An electroencephalogram signal information processing method and system applied to neuromodulation

Abstract

This application provides a method and system for processing electroencephalogram (EEG) signals applied to neuromodulation. The method includes: based on standardized key patient information, determining the degree of difference between the key patient information and the information reflecting the inherent characteristics of the data source, and the degree of difference between the key patient information and the information represented by the established common disease pattern, respectively, according to pre-stored information reflecting the inherent characteristics of the data source and information represented by the common disease pattern; judging whether the key patient information is affected by the inherent characteristics of a specific data source and whether it deviates from the common disease pattern based on the determined degree of difference; and assessing the applicability of a preliminary neuromodulation plan generated for the patient based on the judgment results; triggering strategy adjustment when the applicability assessment results of the preliminary neuromodulation plan show insufficient applicability. This application can improve the personalization and effectiveness of neuromodulation therapy.

Description

Technical Field

[0001] Electroencephalogram (EEG) signals used in neuromodulation are a non-invasive brain activity observation technology. They offer advantages such as minimal harm to patients and ease of operation, and are widely used in the diagnosis of neurological diseases, fatigue and depression detection, and research on brain functional areas. In the field of modern neuromedicine, to improve the accuracy and personalization of neuromodulation therapy, an intelligent information processing system based on artificial intelligence and medical informatics has been designed and deployed. This system aims to provide optimized neuromodulation treatment plans for patients with neurological diseases by deeply integrating diverse medical data from patients. This application relates to the field of neuromodulation technology, specifically to a method and system for processing EEG signal information applied to neuromodulation. Background Technology

[0002] In the deployment and operation of intelligent information processing systems based on artificial intelligence and medical informatics, although participating institutions adhere to strict data standardization protocols during the data collection phase, the large-scale historical datasets collected are not entirely consistent due to subtle differences in equipment models, operating procedures, collection environments, and clinical assessment habits. These seemingly minor, non-obvious data differences accumulate in massive amounts of data, forming complex background noise or implicit biases. While the intelligent processing logic can learn general patterns from these differing data during the learning process, it inevitably internalizes these implicit biases into its decision-making logic. This means that the optimal treatment plan established by the intelligent processing logic may be derived from an average dataset with a specific bias, rather than being constructed from absolutely pure, idealized data. Summary of the Invention

[0003] This application discloses a method and system for processing electroencephalogram (EEG) signals applied to neuromodulation, in order to solve at least one of the above-mentioned technical problems in the prior art.

[0004] To achieve the above objectives, this application adopts the following technical solution:

[0005] In a first aspect, this application discloses a method for processing electroencephalogram (EEG) signals applied to neural modulation, comprising:

[0006] Before processing patient medical data, acquire and store the inherent characteristics of the data source, specifically:

[0007] Medical data is acquired from multiple sources, including EEG data. Based on the medical data, information reflecting the inherent characteristics of each data source is extracted and constructed. For EEG data, information on EEG fingerprint characteristics is obtained by analyzing the noise spectrum characteristics, baseline drift patterns, and frequency of occurrence of specific artifacts from different data sources.

[0008] By storing the characteristic information of brainwave fingerprints, information reflecting the inherent characteristics of the data source can be obtained;

[0009] Acquire patient medical data to be processed, extract key information from the patient medical data, and standardize the key information;

[0010] Based on standardized key patient information, and according to pre-stored information reflecting the inherent characteristics of the data source and information represented by established common disease patterns, the degree of difference between key patient information and information reflecting the inherent characteristics of the data source and the degree of difference between key patient information and information represented by common disease patterns are determined respectively.

[0011] Based on the determined degree of difference, determine whether the patient's key information is affected by the inherent characteristics of the specific data source and whether it deviates from the general disease pattern. Based on the judgment results, assess the applicability of the preliminary neuromodulation plan generated for the patient.

[0012] When the initial suitability assessment of the neuromodulation protocol indicates inadequacy, strategy adjustment is triggered. This adjustment includes issuing prompts and generating alternative treatment options.

[0013] Secondly, this application also discloses a brainwave signal information processing system for neural modulation, comprising:

[0014] The data source characteristic information acquisition and storage module is used to acquire and store the inherent characteristic information of the data source before processing patient medical data, specifically:

[0015] Medical data is obtained from multiple sources, including electroencephalogram (EEG) data.

[0016] Based on medical data, information reflecting the inherent characteristics of each data source is extracted and constructed. For EEG data, information on EEG fingerprint characteristics is obtained by analyzing the noise spectrum characteristics, baseline drift patterns, and frequency of occurrence of specific artifacts from different data sources.

[0017] The characteristic information of brainwave fingerprints is stored to obtain information that reflects the inherent characteristics of the data source;

[0018] The patient medical data processing module is used to acquire patient medical data to be processed, extract key information from the patient medical data, and standardize the key information.

[0019] The difference determination module is used to determine the degree of difference between the patient's key information and the information reflecting the inherent characteristics of the data source, as well as the degree of difference between the patient's key information and the information representing the common disease pattern, based on standardized key patient information, as well as pre-stored information reflecting the inherent characteristics of the data source and information representing the common disease pattern.

[0020] The protocol applicability assessment module is used to determine, based on the determined degree of difference, whether the patient's key information is affected by the inherent characteristics of a specific data source, and whether it deviates from the general disease pattern. Based on the assessment results, it evaluates the applicability of the preliminary neuromodulation protocol generated for the patient.

[0021] The strategy adjustment module is used to trigger strategy adjustment when the initial suitability assessment of the neuromodulation protocol shows insufficient suitability. The strategy adjustment includes issuing prompts and generating alternative treatment plans.

[0022] Compared with the prior art, this application has at least the following beneficial effects:

[0023] This application, by acquiring and storing inherent characteristic information of the data source before processing patient medical data, can effectively identify and quantify implicit biases or background noise that may be introduced by different data sources. After acquiring the patient medical data to be processed and extracting key information for standardization, this application can accurately determine the degree of difference between the patient's key information and the aforementioned biases and common disease patterns based on pre-stored inherent characteristic information of the data source and common disease patterns. Accordingly, it can determine whether the patient's key information is affected by the inherent characteristics of a specific data source and whether it deviates from common disease patterns.

[0024] By assessing the degree of difference, the applicability of the initial neuromodulation plan generated for the patient can be evaluated. When the evaluation results indicate insufficient applicability, strategy adjustments can be triggered in a timely manner. Attached Figure Description

[0025] Figure 1 A flowchart illustrating a brainwave signal information processing method for neural modulation provided in this application;

[0026] Figure 2 This is a schematic diagram of the structure of an electroencephalogram (EEG) signal information processing system for neural modulation, provided in this application. Detailed Implementation

[0027] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0028] like Figure 1 As shown in the embodiments of this application, a method for processing electroencephalogram (EEG) signals applied to neural modulation is proposed, comprising:

[0029] Before processing patient medical data, acquire and store the inherent characteristics of the data source, specifically:

[0030] Medical data is obtained from multiple sources, including electroencephalogram (EEG) data.

