural networks. Using these two parameters in combination allows for more accurate identification of pathological states requiring intervention.

[0038] In one alternative implementation, preprocessing employs an adaptive filtering method that uses the Least Mean Square (LMS) algorithm to dynamically track and eliminate stimulus artifacts. When stimulation and recording occur simultaneously, the adaptive filter estimates the artifact template in real time based on the temporal information of the stimulus pulses and subtracts it from the recorded signal.

[0039] In one alternative implementation, the frequency domain analysis employs the multitaper method, which uses multiple orthogonal window functions to calculate the power spectrum separately and then averages it, thereby reducing the estimation variance while maintaining frequency resolution.

[0040] In one alternative implementation, the phase analysis employs a wavelet phase coherence method, which uses continuous wavelet transform to simultaneously obtain phase information in the time and frequency domains, making it suitable for analyzing the instantaneous phase synchronization characteristics of non-stationary signals.

[0041] Step 204: Determine the individual baseline value based on historical data of β-band power spectral density, and determine the preset threshold based on the normal range of γ-band phase synchronization index. When the β-band power spectral density exceeds the individual baseline value and the γ-band phase synchronization index is lower than the preset threshold, output the stimulation modulation trigger signal.

[0042] The individual baseline value is a β-band power spectral density assessment standard determined based on the target patient's historical data. Due to differences in deep brain nuclei structure, electrode implantation location, and disease severity among different target patients, their baseline β-band power spectral density levels also vary. The individual baseline value is determined by analyzing the target patient's β-band power spectral density data during symptom remission (non-pathological state) and represents the upper limit of power in the target patient's normal state. When the real-time monitored β-band power spectral density exceeds this baseline value, it indicates the presence of pathological β-oscillation enhancement, a typical electrophysiological manifestation of movement disorders such as Parkinson's disease.

[0043] The preset threshold is the lower limit of the judgment standard for the gamma-band phase synchronization index. The gamma-band phase synchronization index reflects the degree of phase consistency of a neuronal population in the 30-100Hz frequency band. Under normal circumstances, this index is maintained at a high level, indicating that the neural network has good functional connectivity and information transmission capabilities. The preset threshold is determined by analyzing the normal range of the gamma-band phase synchronization index during the symptom relief period of the target patient, representing the lower boundary of normal synchronicity. When the real-time monitored gamma-band phase synchronization index is lower than this threshold, it indicates an abnormally reduced synchronicity of the neural network, which is closely related to motor dysfunction.

[0044] Historical data consists of a sequence of neural activity characteristic parameters continuously collected and stored over a specific time period. This data must be collected during a period when the target patient's condition is clear and stable, ideally during symptom remission, when the patient's motor function is close to normal and neural activity is in a non-pathological state. The collection duration of historical data needs to be sufficiently long (e.g., 30 minutes) to cover physiological fluctuations and ensure the representativeness of the obtained statistical characteristics. Statistical analysis of historical data allows for the development of individualized judgment criteria, avoiding potential misjudgments that might arise from using group averages.

[0045] The stimulation adjustment trigger signal is the output control signal used to determine whether to initiate stimulation parameter adjustment. This signal is binary: when both conditions are met simultaneously—the β-band power spectral density exceeding the individual baseline value and the γ-band phase synchronization index below a preset threshold—the output value is 1, indicating a pathological state has been detected and stimulation intervention is required; otherwise, the output value is 0, indicating a normal state, maintaining the existing stimulation parameters or not stimulating. This dual-parameter joint judgment design improves the accuracy of pathological state identification, as using either parameter alone may lead to misjudgment. The trigger signal is directly transmitted to the parameter optimization module in step 206 to initiate the real-time calculation process of the stimulation parameters.

[0046] Specifically, determining the individual baseline value based on historical data of the β-band power spectral density and determining the preset threshold based on the normal range of the γ-band phase synchronization index includes:

[0047] First, the β-band power spectral density was collected during the symptom relief period of the target patient to form the first dataset. The symptom relief period refers to the time when the target patient's motor symptoms are well controlled, typically within 1-2 hours after medication administration. During this time, symptoms such as tremor and rigidity are significantly reduced, and motor function is close to normal. Data collected during this period reflects the baseline level of the target patient's non-pathological state. The data collection process lasted for a certain duration, for example, continuous recording for 30 minutes, with a sampling interval of 1 second, resulting in 1800 β-band power spectral density data points, constituting the first dataset.

[0048] For example, a target patient with Parkinson's disease enters a symptom remission phase one hour after taking levodopa in the morning. At this time, local field potential signals are continuously acquired by electrodes implanted in the subthalamic nucleus, and the β-band power spectral density sequence is obtained after processing in step 202. During the 30-minute recording, the β-band power spectral density fluctuates between 15-25 μV² / Hz, with an average value of approximately 20 μV² / Hz, forming a first dataset containing 1800 data points.

[0049] Secondly, the gamma-band phase synchronization index was collected during the symptom relief period of the target patient to form a second dataset, and the data range during the symptom relief period was taken as the normal range. This was done simultaneously with the collection of the first dataset, recording the gamma-band phase synchronization index within the same time window. The second dataset also contains 1800 data points, reflecting the normal changes in neural network synchronization in the target patient under good conditions.

[0050] For example, during the same recording process described above, the gamma-band phase synchronization index was extracted synchronously. The gamma-band phase synchronization index of this target patient remained between 0.6 and 0.8 during the symptom remission period, with an average of approximately 0.7, indicating good functional connectivity of the neural network. This range of 0.6-0.8 is defined as the normal range for the gamma-band phase synchronization index of this target patient.

[0051] Next, determine whether the distribution width of the first dataset is greater than that of the second dataset. Distribution width reflects the dispersion of the data and can be measured by statistics such as standard deviation, interquartile range, or coefficient of variation. For example, calculate the standard deviation σ1 of the first dataset and the standard deviation σ2 of the second dataset. If σ1 > σ2, it indicates that the power spectral density in the β band has greater variability.

[0052] For example, the standard deviation of the first dataset (β-band power spectral density) is calculated to be 3.5 μV² / Hz, and the standard deviation of the second dataset (γ-band phase synchronization index) is 0.08. Since the units are different, the coefficients of variation need to be calculated for comparison: the coefficient of variation for the first dataset is 3.5 / 20 = 0.175, and the coefficient of variation for the second dataset is 0.08 / 0.7 = 0.114. Therefore, the distribution width of the first dataset is greater than that of the second dataset.

[0053] If the distribution width of the first dataset is greater than that of the second dataset, then the high quantile of the first dataset is selected as the individual baseline value, and the low quantile of the second dataset is selected as the preset threshold. A high quantile refers to a higher percentile in the data distribution, such as the 75th percentile (Q3) or the 80th percentile. A low quantile refers to a lower percentile in the data distribution, such as the 25th percentile (Q1) or the 20th percentile.

[0054] For example, due to the wide distribution width of the first dataset, its 75th percentile is selected as the individual baseline value. After sorting the 1800 β-band power spectral density values, the value of the 1350th data point is 23 μV² / Hz, which is the individual baseline value for this target patient. For the second dataset, the 25th percentile is selected as the preset threshold, and the value of the 450th data point after sorting is 0.65, which is the preset threshold for the γ-band phase synchronization index.

[0055] Otherwise, the individual baseline value is determined based on the central tendency measure of the first dataset, and the preset threshold is determined based on the discrete tendency measure of the second dataset. When the distribution width of the first dataset is less than or equal to that of the second dataset, it indicates that the power spectral density of the β-band is relatively stable, and a central tendency measure, such as the mean or median, can be used as the baseline value. Meanwhile, since the phase synchronization index of the γ-band has a large variability, its discrete tendency needs to be considered.