[0031] Based on medical data, information reflecting the inherent characteristics of each data source is extracted and constructed. For EEG data, information on EEG fingerprint characteristics is obtained by analyzing the noise spectrum characteristics, baseline drift patterns, and frequency of occurrence of specific artifacts from different data sources.

[0032] By storing the characteristic information of brainwave fingerprints, information reflecting the inherent characteristics of the data source can be obtained;

[0033] Acquire patient medical data to be processed, extract key information from the patient medical data, and standardize the key information;

[0034] Based on standardized key patient information, and according to pre-stored information reflecting the inherent characteristics of the data source and information represented by established common disease patterns, the degree of difference between key patient information and information reflecting the inherent characteristics of the data source and the degree of difference between key patient information and information represented by common disease patterns are determined respectively.

[0035] Based on the determined degree of difference, determine whether the patient's key information is affected by the inherent characteristics of the specific data source and whether it deviates from the general disease pattern. Based on the determination results, assess the applicability of the preliminary neuromodulation plan generated for the patient; and

[0036] When the initial suitability assessment of the neuromodulation protocol shows insufficient suitability, a strategy adjustment is triggered, which includes issuing a prompt message and generating an alternative treatment plan.

[0037] Neuromodulation refers to methods of treating neuropsychiatric disorders by regulating the activity of the nervous system through technical means, such as deep brain stimulation (DBS), transcranial magnetic stimulation (TMS), transcranial direct current stimulation, transcranial focused ultrasound, transcranial optical stimulation, vagus nerve stimulator (VNS), and spinal cord stimulation. Its core lies in altering the electrophysiological activity of the brain through external intervention, thereby improving the patient's clinical symptoms. Electroencephalogram (EEG) signal processing refers to a series of operations, including analysis, noise reduction, feature extraction, and pattern recognition, of EEG signals acquired through methods such as electroencephalography (EEG), to obtain useful information related to neural activity. The inherent characteristics of the data source refer to systematic features that, during the medical data acquisition process, are unrelated to the data content itself but affect the data's presentation due to factors such as equipment model, acquisition environment, operating procedures, and clinical assessment habits. For example, different models of EEG equipment may have different noise baselines or frequency response characteristics; clinical assessment scales from different hospitals may have subtle differences. This characteristic information is extracted and constructed to quantify the potential bias of the data source.

[0038] Medical data refers to data generated during a patient's diagnosis and treatment, including electroencephalogram (EEG) data (records of brain activity collected by EEG devices).

[0039] Key information refers to features extracted from raw medical data that are crucial for disease diagnosis and treatment planning, such as the power of specific frequency bands in electroencephalogram (EEG) signals.

[0040] Standardization refers to the process of unifying and standardizing extracted key information to eliminate non-physiological differences caused by different data sources or collection conditions, ensuring the comparability of data during comparison and analysis. This may include operations such as data normalization, detrending, and filtering.

[0041] The information represented by a common disease pattern refers to the typical characteristics or standard patterns of a certain disease established through the analysis and learning of a large amount of patient data. This can be a statistical model, a machine learning classifier, or a feature vector, representing the common manifestation of the disease in the population.

[0042] The degree of difference refers to the quantitative difference between the patient's key information and the information represented by the inherent characteristics of the data source or the general disease pattern. It is usually expressed by distance measures (such as Euclidean distance, cosine similarity) or statistical tests.

[0043] The preliminary neuromodulation plan refers to the neuromodulation treatment plan generated by the system for the patient for the first time based on the patient's current data and preset rules or models, including stimulation parameters, target points, etc.

[0044] Suitability assessment is the process of determining whether a preliminary neuromodulation protocol is suitable for the current patient, taking into account its potential effectiveness, safety, and individual fit.

[0045] Strategy adjustment refers to the corrective measures taken by the system when the initial neuromodulation protocol is not suitable. These measures include issuing warning messages to clinicians and generating new, more suitable alternative treatment options for the patient.

[0046] Specifically, when medical data is collected from different medical institutions or using different models of EEG devices, these data sources may have inherent and systematic differences. To obtain this inherent information, the system acquires medical data from multiple sources, including EEG data. For example, the system might retrieve EEG data from a batch of patients' databases from Hospital A and another batch from Hospital B's database. These data may exhibit subtle but systematic differences due to factors such as the brand and model of the acquisition equipment, electrode placement standards, and environmental electromagnetic interference levels.

[0047] For EEG data, the system can analyze noise spectrum characteristics, baseline drift patterns, and the frequency of specific artifacts from different data sources. These analytical results are integrated to form a quantitative characteristic information that represents the fingerprint of the data source. For example, a feature vector can be constructed, which includes the mean noise power within a specific frequency range, the standard deviation of baseline drift, and the incidence of a specific artifact. This information reflecting the inherent characteristics of the data source is then stored to form a database for subsequent querying and comparison. For example, this information can be stored in a structured database, with each record corresponding to a data source and containing its extracted inherent characteristic information.

[0048] When patient medical data is input into the system, the system acquires this data and extracts key information. For example, for a Parkinson's disease patient, key information might include the beta band power in the resting-state EEG signal, the latency and amplitude of motor-related cortical potentials (MRCP), and the volume and morphological features of the basal ganglia region in MRI images. This extracted key information is then standardized to eliminate non-physiological variations caused by different acquisition conditions or individual differences. For example, the power values ​​of the EEG signal can be Z-score standardized to make them comparable across different patients.

[0049] Subsequently, the Euclidean distance between the patient's EEG feature vector and the noise feature vector from a specific data source can be calculated to quantify whether the patient's data is affected by the inherent noise characteristics of that data source. Simultaneously, the distance between the patient's EEG feature vector and the feature vector of a common Parkinson's disease pattern can be calculated to assess the degree of deviation between the patient's physiological characteristics and the typical disease pattern. These differences can be calculated using various mathematical methods; for example, cosine similarity can be used to measure the similarity between feature vectors, or Mahalanobis distance can be used to consider the covariance between features.

[0050] If patient data shows a high similarity to the noise characteristics of a data source (e.g., the difference is less than a preset threshold), it may indicate that the patient data is contaminated by the inherent noise of that data source. If patient data differs significantly from common disease patterns, it indicates that the patient has high individual specificity. Based on these judgments, the system assesses the applicability of the initial neuromodulation protocol generated for the patient. For example, if the patient data is severely contaminated by noise, or if the patient's physiological characteristics deviate significantly from common disease patterns, then the applicability of the initial neuromodulation protocol generated based on the common pattern will be considered insufficient.