[0056] After determining the individual baseline value and the preset threshold, during real-time monitoring, when the power spectral density of the β band exceeds the individual baseline value and the phase synchronization index of the γ band is lower than the preset threshold, a stimulus modulation trigger signal is output. This dual-parameter joint judgment mechanism is more reliable than single-parameter judgment and reduces the possibility of false triggering.

[0057] For example, during the target patient's daily activities, real-time monitoring shows that at a certain moment the β-band power spectral density reaches 26 μV² / Hz (exceeding the baseline value of 23 μV² / Hz), while the γ-band phase synchronization index drops to 0.60 (below the threshold of 0.65). This indicates that the target patient has entered a pathological state, and a stimulation modulation trigger signal with a value of 1 is output to initiate the subsequent stimulation parameter optimization process.

[0058] In one optional implementation, the distribution width is measured using the interquartile range (IQR), which is the difference between the third quartile and the first quartile. Specifically, the first and second datasets are sorted by numerical value, and the third quartile (Q3) at the 75th percentile and the first quartile (Q1) at the 25th percentile are found. The IQR is then calculated as IQR = Q3 - Q1. For example, if the first dataset has 1800 data points, the 450th data point after sorting corresponds to Q1, and the 1350th data point corresponds to Q3. The interquartile range reflects the distribution range of the middle 50% of the data and is not affected by extreme values. When the IQR of the first dataset is greater than the IQR of the second dataset, Q3 of the first dataset is selected as the individual baseline value, and Q1 of the second dataset is used as a preset threshold.

[0059] In one alternative implementation, the central tendency measure uses a truncated mean. The specific calculation process is as follows: first, the dataset is sorted by numerical value; then, the top 5% and bottom 5% of data points are removed. For example, for 1800 data points, the top 90 and bottom 90 are removed, and the arithmetic mean is calculated for the remaining 1620 data points. This method retains the advantages of the mean while avoiding the interference of extreme values. The truncation ratio can be adjusted according to the data characteristics, for example, using a 10% truncation or a fixed truncation.

[0060] In one optional implementation, a duration requirement is added when determining whether to output a trigger signal. Specifically, a timer is set up. When the power spectral density of the β band exceeds the individual baseline value and the phase synchronization index of the γ band is first detected to be below a preset threshold, the counter starts counting. At each subsequent sampling moment, it continues to check whether these two conditions are simultaneously met. If the conditions are consistently met for 5 consecutive seconds, a trigger signal is output; if either condition is not met, the counter is reset and restarted. This mechanism can effectively filter out false triggers caused by transient interference or physiological fluctuations.

[0061] Preferably, step 204 improves the accuracy and reliability of pathological state identification by establishing individualized judgment thresholds and a dual-parameter joint judgment mechanism. Compared with traditional fixed threshold or population average methods, individual baseline values ​​and preset thresholds are determined based on the patient's historical data during symptom relief, fully considering individual differences; the threshold determination strategy is adaptively selected according to data distribution characteristics, making the threshold setting more scientific and reasonable; the dual-parameter joint judgment requires both the β-band power spectral density and the γ-band phase synchronization index to meet the triggering conditions simultaneously, effectively reducing false triggering that may be caused by a single indicator; this joint judgment only activates stimulation modulation in pathological states that truly require intervention, avoiding unnecessary stimulation, prolonging battery life, and reducing the risk of side effects.

[0062] Step 206: In response to the stimulus modulation trigger signal, the optimal combination of stimulus parameters is calculated in real time using a gradient descent optimization algorithm, and then output to the deep brain stimulator to perform neuromodulation. The optimal combination of stimulus parameters includes frequency, pulse width, and amplitude.

[0063] The gradient descent optimization algorithm is an iterative optimization method that calculates the partial derivatives (gradients) of the objective function with respect to each parameter, updates the parameters along the negative direction of the gradient, gradually decreasing the objective function value until the optimal solution is found. The optimal combination of stimulation parameters refers to the stimulation parameter settings that most effectively improve the patient's neural activity characteristics. Frequency refers to the repetition frequency of the stimulation pulse, typically in the range of 60-185Hz; pulse width refers to the duration of a single stimulation pulse, typically in the range of 60-450μs; amplitude refers to the intensity of the stimulation current or voltage, typically in the range of 0-10mA or 0-10V. These three parameters together determine the energy and pattern of the stimulation, significantly impacting the therapeutic effect.

[0064] For example, when the stimulus modulation trigger signal with an output value of 1 is received in step 204, the parameter optimization process is immediately initiated. The goal of the optimization is to find a set of stimulus parameters that reduce the patient's β-band power spectral density below the individual baseline value while increasing the γ-band phase synchronization index above a preset threshold, thereby restoring normal neural activity patterns.

[0065] Specifically, the step of calculating the optimal combination of stimulus parameters in real time using the gradient descent optimization algorithm includes:

[0066] First, an objective function is constructed. This objective function includes a measure of the difference between the β-band power spectral density and the individual baseline value, and a measure of the difference between the γ-band phase synchronization index and the preset threshold. The objective function is the core of the optimization algorithm, used to quantify the deviation of neural activity characteristic parameters from the desired state under the current stimulus parameters. The difference measure reflects the gap between the measured value and the target value; commonly used measures include the difference of squares and the absolute difference. The objective function is designed as follows:

[0067] ;

[0068] in, The objective function value represents the degree to which the current neural state deviates from the ideal state; This represents the currently measured power spectral density in the β band. The individual baseline value determined in step 204; This represents the currently measured phase synchronization index for the γ-band. The preset threshold determined in step 204; and These are two difference measurement functions; and These are the corresponding weight coefficients. This two-parameter objective function design allows the optimization algorithm to simultaneously consider the suppression of pathological β oscillations and the restoration of γ band synchronization, thus improving the patient's neurological activity more comprehensively compared to single-index optimization.

[0069] For example, a patient's individual baseline value 23μV² / Hz, preset threshold It is 0.65. Under the current stimulation parameters, the measured β-band power spectral density is... The phase synchronization index for the γ band is 28 μV² / Hz. The value is 0.55. The squared difference is used as the measure of difference. , Then the difference measure term for the β band is The difference metric for the γ band is Assuming weighting coefficients , Then the objective function value .

[0070] Secondly, a search space for stimulation parameters is defined, and constraints are established between stimulation frequency, pulse width, and amplitude. The search space defines the range for parameter optimization, and safety and effectiveness must be considered. The frequency search space is typically set between 60-185Hz, as below 60Hz may be ineffective, and above 185Hz may cause side effects. The pulse width search space is typically set between 60-450μs; too short a pulse width cannot effectively activate neurons, while too long a pulse width increases energy consumption and the risk of tissue damage. The amplitude search space needs to be individually set, typically within the range below the sensory threshold and above the motor threshold. Constraints refer to the mutual constraints between parameters, establishing energy constraints:

[0071] ;

[0072] in, As a stimulus energy indicator; Stimulation frequency (unit: Hz); Pulse width (unit: μs); Stimulus amplitude (unit: mA); This is a safe energy limit. This energy constraint ensures that excessive stimulation energy is not generated during the optimization process, avoiding tissue overheating or electrode damage, and guaranteeing the safety of the treatment.

[0073] For example, for the above patients, the frequency search space is set as follows: Hz, pulse width search space is μs, amplitude search space is mA. Establish energy constraints: To ensure the stimulation energy is within a safe range, the initial parameters were set as follows: frequency 130Hz, pulse width 90μs, and amplitude 2.5mA.

[0074] Then, gradient descent iterative optimization is performed. In each iteration, the partial derivatives of the objective function with respect to the three stimulus parameters are calculated, forming the gradient vector. The gradient vector indicates the direction of the fastest growth of the objective function; therefore, updating the parameters along the negative direction of the gradient can reduce the value of the objective function. The parameter update formula is:

[0075] ;

[0076] in, The stimulus parameter vector contains three components: frequency, pulse width, and amplitude; subscripts... Indicates the first The next iteration; The learning rate controls the step size for parameter updates; For the objective function in The gradient vector at that point, i.e. Through this iterative update mechanism, the stimulation parameters are gradually adjusted in a direction that can improve the characteristics of neural activity, thus achieving adaptive optimization.