[0051] Strategy adjustments include issuing alerts and generating alternative treatment plans. For example, the system can alert clinicians that patient data may be affected by specific device noise, casting doubt on the applicability of the initial treatment plan, and simultaneously activate a module to generate one or more alternative treatment plans specific to that patient or data bias. These alternative treatment plans may consider adjusting the stimulation target, the range of stimulation parameters, or employing different neuromodulation techniques.

[0052] In some of the above embodiments, the steps of determining whether the patient's key information is affected by the inherent characteristics of a specific data source and whether it deviates from a common disease pattern based on the determined degree of difference, and assessing the applicability of the preliminary neuromodulation plan generated for the patient based on the determination results, include:

[0053] While collecting patients' EEG data, an environmental sensor array is activated to monitor the electromagnetic environment in the acquisition room in real time, capture the intensity, frequency distribution and instantaneous fluctuations of the electromagnetic spectrum, generate real-time environmental noise information of the current environment, and synchronize this real-time environmental noise information with the EEG data acquisition.

[0054] Interference components that highly match real-time environmental noise information are separated from the raw EEG data to obtain the denoised EEG signal;

[0055] If the statistical characteristics of the interference components are highly consistent with the real-time environmental noise information, and the characteristics of the interference components show a high degree of similarity with the inherent circuit noise characteristics recorded by the characteristic information of a certain historical data source in the system database, it is determined that the EEG data is affected by the inherent characteristics of a specific data source, and the deviation in the EEG data is mainly due to the instantaneous interference of the external environment.

[0056] Electrophysiological features are extracted from the denoised EEG signals to form a high-purity patient feature vector, and the distance between the patient feature vector and the information represented by the common disease pattern is calculated to obtain the first distance;

[0057] Simultaneously, the distance between the electrophysiological characteristics of the raw EEG data and the information represented by the common disease pattern is calculated to obtain the second distance, and the distance between the electrophysiological characteristics of the raw EEG data and the information of the inherent characteristics of each data source is calculated to obtain multiple third distances.

[0058] If the deviation of the second distance is significantly greater than that of the first distance, and the distance between the electrophysiological characteristics of the original EEG data and the information of the inherent characteristics of a certain historical data source is small, and the characteristics of the information of the inherent characteristics of the historical data source are highly similar to the real-time environmental noise information, then it is clearly determined that the original EEG data is affected by the inherent characteristics of a specific data source, and the deviation in the original EEG data is mainly caused by instantaneous external environmental interference.

[0059] After removing the influence of transient external environmental interference, the true physiological characteristics of patients are assessed based on the degree of deviation of the denoised patient characteristics.

[0060] Based on the patient's actual physiological characteristics, calculate the purity score and applicability risk index of the preliminary neuromodulation plan;

[0061] If the purity score is below the preset limit or the applicability risk index exceeds the acceptable threshold, a strategy adjustment is triggered. The strategy adjustment includes issuing an alert to the clinician, clearly indicating that the patient's EEG data was affected by transient external environmental interference but has been successfully denoised, and that the post-denoising assessment shows that the patient still has high physiological specificity. It is recommended to activate the exploratory treatment plan generation module and consider adjusting the stimulation target or parameter range to adapt to the patient's unique physiological pattern. At the same time, it is recommended that the physician check the acquisition environment.

[0062] Specifically, based on the patient's actual physiological characteristics, the purity score and applicability risk index of the preliminary neuromodulation plan are calculated, which can be:

[0063] The distance between the patient’s actual physiological characteristics and the information represented by the common disease pattern is calculated, and the distance is compared with a preset distance threshold. Based on the comparison results, a purity score and an applicability risk index are obtained. For example, if the distance is greater than the preset distance threshold, the purity score is rated as below the preset limit, and the applicability risk index is rated as exceeding the acceptable threshold.

[0064] If the distance value is less than the preset distance threshold, the purity score will be rated as higher than the preset limit, and the applicability risk index will be rated as not exceeding the acceptable threshold.

[0065] Specifically, an environmental sensor array can be understood as a group of devices used to detect and quantify electromagnetic interference in the acquired environment, including, for example, radio frequency (RF) sensors, magnetic field sensors, and power line noise sensors. These sensors are configured to operate synchronously during EEG signal acquisition to capture the intensity, frequency distribution, and transient fluctuations of the electromagnetic spectrum. Real-time environmental noise information refers to the environmental interference data recorded by these sensors at specific points in time, which is time-synchronized with the EEG acquisition data to ensure that subsequent analysis can accurately correlate noise with artifacts in the EEG signals.

[0066] Separating interfering components from raw EEG signals can be achieved through various signal processing techniques, such as adaptive filtering, independent component analysis (ICA), or wavelet denoising. The aim is to accurately identify and remove interference that closely matches real-time environmental noise, thereby obtaining a cleaner denoised EEG signal.

[0067] Determining whether the statistical characteristics of interference components are highly consistent with real-time environmental noise information and highly similar to the inherent circuit noise characteristics recorded in historical data sources can be achieved by calculating correlation coefficients, spectral similarity, or pattern recognition algorithms. For example, if the spectral characteristics of interference components highly match the spectral characteristics of known environmental noise sources (such as 50 / 60Hz power frequency interference) or the inherent noise of specific equipment, their source can be determined.

[0068] Furthermore, a high-purity patient feature vector refers to a set of features extracted from denoised EEG signals that accurately reflects the patient's electrophysiological state, such as power spectral density in specific frequency bands, latency and amplitude of event-related potentials (ERPs), and connectivity indices. Calculating the distance between the patient feature vector and the information represented by a common disease pattern aims to quantify the degree of deviation between the patient's physiological state and the standard disease pattern.

[0069] Simultaneously, calculating the distance between the raw EEG characteristics and the information represented by common disease patterns, as well as the information inherent in the characteristics of each data source, is for comparative analysis. If the deviation of the raw data is significantly greater than the deviation of the denoised data, and the distance between the raw EEG characteristics and the information inherent in a certain historical data source is small, and the characteristics of the information inherent in the historical data source are highly similar to the real-time environmental noise information, then it can be clearly determined that the deviation in the raw data is mainly caused by instantaneous external environmental interference, which may be misjudged as an inherent bias of the historical data source. This means that without environmental noise monitoring and denoising, the system may incorrectly attribute the signal deviation caused by environmental interference to the patient's own physiological characteristics or an inherent bias of a certain data source.

[0070] Therefore, after identifying and eliminating the influence of transient environmental interference, assessing the patient's true physiological characteristics based on the degree of deviation of the denoised patient features can more accurately reflect the patient's neurophysiological state. Based on the patient's true physiological characteristics, a purity score and applicability risk index for the preliminary neuromodulation protocol are calculated to quantify the degree of match between the treatment protocol and the patient's true physiological state, as well as potential risks. The purity score reflects the protocol's targeting of the patient's core physiological problems, while the applicability risk index assesses potential side effects or discomfort caused by the protocol.

[0071] If the purity score falls below a preset limit or the applicability risk index exceeds an acceptable threshold, the system will trigger a strategy adjustment. This adjustment includes alerting the clinician, clearly indicating that the patient's EEG data was affected by transient environmental interference, but that noise reduction has been successful, and that post-denoising assessment shows the patient still possesses high physiological specificity. Simultaneously, the system suggests activating the exploratory treatment plan generation module and considering adjusting the stimulation target or parameter range to suit the patient's unique physiological patterns. It also advises the physician to examine the acquisition environment to avoid similar interference in the future.

[0072] In some of the above embodiments, when the preliminary suitability assessment of the neuromodulation protocol shows insufficient suitability, the strategy adjustment is triggered. The strategy adjustment includes issuing a prompt message and generating an alternative treatment plan, and further includes:

[0073] After generating alternative treatment plans, physiological feedback data of patients during the period of receiving alternative treatment plans are collected;

[0074] Pre-defined physiological response patterns for alternative treatment options;

[0075] The physiological feedback data is compared with the expected physiological response pattern to determine the degree of deviation between the two.

[0076] The early effects of alternative treatment options can be assessed based on the degree of deviation.

[0077] When the initial results are unsatisfactory, new alternative treatment options are selected from the pool of alternative treatment options based on the specific physiological indicators and degree of deviation.

[0078] An alert is issued when two consecutive alternative treatments show poor early results.

[0079] Specifically, physiological feedback data can include, but is not limited to, the patient's real-time electroencephalogram (EEG) activity, heart rate variability, skin conductance, behavioral scores, and neurotransmitter levels. This data objectively reflects the patient's immediate physiological response to the treatment plan. Simultaneously, the system presets an expected physiological response pattern for the alternative treatment plan. This pattern, constructed based on extensive clinical data and disease models, represents the ideal physiological response the patient should have to the plan. The actual collected physiological feedback data is then compared with the expected physiological response pattern to quantify the degree of deviation. The degree of deviation can be determined by calculating statistical distance, similarity indices, or predicting using machine learning models. Based on the degree of deviation, the system can assess the early effectiveness of the alternative treatment plan. If the deviation exceeds a preset threshold, the early effectiveness is considered poor. When the early effectiveness is poor, the system will select new alternative treatment plans from a pre-established pool of candidates based on the specific physiological indicators of deviation (e.g., no improvement in EEG waveform, abnormal heart rate fluctuations) and the degree of deviation (e.g., large deviation magnitude, long duration), aiming to find a more suitable plan for the patient. In addition, to ensure the effectiveness and safety of treatment, if two consecutive alternative treatments fail to show good early results, the system will issue an alert, prompting clinicians to consider a more in-depth reassessment of the patient's diagnosis or treatment strategy.

[0080] In some of the above embodiments, the step of selecting new alternative treatment options from the pool of alternative treatments based on the specific physiological indicators and degree of deviation when the early effects are unsatisfactory includes:

[0081] Obtain the patient's clinical background information, including the patient's comorbidities, long-term medication history, age, and physiological changes.

[0082] A pre-defined set of clinical safety and feasibility assessment rules is provided, which includes compatibility limitations and risk warnings between alternative treatment options and clinical context information.

[0083] Each alternative treatment option in the pool of options is compared with the patient's clinical background information;

[0084] The safety and feasibility of each alternative treatment option was scored using a set of clinical safety and feasibility assessment rules.

[0085] Alternative treatment options are filtered and ranked based on safety and feasibility scores to select options that meet preset safety thresholds and have high feasibility scores.

[0086] Specifically, obtaining a patient's clinical background information refers to the collection of multifaceted data related to the patient's health status, either automatically or through manual input. Clinical background information can be understood as a comprehensive health record for the patient, aiming to provide a comprehensive and individualized reference for subsequent treatment plan evaluation. For example, comorbidities may include chronic diseases such as diabetes, hypertension, and heart disease; long-term medication history can record the types, dosages, and durations of medications the patient is currently taking or has taken, especially those that may interact with neuromodulation therapy; age is an important factor in assessing treatment tolerance and risk; physiological change data can cover body mass index, liver and kidney function indicators, allergy history, etc., all of which collectively depict the patient's overall physiological state and potential risks.

[0087] The pre-defined set of clinical safety and feasibility assessment rules can be understood as a set of predefined decision-making logics to guide the selection of treatment options. The rule set aims to ensure the safety and operability of the selected treatment in clinical practice, and its purpose is to avoid adverse events caused by incompatibility between the treatment and the individual patient's condition. Specifically, the rule set may include various conditional judgments. For example, if a patient has a pacemaker, certain electrical stimulation neuromodulation protocols may be marked as contraindicated; if a patient has severe hepatic or renal insufficiency, certain adjunctive therapies requiring hepatic or renal metabolism should be avoided; if the patient is too old or too young, the risk warning level for certain invasive treatments will be correspondingly increased. These rules can be constructed and continuously updated based on medical guidelines, expert experience, and historical clinical data.

[0088] Comparing each alternative treatment option in the pool of options with the patient's clinical background information means that the system cross-checks the characteristics of the options with the patient's individual circumstances. The purpose is to identify potential conflicts or incompatibilities. For example, the system might check whether a drug component of an alternative treatment interacts with the patient's long-term medication history; or whether an invasive surgical procedure conflicts with the patient's comorbidities (such as coagulation disorders).

[0089] Applying a set of clinical safety and feasibility assessment rules to score the safety and feasibility of each alternative treatment option means that the system quantifies the safety and implementation difficulty of each candidate option for a specific patient based on preset rules. The score can be a comprehensive numerical value or a combination of scores from multiple dimensions, aiming to provide a quantitative basis for subsequent option selection. For example, an option may score highly in safety but low in feasibility (such as operational complexity or equipment accessibility).

[0090] Therefore, filtering and ranking alternative treatment options based on safety and feasibility scores to select those that meet a preset safety threshold and have a high feasibility score means that the system prioritizes candidate options or directly excludes those that do not meet the criteria based on quantitative assessment results. The goal is to ensure that the final selected option has a high probability of successful implementation while guaranteeing patient safety. For example, the system can set a minimum safety score threshold; any option below this threshold will be directly filtered out. Then, among the remaining options, they are ranked according to feasibility scores, with the option with the highest feasibility score being given priority.

[0091] In some of the above embodiments, the step of filtering and ranking alternative treatment options based on safety and feasibility scores to select options that meet a preset safety threshold and have a high feasibility score includes:

[0092] Acquire the patient's current physiological status data, including the patient's real-time electroencephalogram (EEG) activity patterns, heart rate variability, and skin conductance response;

[0093] Based on physiological status data, the preset safety threshold and the weight of the feasibility score are adjusted. The adjustment process includes: if the patient shows an acute physiological stress response to the current treatment, the weight of the safety score is increased and the weight of the expected efficacy is decreased; if the patient's physiological status is stable and there is no obvious response to the existing treatment, the weight of the expected efficacy is increased and the weight of the safety score is decreased.