[0077] If the magnitude of the gradient vector of the objective function is less than the preset convergence criterion, then the current combination of stimulus parameters is output as the optimal combination of stimulus parameters; otherwise, gradient update continues. The magnitude of the gradient vector is defined as:

[0078] ;

[0079] when It is considered to converge at this time, where This serves as a pre-defined convergence criterion. The magnitude reflects the rate of change of the objective function at the current point. When the magnitude is sufficiently small, it indicates that the optimum has been approached, and further iteration will not bring significant improvement.

[0080] For example, after 15 iterations, the magnitude of the gradient vector decreased from the initial 5.2 to 0.008, which is less than the convergence criterion. The parameters at this point are: frequency 145Hz, pulse width 85μs, amplitude 2.8mA. Under these parameters, the predicted β-band power spectral density will decrease to 22μV² / Hz, and the γ-band phase synchronization index will increase to 0.66, both reaching the target range. This set of parameters is output as the optimal combination of stimulation parameters.

[0081] In the difference metric, the coefficient of the difference metric between the β-band power spectral density and the individual baseline value is greater than the coefficient of the difference metric between the γ-band phase synchronization index and the preset threshold, i.e. This weighting reflects clinical priorities: abnormally increased β-band power is the most important electrophysiological marker of Parkinson's disease and is directly related to motor symptoms, therefore it is given higher weight in the optimization. The larger weighting coefficients cause the optimization algorithm to prioritize reducing β-band power while also considering the restoration of γ-band synchronicity.

[0082] For example, setting , This indicates that the importance of improving the β band is 2.33 times that of the γ band. This weighting means that during the optimization process, even if the improvement in the γ band is not significant, as long as the β band is effectively suppressed, the objective function value will still decrease, and the algorithm tends to choose this parameter combination.

[0083] Finally, the calculated optimal combination of stimulation parameters is output to the deep brain stimulator. Upon receiving the parameters, the stimulator generates a corresponding stimulation pulse sequence, which is applied to the target nucleus via implanted electrodes. The stimulation signal is a biphasic rectangular pulse, initially negative then positive, ensuring charge balance and preventing electrode corrosion and tissue damage.

[0084] Preferably, step 206 achieves real-time intelligent optimization of stimulation parameters using the gradient descent optimization algorithm. The constructed two-parameter objective function simultaneously considers the suppression of β-band oscillations and the restoration of γ-band synchronization, making the optimization objective more comprehensive. The gradient descent algorithm can automatically search for the optimal combination in the three-dimensional parameter space (frequency, pulse width, amplitude), avoiding the subjectivity and limitations of manual parameter tuning. By setting energy constraints and charge density limits, the optimization results are ensured to be within a safe range, preventing tissue damage. The differentiated setting of weight coefficients (β-band weight is greater than γ-band weight) reflects the priority of clinical treatment, making the optimization process more in line with actual treatment needs. The entire optimization process can be completed within seconds, achieving true real-time adaptive adjustment.

[0085] Step 208: Monitor the changes in local field potential signals after performing neural modulation, calculate the relative rate of change of neural activity characteristic parameters before and after modulation, and update the individual baseline value and preset threshold based on the relative rate of change.

[0086] The relative rate of change is a quantitative indicator of the stimulation effect, reflecting the degree of change in neural activity characteristic parameters before and after stimulation. For the β-band power spectral density, a negative relative rate of change indicates a decrease in power, i.e., suppression of pathological oscillations; for the γ-band phase synchronization index, a positive relative rate of change indicates enhanced synchronization, i.e., improved neural network function. By monitoring the relative rates of change of these two parameters, the therapeutic effect of stimulation can be objectively assessed. Updating individual baseline values ​​and preset thresholds is a key mechanism for achieving adaptation, enabling the judgment criteria to be dynamically adjusted according to changes in patient condition and disease progression.