[0094] The adjusted weights were applied to the safety and feasibility scores to calculate the overall optimal score for each alternative treatment option;

[0095] Based on the comprehensive evaluation score, the alternative treatment options are ranked, and the option with the highest comprehensive evaluation score is selected.

[0096] If the overall optimal scores of multiple options differ within a pre-defined range of statistical uncertainty, then these multiple options are considered as a candidate set. Clinicians are then alerted to this candidate set, and a detailed evaluation report for each option is provided. This report includes the safety, feasibility, expected efficacy, and potential side effects of the option.

[0097] Specifically, real-time EEG activity patterns can reveal the immediate functional state of the brain and its response to stimuli; heart rate variability can reflect the activity of the autonomic nervous system and the patient's stress level; and skin conductance can indicate the excitation level of the sympathetic nervous system. These data are used to comprehensively assess the patient's current physiological condition. The system adjusts preset safety thresholds and feasibility scores based on physiological data to dynamically respond to the patient's immediate physiological needs during the treatment selection process. For example, when a patient exhibits an acute physiological stress response to the current treatment, the system prioritizes safety by increasing the weight of the safety score and decreasing the weight of the expected efficacy score to avoid treatments that might exacerbate patient discomfort or risk. Conversely, if the patient's physiological state is stable but there is no significant response to existing treatment, the system increases the weight of the expected efficacy score and decreases the weight of the safety score to encourage the selection of treatments that may bring more significant therapeutic effects, thereby avoiding treatment inertia. In practical application, the adjusted weights are applied to the safety and feasibility scores to calculate a comprehensive optimal score for each alternative treatment plan, which comprehensively reflects the merits of the plan in the patient's current physiological state. Subsequently, alternative treatment options are ranked according to their comprehensive optimization scores, and the option with the highest comprehensive optimization score is selected, ensuring the optimality of the selected option in the current context. Furthermore, if the comprehensive optimization scores of multiple options differ within a pre-defined range of statistical uncertainty, these multiple options are treated as a candidate set, and a prompt message is sent to the clinician, indicating the candidate set and providing a detailed evaluation report for each option. The aim is to provide clinicians with more comprehensive decision support when multiple near-optimal options exist, avoiding the limitations of a single choice and allowing physicians to make a final decision based on their professional judgment.

[0098] In some of the above embodiments, the step of adjusting the preset safety threshold and feasibility score weights based on physiological state data includes:

[0099] After acquiring the patient's current physiological state data, the physiological state data is preprocessed to obtain denoised physiological state data. The preprocessing includes:

[0100] Time-domain filtering is performed on physiological state data to remove high-frequency noise and transient spikes;

[0101] Frequency domain analysis of physiological state data is performed to identify and separate specific frequency components associated with known interference sources;

[0102] Baseline drift correction was applied to physiological state data to eliminate slow-wave artifacts;

[0103] Anomaly detection is performed on physiological state data to identify and mark instantaneous data points that exceed the normal physiological range;

[0104] Based on the outlier detection results, the marked instantaneous data points are interpolated or removed.

[0105] Based on the denoised physiological state data, the stability index of the patient's physiological state is calculated. The stability index reflects the fluctuation range and rate of change of the physiological state data over a period of time.

[0106] Based on the stability index, the preset safety threshold and the weight of the feasibility score are dynamically adjusted. The adjustment process includes:

[0107] When stability indicators show that the patient's physiological state fluctuates greatly or is abnormal, the system increases the weight of the safety score and decreases the weight of the expected efficacy.

[0108] When the stability index shows that the patient's physiological state is stable and there are no obvious abnormalities, the system increases the weight of the expected therapeutic effect and decreases the weight of the safety score.

[0109] Specifically, physiological state data can be understood as various biosignals reflecting a patient's current physiological condition, such as real-time electroencephalogram (EEG) activity patterns, heart rate variability, and skin conductance responses. This data is crucial for assessing a patient's immediate response to treatment and overall physiological load. Preprocessing of physiological state data aims to eliminate various noises and artifacts in the raw data, ensuring the accuracy of subsequent analysis. Specifically, time-domain filtering, using low-pass, high-pass, or band-pass filters, effectively removes high-frequency noise (such as electromyographic interference) and transient spikes (such as power frequency interference), aiming to improve the signal-to-noise ratio. Frequency-domain analysis, using methods such as Fourier transform, converts the signal from the time domain to the frequency domain, thereby identifying and separating specific frequency components associated with known interference sources (such as power line interference and environmental electromagnetic radiation), aiming to accurately remove interference at specific frequencies. Baseline drift correction removes slow-wave components from the signal, eliminating baseline drift caused by poor electrode contact, respiratory movements, or changes in body position, aiming to ensure the stability of the signal amplitude. Outlier detection uses statistical methods or machine learning algorithms to identify and label transient data points that exceed the normal physiological range, such as abnormal readings caused by sudden patient movement or equipment malfunction. Its purpose is to prevent abnormal data from misleading the overall assessment. Based on the outlier detection results, the labeled transient data points are interpolated (e.g., linear interpolation, spline interpolation) or removed to repair or remove damaged data and maintain the integrity and validity of the data sequence. In practical applications, stability indices can be understood as parameters that quantify the stability of a patient's physiological state over a period of time. These indices can be obtained by calculating the standard deviation, coefficient of variation, mean absolute deviation, or fluctuation range of power in a specific frequency band of denoised physiological state data within a specific time window. Their purpose is to objectively reflect the fluctuation range and rate of change of the patient's physiological state. For example, the standard deviation of heart rate variability and the short-term fluctuation amplitude of power in a specific frequency band of electroencephalogram (EEG) signals can both serve as stability indices.

[0110] In some of the above embodiments, the process of dynamically adjusting the weights of the preset safety threshold and feasibility score based on the stability index includes: when the stability index shows that the patient's physiological state fluctuates significantly or is abnormal, the system increases the weight of the safety score and decreases the weight of the expected therapeutic effect; and when the stability index shows that the patient's physiological state is stable and without obvious abnormalities, the system increases the weight of the expected therapeutic effect and decreases the weight of the safety score.

[0111] When the stability index shows that the patient's physiological state fluctuates greatly or is abnormal, the system starts an observation period with a preset time range, such as 1 hour, to continuously monitor the changes in the stability index.

[0112] If the stability index recovers to the preset stability range within the preset time range observation period, the system maintains the current safety and expected efficacy weights without adjustment;

[0113] If the stability index still shows large fluctuations or abnormalities after the observation period within the preset time range, the system will gradually increase the weight of the safety score and gradually decrease the weight of the expected efficacy.

[0114] When the stability index shows that the patient's physiological state is stable and there are no obvious abnormalities, the system increases the weight of the expected therapeutic effect and decreases the weight of the safety score.

[0115] Specifically, after the patient's physiological data is preprocessed and a stability index is calculated, if the stability index indicates significant fluctuations or abnormalities in the patient's physiological state, the system will not immediately adjust the weights of the safety threshold and feasibility score. Instead, the system will initiate a one-hour observation period to provide a buffer time, thereby distinguishing between transient, non-continuous physiological fluctuations and persistent physiological abnormalities requiring intervention. During this observation period, the system will continuously monitor changes in the stability index. For example, this preset time range observation period can be set from several minutes to several hours, with the specific duration determined based on clinical experience and the real-time requirements of the patient's physiological data.

[0116] If, within the preset timeframe of the observation period, the stability index spontaneously recovers to the preset stability range, it indicates that the previously detected fluctuations or abnormalities may have been merely transient, self-limiting physiological disturbances, rather than persistent risk signals. In this case, the system will maintain the current safety and expected efficacy weights without making any adjustments to avoid overreacting to transient fluctuations and thus maintain the stability of the treatment plan assessment.

[0117] However, if, after the observation period within the preset time frame, the stability indicators still show significant fluctuations in the patient's physiological state or persistent abnormalities, it indicates that the physiological abnormality is persistent or severe, requiring systemic intervention. In this case, the system will no longer maintain the original weights but will gradually increase the weight of the safety score and gradually decrease the weight of the expected therapeutic effect. This gradual adjustment strategy aims for a smooth transition, avoiding assessment turbulence that may result from sudden weight changes, while ensuring that the system can place greater emphasis on patient safety when the risk is confirmed to persist.

[0118] In addition, when the stability index shows that the patient's physiological state is stable and there are no obvious abnormalities, the system will increase the weight of the expected efficacy and decrease the weight of the safety score according to the established logic, so as to encourage the system to more actively explore and optimize the efficacy of the treatment plan under the premise of safety.

[0119] In some of the above embodiments, the step of increasing the weight of the expected therapeutic effect and decreasing the weight of the safety score when the stability index shows that the patient's physiological state is stable and without obvious abnormalities includes:

[0120] Continuous monitoring of patients' physiological response patterns, including the power change trend of specific frequency bands of EEG signals, the dynamic range of neurotransmitter-related indicators, and the long-term rate of change of behavioral scores;

[0121] Identify adaptive or tolerable characteristics in the patient's physiological response pattern. Adaptive or tolerable characteristics include: (1) Physiological indicators show significant improvement in the early stage of treatment. For example, the range of change of physiological indicators in the early stage of treatment exceeds the preset target range, which indicates significant improvement. The preset target range can be set as needed and is not limited here. However, the physiological indicators tend to level off during the treatment period. For example, the range of change of physiological indicators does not exceed the preset target range during the treatment period, which indicates leveling off and not reaching the preset ideal treatment goal; (2) Physiological indicators no longer show sensitivity to parameter adjustments of the treatment plan. When adaptive or tolerable characteristics are identified, an inertial assessment of the current treatment plan is triggered. The inertial assessment includes: comparing the patient's current physiological response pattern with the preset adaptive and tolerable physiological response pattern and calculating the degree of matching.

[0122] If the inertia assessment results show that the matching degree exceeds the preset threshold, the true effectiveness of the current treatment plan will be reassessed, and the weight adjustment strategy will be adjusted. The adjustment strategy includes: appropriately reducing the weight of the expected efficacy and appropriately increasing the weight of the safety score, while suggesting the generation of alternative treatment plans with different mechanisms of action.

[0123] If the inertial evaluation results show that the matching degree does not exceed the preset threshold, the current weight adjustment strategy will be maintained.

[0124] Specifically, the power variation trend of specific frequency bands in electroencephalogram (EEG) signals can reflect the dynamic changes in brain activity across different frequency ranges. For example, changes in the power of theta or alpha waves may be related to emotion, cognition, or pain perception. The dynamic range of neurotransmitter-related indicators can be assessed through biochemical tests or indirect physiological indicators (such as pupillary response and heart rate variability) to reflect changes in the activity of neurotransmitter systems (such as dopamine and serotonin). The long-term rate of change in behavioral scores is obtained through regular behavioral scale assessments (such as depression scales and anxiety scales) to measure the improvement or deterioration trend of patients' clinical symptoms. The comprehensive monitoring of these multimodal data aims to fully capture the subtle physiological and behavioral feedback of patients to treatment.

[0125] Identifying adaptive or tolerant characteristics means that the system can recognize signs that a patient has adapted to or tolerated the current treatment regimen. For example, when a patient shows significant improvement in physiological indicators at the beginning of treatment, but these improvements gradually level off or even stagnate as treatment continues, failing to reach the preset ideal treatment goal, this is considered an adaptive characteristic. Another scenario is when the patient's physiological indicators no longer show a sensitive response when the parameters of the treatment regimen (such as stimulation intensity and frequency) are adjusted, indicating that tolerance to the treatment may have formed. Identifying these characteristics is crucial for determining whether a treatment regimen has entered an inertia phase.

[0126] When adaptive or tolerable characteristics are identified, the system triggers an inertial assessment of the current treatment plan. This assessment compares the patient's current physiological response pattern with pre-established physiological response patterns representing adaptive and tolerable states, calculating the degree of match. The pre-defined adaptive and tolerable physiological response patterns can be constructed based on extensive clinical data and expert experience, encompassing typical characteristics of different types of adaptive or tolerable manifestations. The degree of match can be calculated using various similarity measurement algorithms, such as cosine similarity and Euclidean distance, to quantify the closeness of the current patient state to known inertial patterns.

[0127] If the indolence assessment shows a match exceeding a preset threshold, indicating that the patient has developed significant adaptation or tolerance to the current treatment plan, the system will reassess the true effectiveness of the current treatment. Based on this, the weighting strategy will be adjusted accordingly, specifically by moderately reducing the weight of expected efficacy and moderately increasing the weight of safety scores. This means the system no longer excessively pursues improved treatment efficacy but focuses more on treatment safety to avoid potential risks from ineffective treatment. Simultaneously, the system will suggest generating alternative treatment plans with different mechanisms of action, aiming to break the patient's adaptation or tolerance through novel interventions. Conversely, if the indolence assessment shows a match not exceeding the preset threshold, indicating that the patient has not yet developed significant adaptation or tolerance to the current treatment plan, the system will maintain the current weighting strategy and continue optimization according to the original logic.

[0128] Based on the same inventive concept, embodiments of this application also disclose an electroencephalogram (EEG) signal information processing system applied to neural modulation, such as... Figure 2 As shown, it includes:

[0129] The data source characteristic information acquisition and storage module 1 is used to acquire and store the inherent characteristic information of the data source before processing patient medical data, specifically:

[0130] Medical data is obtained from multiple sources, including electroencephalogram (EEG) data.