[0087] Specifically, the calculation of the relative rate of change of the neural activity characteristic parameters before and after modulation includes:

[0088] First, a representative value of the β-band power spectral density within a preset time period before neural modulation is obtained as the first characteristic value, and a representative value of the γ-band phase synchronization index within the same preset time period is obtained as the second characteristic value. The preset time period refers to a time window used to assess the baseline state; it needs to be long enough to obtain stable statistical characteristics, but not too long to avoid including state changes. The representative value refers to a statistical quantity that can characterize the typical level of the parameter within that time period, such as the mean, median, or mode. The first characteristic value reflects the pathological oscillation level of the β-band before stimulation, and the second characteristic value reflects the synchronization state of the γ-band before stimulation.

[0089] For example, the preset time period is set to 5 seconds before the stimulus begins. Within these 5 seconds, the power spectral density of the β band is calculated at 1-second intervals, resulting in 5 data points: [26, 27, 26.5, 27.5, 26] μV² / Hz. The average value of 26.6 μV² / Hz is taken as the first characteristic value. Simultaneously, the phase synchronization index of the γ band is calculated, resulting in [0.58, 0.60, 0.59, 0.57, 0.61], and the average value of 0.59 is taken as the second characteristic value.

[0090] Secondly, a representative value of the power spectral density of the β band within a preset time period after neural modulation is obtained as the third characteristic value, and a representative value of the phase synchronization index of the γ band within the preset time period after neural modulation is obtained as the fourth characteristic value. The selection of the preset time period after neural modulation needs to consider the establishment time of the stimulus effect. Typically, a certain time (e.g., 10 seconds) is waited after the stimulus begins to allow neural activity to reach a new steady state before selecting a time window for evaluation. The third and fourth characteristic values ​​reflect the characteristic parameter levels of the two frequency bands after stimulation, respectively.

[0091] For example, a 5-second time window is selected for evaluation 30 seconds after the stimulus begins. At this time, the β-band power spectral density sequence is [21, 20.5, 21.5, 20, 21] μV² / Hz, with an average value of 20.8 μV² / Hz as the third characteristic value. The γ-band phase synchronization index sequence is [0.68, 0.70, 0.69, 0.67, 0.71], with an average value of 0.69 as the fourth characteristic value.

[0092] Then, the relative rate of change of the power spectral density in the β band is calculated based on the first and third eigenvalues, and the relative rate of change of the phase synchronization index in the γ band is calculated based on the second and fourth eigenvalues. The formula for calculating the relative rate of change is:

[0093] ;

[0094] ;

[0095] in, The relative rate of change of the power spectral density in the β band; This represents the representative value of the β-band power spectral density after stimulation; This represents the power spectral density of the β-band before stimulation; The first characteristic value (representative value of the β band before stimulation); This is the third characteristic value (representative value of the β band after stimulation); This represents the relative rate of change of the phase synchronization index in the γ-band. This is the second characteristic value (the representative value of the γ band before stimulation); This is the fourth characteristic value (representative value of the γ band after stimulation); This represents the value of the phase synchronization index in the γ-band after stimulation. This represents the pre-stimulation gamma-band phase synchronization index. The sign and magnitude of the relative rate of change directly reflect the effectiveness of the stimulus.

[0096] For example, the calculation is based on the above values: This indicates that the power spectral density in the β band decreased by 21.8%. This indicates that the phase synchronization index in the γ band increased by 16.9%. The negative relative change rate in the β band and the positive relative change rate in the γ band indicate that the stimulus achieved a good effect.

[0097] Specifically, updating the individual baseline value and the preset threshold based on the relative rate of change includes:

[0098] First, an update function is determined based on the combination of the relative rate of change of the power spectral density in the β band and the relative rate of change of the phase synchronization index in the γ band. The update function is a mathematical relationship that maps the relative rates of change to a threshold adjustment amount; its form depends on the positive and negative combinations of the two relative rates of change. Different combinations reflect different treatment response patterns, requiring corresponding update strategies.