[0131] Based on medical data, information reflecting the inherent characteristics of each data source is extracted and constructed. For EEG data, information on EEG fingerprint characteristics is obtained by analyzing the noise spectrum characteristics, baseline drift patterns, and frequency of occurrence of specific artifacts from different data sources.

[0132] The characteristic information of brainwave fingerprints is stored to obtain information that reflects the inherent characteristics of the data source;

[0133] Patient medical data processing module 2 is used to acquire patient medical data to be processed, extract key information from the patient medical data, and standardize the key information.

[0134] The difference determination module 3 is used to determine the degree of difference between the patient's key information and the information reflecting the inherent characteristics of the data source, as well as the degree of difference between the patient's key information and the information representing the common disease pattern, based on the standardized key patient information, as well as the pre-stored information reflecting the inherent characteristics of the data source and the information representing the common disease pattern.

[0135] The applicability assessment module 4 is used to determine, based on the determined degree of difference, whether the patient's key information is affected by the inherent characteristics of the specific data source and whether it deviates from the general disease pattern, and based on the judgment results, to assess the applicability of the preliminary neuromodulation plan generated for the patient.

[0136] The strategy adjustment module 5 is used to trigger strategy adjustment when the initial suitability assessment of the neuromodulation protocol shows insufficient suitability. The strategy adjustment includes issuing prompts and generating alternative treatment plans.

[0137] This application enables intelligent processing of electroencephalogram (EEG) signal information in neuromodulation.

Claims

1. A method for processing electroencephalogram (EEG) signals applied to neural modulation, characterized in that, include: Before processing patient medical data, acquire and store the inherent characteristics of the data source, specifically: Medical data is acquired from multiple sources, including EEG data. Based on the medical data, information reflecting the inherent characteristics of each data source is extracted and constructed. For EEG data, information on EEG fingerprint characteristics is obtained by analyzing the noise spectrum characteristics, baseline drift patterns, and frequency of occurrence of specific artifacts from different data sources. By storing the characteristic information of brainwave fingerprints, information reflecting the inherent characteristics of the data source can be obtained; Acquire patient medical data to be processed, extract key information from the patient medical data, and standardize the key information; Based on standardized key patient information, and according to pre-stored information reflecting the inherent characteristics of the data source and information represented by established common disease patterns, the degree of difference between key patient information and information reflecting the inherent characteristics of the data source and the degree of difference between key patient information and information represented by common disease patterns are determined respectively. Based on the determined degree of difference, determine whether the patient's key information is affected by the inherent characteristics of the specific data source and whether it deviates from the general disease pattern. Based on the judgment results, assess the applicability of the preliminary neuromodulation plan generated for the patient. When the initial suitability assessment of a neuromodulation protocol shows inadequacy, a strategy adjustment is triggered, which includes issuing prompts and generating alternative treatment options.

2. The EEG signal information processing method for neural modulation according to claim 1, characterized in that, The steps of determining whether the patient's key information is affected by the inherent characteristics of a specific data source and whether it deviates from a common disease pattern based on the determined degree of difference, and assessing the applicability of the preliminary neuromodulation plan generated for the patient based on the determination results, include: While collecting patients' electroencephalogram (EEG) data, an environmental sensor array is activated to monitor the electromagnetic environment in the collection room in real time, capturing the intensity, frequency distribution, and instantaneous fluctuations of the electromagnetic spectrum. Interference components that highly match real-time environmental noise information are separated from the raw EEG data to obtain the denoised EEG signal; If the statistical characteristics of the interference components are highly consistent with the real-time environmental noise information, and the characteristics of the interference components show a high degree of similarity with the inherent circuit noise characteristics recorded by the characteristic information of a certain historical data source in the system database, it is determined that the EEG data is affected by the inherent characteristics of a specific data source. Electrophysiological features are extracted from the denoised EEG signals to form a high-purity patient feature vector, and the distance between the patient feature vector and the information represented by the common disease pattern is calculated to obtain the first distance; Simultaneously, the distance between the electrophysiological characteristics of the raw EEG data and the information represented by the common disease pattern is calculated to obtain the second distance, and the distance between the electrophysiological characteristics of the raw EEG data and the information of the inherent characteristics of each data source is calculated to obtain multiple third distances. If the deviation of the second distance is significantly greater than that of the first distance, and the distance between the electrophysiological characteristics of the original EEG data and the information of the inherent characteristics of a certain historical data source is small, and the characteristics of the information of the inherent characteristics of the historical data source are highly similar to the real-time environmental noise information, then it is clearly determined that the original EEG data is affected by the inherent characteristics of a specific data source, and the deviation in the original EEG data is mainly caused by instantaneous external environmental interference. After removing the influence of transient external environmental disturbances, the patient's true physiological characteristics are obtained; Based on the patient's actual physiological characteristics, the purity score and applicability risk index of the preliminary neuromodulation plan are calculated. The purity score is used to reflect the target of the preliminary neuromodulation plan for the patient's core physiological problems, and the applicability risk index is used to assess the possible side effects or discomfort caused by the plan. If the purity score is below the preset limit or the applicability risk index exceeds the acceptable threshold, a strategy adjustment is triggered. The strategy adjustment includes issuing an alert to the clinician, clearly indicating that the patient's EEG data was affected by transient external environmental interference but has been successfully denoised, and that the post-denoising assessment shows that the patient still has high physiological specificity. It is recommended to activate the exploratory treatment plan generation module and consider adjusting the stimulation target or parameter range to adapt to the patient's unique physiological pattern. It is also recommended to check the acquisition environment.

3. The brainwave signal information processing method for neural modulation according to claim 1, characterized in that, When the initial suitability assessment of the neuromodulation protocol shows insufficient suitability, strategy adjustment is triggered. This strategy adjustment includes issuing prompts and generating alternative treatment plans. After generating alternative treatment plans, physiological feedback data of patients during the period of receiving alternative treatment plans are collected; Pre-defined physiological response patterns for alternative treatment options; The physiological feedback data is compared with the expected physiological response pattern to determine the degree of deviation between the two. The early effects of alternative treatment options can be assessed based on the degree of deviation. When the initial results are unsatisfactory, new alternative treatment options are selected from the pool of alternative treatment options based on the specific physiological indicators and degree of deviation. An alert is issued when two consecutive alternative treatments show poor early results.

4. The method for processing electroencephalogram (EEG) signals for neural modulation according to claim 3, characterized in that, When early results are unsatisfactory, the step of selecting new alternative treatment options from the pool of alternatives based on the specific physiological indicators and degree of deviation includes: Obtain the patient's clinical background information, including the patient's comorbidities, long-term medication history, age, and physiological changes. A pre-defined set of clinical safety and feasibility assessment rules is provided, which includes compatibility limitations and risk warnings between alternative treatment options and clinical context information. Each alternative treatment option in the pool of options is compared with the patient's clinical background information; The safety and feasibility of each alternative treatment option was scored using a set of clinical safety and feasibility assessment rules. Alternative treatment options are filtered and ranked based on safety and feasibility scores to select options that meet preset safety thresholds and have high feasibility scores.

5. The EEG signal information processing method for neural modulation according to claim 4, characterized in that, The step of filtering and ranking alternative treatment options based on safety and feasibility scores to select options that meet a preset safety threshold and have a high feasibility score includes: Acquire the patient's current physiological status data, including the patient's real-time electroencephalogram (EEG) activity patterns, heart rate variability, and skin conductance response; Based on physiological status data, the preset safety threshold and the weight of the feasibility score are adjusted. The adjustment process includes: if the patient shows an acute physiological stress response to the current treatment, the weight of the safety score is increased and the weight of the expected efficacy is decreased; if the patient's physiological status is stable and there is no obvious response to the existing treatment, the weight of the expected efficacy is increased and the weight of the safety score is decreased. The adjusted weights were applied to the safety and feasibility scores to calculate the overall optimal score for each alternative treatment option; Based on the comprehensive evaluation score, the alternative treatment options are ranked, and the option with the highest comprehensive evaluation score is selected. If the overall optimal scores of multiple options differ within a pre-defined range of statistical uncertainty, then these multiple options are considered as a candidate set. Clinicians are then alerted to this candidate set, and a detailed evaluation report for each option is provided. This report includes the safety, feasibility, expected efficacy, and potential side effects of the option.

6. The EEG signal information processing method for neural modulation according to claim 5, characterized in that, The step of adjusting the preset safety threshold and feasibility score weights based on physiological state data includes: After acquiring the patient's current physiological state data, the physiological state data is preprocessed to obtain denoised physiological state data. The preprocessing includes: Time-domain filtering is performed on physiological state data to remove high-frequency noise and transient spikes; Frequency domain analysis of physiological state data is performed to identify and separate specific frequency components associated with known interference sources; Baseline drift correction was applied to physiological state data to eliminate slow-wave artifacts; Anomaly detection is performed on physiological state data to identify and mark instantaneous data points that exceed the normal physiological range; Based on the outlier detection results, the marked instantaneous data points are interpolated or removed. Based on the denoised physiological state data, the stability index of the patient's physiological state is calculated. The stability index reflects the fluctuation range and rate of change of the physiological state data over a period of time. Based on the stability index, the preset safety threshold and feasibility score weights are dynamically adjusted. The adjustment process includes: When stability indicators show that the patient's physiological state fluctuates greatly or is abnormal, the system increases the weight of the safety score and decreases the weight of the expected efficacy. When the stability index shows that the patient's physiological state is stable and there are no obvious abnormalities, the system increases the weight of the expected therapeutic effect and decreases the weight of the safety score.

7. The EEG signal information processing method for neural modulation according to claim 6, characterized in that, The step of dynamically adjusting the weights of the preset safety threshold and feasibility score based on the stability index includes: when the stability index shows that the patient's physiological state fluctuates greatly or is abnormal, the system increases the weight of the safety score and decreases the weight of the expected efficacy; and when the stability index shows that the patient's physiological state is stable and without obvious abnormalities, the system increases the weight of the expected efficacy and decreases the weight of the safety score. When the stability index shows that the patient's physiological state fluctuates greatly or is abnormal, the system starts a one-hour observation cycle to continuously monitor the changes in the stability index. If the stability index recovers to the preset stability range within the preset time range observation period, the system maintains the current safety and expected efficacy weights without adjustment; If the stability index still shows large fluctuations or abnormalities after the observation period within the preset time range, the system will gradually increase the weight of the safety score and gradually decrease the weight of the expected efficacy. When the stability index shows that the patient's physiological state is stable and there are no obvious abnormalities, the system increases the weight of the expected therapeutic effect and decreases the weight of the safety score.

8. The method for processing electroencephalogram (EEG) signals for neural modulation according to claim 6, characterized in that, When the stability index shows that the patient's physiological state is stable and without significant abnormalities, the steps of increasing the weight of the expected therapeutic effect and decreasing the weight of the safety score include: Continuous monitoring of patients' physiological response patterns, including the power change trend of specific frequency bands of EEG signals, the dynamic range of neurotransmitter-related indicators, and the long-term rate of change of behavioral scores; Identify adaptive or tolerable characteristics in the patient's physiological response pattern. Adaptive or tolerable characteristics include: (1) the range of change of physiological indicators in the early stage of treatment exceeds the preset target range relative to the untreated range, but the range of change of physiological indicators does not exceed the preset target range during the subsequent treatment period, and the preset ideal treatment goal is not achieved; (2) physiological indicators no longer show sensitivity to parameter adjustments of the treatment plan; when adaptive or tolerable characteristics are identified, trigger an inertial assessment of the current treatment plan. The inertial assessment includes: comparing the patient's current physiological response pattern with the preset adaptive and tolerable physiological response pattern and calculating the degree of matching. If the inertia assessment results show that the matching degree exceeds the preset threshold, the true effectiveness of the current treatment plan will be reassessed, and the weight adjustment strategy will be adjusted. The adjustment strategy includes: appropriately reducing the weight of the expected efficacy and appropriately increasing the weight of the safety score, while suggesting the generation of alternative treatment plans with different mechanisms of action. If the inertial evaluation results show that the matching degree does not exceed the preset threshold, the current weight adjustment strategy will be maintained.

9. A brainwave signal information processing system for neural modulation, characterized in that, include: The data source characteristic information acquisition and storage module is used to acquire and store the inherent characteristic information of the data source before processing patient medical data, specifically: Medical data is obtained from multiple sources, including electroencephalogram (EEG) data. Based on medical data, information reflecting the inherent characteristics of each data source is extracted and constructed. For EEG data, information on EEG fingerprint characteristics is obtained by analyzing the noise spectrum characteristics, baseline drift patterns, and frequency of occurrence of specific artifacts from different data sources. The characteristic information of brainwave fingerprints is stored to obtain information that reflects the inherent characteristics of the data source; The patient medical data processing module is used to acquire patient medical data to be processed, extract key information from the patient medical data, and standardize the key information. The difference determination module is used to determine the degree of difference between the patient's key information and the information reflecting the inherent characteristics of the data source, as well as the degree of difference between the patient's key information and the information representing the common disease pattern, based on standardized key patient information, as well as pre-stored information reflecting the inherent characteristics of the data source and information representing the common disease pattern. The protocol applicability assessment module is used to determine, based on the determined degree of difference, whether the patient's key information is affected by the inherent characteristics of a specific data source, and whether it deviates from the general disease pattern. Based on the assessment results, it evaluates the applicability of the preliminary neuromodulation protocol generated for the patient. The strategy adjustment module is used to trigger strategy adjustment when the initial suitability assessment of the neuromodulation protocol shows insufficient suitability. The strategy adjustment includes issuing prompts and generating alternative treatment plans.

Images