[0099] Secondly, a new individual baseline value is calculated based on the update function and the current individual baseline value. The update process can be represented as follows:

[0100] ;

[0101] in, The updated individual baseline value; This is the current individual baseline value; The adjustment amount is determined based on two relative rates of change. and The specific function form is determined based on different combinations.

[0102] Next, a new preset threshold is calculated based on the update function and the current preset threshold. The update process is as follows:

[0103] ;

[0104] in, The updated preset threshold; The current preset threshold; This is the adjustment amount determined based on the relative rate of change.

[0105] Specifically, when the relative rate of change of the β-band power spectral density is negative and the relative rate of change of the γ-band phase synchronization index is positive, the update function ensures that the new individual baseline value is greater than the current value and the new preset threshold is less than the current value. This indicates a good stimulation effect and that the patient is sensitive to the current stimulus. The trigger threshold can be appropriately increased, and the stimulation frequency reduced, thus conserving energy and extending battery life while maintaining therapeutic efficacy.

[0106] For example, when (Negative value) and A positive value indicates that the treatment is effective. The update function is set as follows: μV² / Hz, the new individual baseline value is 23.5 μV² / Hz; The new preset threshold is 0.639.

[0107] Through this feedback and update mechanism, this application can dynamically adjust the judgment threshold based on the actual effect of each stimulus, achieving true adaptive control. As treatment progresses, the threshold gradually converges to the level most suitable for the patient's current state, ensuring treatment effectiveness while avoiding overstimulation.

[0108] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0109] Based on the same inventive concept, this application also provides an adaptive adjustment-based deep brain stimulation system. The solution provided by this system is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the adaptive adjustment-based deep brain stimulation system provided below can be found in the above-described limitations of the adaptive adjustment-based deep brain stimulation system, and will not be repeated here.

[0110] In one exemplary embodiment, such as Figure 3 As shown, an adaptive deep brain stimulation system is provided, comprising a signal acquisition module, a threshold judgment module, a parameter optimization module, and a feedback update module, wherein:

[0111] The signal acquisition module acquires local field potential signals in the target area of ​​the deep brain of the target patient in real time through implanted electrodes, and extracts the β-band power spectral density and γ-band phase synchronization index of the local field potential signal as neural activity characteristic parameters.

[0112] The threshold judgment module determines the individual baseline value based on historical data of the β-band power spectral density and determines the preset threshold based on the normal range of the γ-band phase synchronization index. When the β-band power spectral density exceeds the individual baseline value and the γ-band phase synchronization index is lower than the preset threshold, it outputs a stimulus modulation trigger signal.

[0113] The parameter optimization module responds to the stimulus modulation trigger signal, calculates the optimal stimulus parameter combination in real time through the gradient descent optimization algorithm, and outputs the optimal stimulus parameter combination to the deep brain stimulator to perform neuromodulation.

[0114] The feedback update module monitors changes in local field potential signals after neural modulation, calculates the relative rate of change of neural activity characteristic parameters before and after modulation, and updates individual baseline values ​​and preset thresholds based on the relative rate of change.

[0115] The modules in the aforementioned adaptive deep brain stimulation system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can invoke and execute the corresponding operations of each module.

[0116] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 4As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements an adaptive deep brain stimulation system. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0117] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0118] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0119] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0120] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0121] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0122] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0123] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0124] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A deep brain stimulation system based on adaptive adjustment, characterized in that, include: The signal acquisition module is used to acquire local field potential signals in the target area of ​​the deep brain of the target patient in real time through implanted electrodes, and extract the β-band power spectral density and γ-band phase synchronization index of the local field potential signals as neural activity characteristic parameters. The threshold judgment module determines the individual baseline value based on the historical data of the β-band power spectral density and determines the preset threshold based on the normal range of the γ-band phase synchronization index. When the β-band power spectral density exceeds the individual baseline value and the γ-band phase synchronization index is lower than the preset threshold, it outputs a stimulation modulation trigger signal. The parameter optimization module, in response to the stimulus modulation trigger signal, calculates the optimal combination of stimulus parameters in real time using a gradient descent optimization algorithm, including: Construct an objective function; the objective function includes a difference measure term between the β-band power spectral density and the individual baseline value and a difference measure term between the γ-band phase synchronization index and the preset threshold; in the difference measure term, the coefficient of the difference measure term between the β-band power spectral density and the individual baseline value is greater than the coefficient of the difference measure term between the γ-band phase synchronization index and the preset threshold; Define the search space for stimulation parameters and establish the constraint relationship between stimulation frequency, pulse width, and amplitude; If the magnitude of the gradient vector of the objective function is less than the preset convergence criterion, then the current combination of stimulus parameters is output as the optimal combination of stimulus parameters; otherwise, gradient update continues. The optimal combination of stimulation parameters is output to a deep brain stimulator to perform neuromodulation; wherein, the optimal combination of stimulation parameters includes frequency, pulse width, and amplitude; The feedback update module is used to monitor the changes in the local field potential signal after the neural modulation is performed, and to calculate the relative rate of change of the neural activity characteristic parameters before and after modulation, including: The representative value of the power spectral density of the β band within a preset time period before neural modulation is obtained as the first feature value, and the representative value of the phase synchronization index of the γ band within a preset time period before neural modulation is obtained as the second feature value. The representative value of the power spectral density of the β band within a preset time period after neural modulation is obtained as the third feature value, and the representative value of the phase synchronization index of the γ band within a preset time period after neural modulation is obtained as the fourth feature value. The relative rate of change of the power spectral density of the β band is calculated based on the first characteristic value and the third characteristic value, and the relative rate of change of the phase synchronization index of the γ band is calculated based on the second characteristic value and the fourth characteristic value. Updating the individual baseline value and the preset threshold based on the relative rate of change includes: The update function is determined based on the combination of the relative rate of change of the power spectral density in the β band and the relative rate of change of the phase synchronization index in the γ band; Calculate the new individual baseline value based on the update function and the current individual baseline value; Calculate a new preset threshold based on the update function and the current preset threshold; Specifically, when the relative rate of change of the power spectral density of the β band is negative and the relative rate of change of the phase synchronization index of the γ band is positive, the update function makes the new individual baseline value greater than the current value and the new preset threshold less than the current value.

2. The deep brain stimulation system based on adaptive adjustment as described in claim 1, characterized in that: The extraction of the β-band power spectral density and γ-band phase synchronization index of the local field potential signal as neural activity characteristic parameters includes: The local field potential signal is preprocessed; Frequency domain analysis is performed on the preprocessed local field potential signal to extract the power characteristics of the β band and obtain the power spectral density of the β band. Phase analysis is performed on the preprocessed local field potential signal to extract the phase consistency characteristics of the γ band and obtain the phase synchronization index of the γ band.

3. The deep brain stimulation system based on adaptive adjustment as described in claim 2, characterized in that: The process of determining an individual baseline value based on historical data of the β-band power spectral density and determining a preset threshold based on the normal range of the γ-band phase synchronization index includes: The power spectral density of the β band was collected during the symptom relief period of the target patients to form the first dataset; During the symptom relief period of the target patient, the phase synchronization index of the γ band is collected to form a second dataset, and the data range during the symptom relief period is taken as the normal range. If the distribution width of the first dataset is greater than the distribution width of the second dataset, then the high quantile of the first dataset is selected as the individual baseline value, and the low quantile of the second dataset is selected as the preset threshold. Otherwise, the individual baseline value is determined based on the central tendency metric of the first dataset, and the preset threshold is determined based on the discrete tendency metric of the second dataset.

4. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the adaptive adjustment-based deep brain stimulation system as described in any one of claims 1 to 3.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the adaptive adjustment-based deep brain stimulation system as described in any one of claims 1 to 3.

Images