Regulation data denoising method and system based on implantable medical device

Abstract

The application discloses a kind of regulation and control data denoising method and system based on implantable medical device, belong to data processing technical field.The application first generates several data optimization execution strategies with denoising activation features and denoising temporary stop features, can intelligently identify and respond to the data noise characteristics under different space-time monitoring regulation and control tasks, to improve the pertinence and effectiveness of denoising processing.Secondly, the correlation atlas between these data optimization execution strategies is further determined, so that the selection and execution of denoising strategy are more scientific and reasonable, and potential conflicts and misoperations in the denoising process are avoided.Finally, the target denoising instruction text related to each space-time monitoring regulation and control task is generated with the target living body monitoring information as the denoising benchmark, and this highly customized data optimization scheme is issued to the edge side data optimization equipment, realizing accurate preprocessing and optimization of original data, and significantly improving the accuracy and reliability of data.

Description

Technical Field

[0001] This application belongs to the field of data processing technology, specifically relating to a method and system for denoising control data based on implantable medical devices. Background Technology

[0002] With the rapid development of modern medical technology, implantable neuromonitoring and modulation technology, as an innovative treatment method, is gradually demonstrating its enormous potential. However, due to the complexity and variability of the internal environment of organisms, and the various unavoidable interference factors during the monitoring and modulation process, the data collected by implantable monitoring and modulation systems often contains a large amount of noise. This seriously affects the accuracy and reliability of the data, making it difficult to provide a stable and reliable data source for subsequent data management and analysis (such as data sharing, data storage, data visualization, and data textification). Therefore, effective noise reduction of implantable neuromonitoring and modulation data has become an urgent problem to be solved. Summary of the Invention

[0003] This application provides a method and system for denoising control data based on implantable medical devices, which can solve or partially solve the technical problems involved in the background art mentioned above.

[0004] This application provides a method for denoising control data based on implantable medical devices, applied to a control data denoising system. The method includes:

[0005] The data denoising system acquires target liveness monitoring information from the implanted monitoring and control information pool and several spatiotemporal monitoring and control tasks belonging to the implanted monitoring and control information pool, with some overlap between the several spatiotemporal monitoring and control tasks.

[0006] The data denoising system generates several data optimization execution strategies with denoising activation and denoising pause features based on the target liveness monitoring information and the several spatiotemporal monitoring and control tasks.

[0007] The data denoising system determines the correlation map corresponding to the several data optimization execution strategies based on the denoising activation features and denoising pause features corresponding to the several data optimization execution strategies.

[0008] The data denoising system generates target denoising instruction text based on the correlation map, using the target liveness monitoring information as the denoising benchmark, and covering various spatiotemporal monitoring and control tasks.

[0009] The control data denoising system sends the target denoising instruction text to the edge-side data optimization device, so that the edge-side data optimization device can optimize the control data of the several spatiotemporal domain monitoring and control tasks through the target denoising instruction text.

[0010] Optionally, the generation of several data optimization execution strategies with denoising activation and denoising pause features based on the target liveness monitoring information and the several spatiotemporal monitoring and control tasks includes:

[0011] Based on the monitoring expectation mode of the spatiotemporal domain monitoring and control tasks, the several spatiotemporal domain monitoring and control tasks are respectively mapped into several first task knowledge graphs containing denoising activation features and denoising pausing features.

[0012] The target liveness monitoring information is mapped into a second task knowledge graph containing the same denoising activation features and denoising pause features;

[0013] The data optimization execution strategies are generated based on the aforementioned first task knowledge graphs and second task knowledge graphs.

[0014] Optionally, determining the correlation map corresponding to the plurality of data optimization execution strategies based on the denoising activation features and denoising pause features corresponding to the plurality of data optimization execution strategies includes:

[0015] Obtain the denoising activation features and denoising pause features corresponding to each data optimization execution strategy;

[0016] Based on the denoising activation features and denoising pause features corresponding to each of the data optimization execution strategies, potential noisy image descriptors are determined.

[0017] Based on the denoising activation features and denoising pause features corresponding to each of the data optimization execution strategies, the correlation degree between the first strategy identifier of each data optimization execution strategy and the second strategy identifier of the remaining data optimization execution strategies is determined.

[0018] The latent noise image descriptors are mapped into the correlation map based on the determined correlation degree.

[0019] Optionally, obtaining the denoising activation features and denoising pause features corresponding to each data optimization execution strategy includes:

[0020] From the aforementioned data optimization execution strategies, a target data optimization execution strategy that is related to the historical execution strategies in the embedded monitoring and control information pool in the monitoring and control dimension is obtained; based on the denoising activation features and denoising pause features of the historical execution strategies, the denoising activation features and denoising pause features corresponding to the target data optimization execution strategy are determined;

[0021] or,

[0022] Using the topological labels corresponding to the target liveness monitoring information in the aforementioned data optimization execution strategies as a reference, a target feature space is determined; the strategy node variables of each data optimization execution strategy other than the topological labels corresponding to the target liveness monitoring information are obtained; and the denoising activation features and denoising pause features of each data optimization execution strategy in the target feature space are determined based on the strategy node variables of each data optimization execution strategy.

[0023] Optionally, mapping the latent noise image descriptor to the correlation map based on the determined correlation degree includes:

[0024] Mining target denoising activation features and target denoising pause features from the latent noise image descriptor;

[0025] If the target denoising activation feature and the target denoising pause feature correspond to the same data optimization execution strategy, then the graph semantic unit corresponding to the target denoising activation feature and the target denoising pause feature is determined to be a preset semantic code.

[0026] If the target denoising activation feature and the target denoising pause feature correspond to different data optimization execution strategies, then the cosine similarity between the target denoising pause feature and the target denoising activation feature is taken from the determined correlation and used as the graph semantic unit corresponding to the target denoising activation feature and the target denoising pause feature, so as to obtain the correlation graph.

[0027] Optionally, the step of acquiring target liveness monitoring information from the implantable monitoring and control information pool and several spatiotemporal monitoring and control tasks belonging to the implantable monitoring and control information pool includes:

[0028] Obtain the historical execution strategies created by the noise reduction strategy planning system from the implanted monitoring and control information pool;

[0029] The execution strategies that exclude the historical execution strategies from the implanted monitoring and control information pool are taken as the execution strategies to be processed.

[0030] Based on the information involvement weight of the execution strategy to be processed, the implanted monitoring and control information pool is decomposed into several monitoring and control information sets.

[0031] The several spatiotemporal monitoring and control tasks are determined based on the pending execution strategies within the target monitoring and control information set, and the target liveness monitoring information is determined using the target monitoring and control information set.

[0032] Optionally, the step of acquiring target liveness monitoring information from the implantable monitoring and control information pool and several spatiotemporal monitoring and control tasks belonging to the implantable monitoring and control information pool includes:

[0033] Knowledge mining is performed on the in vivo physiological attribute data corresponding to the implantable monitoring and regulation information pool to obtain a knowledge relationship network of in vivo physiological attributes;

[0034] Based on the in vivo physiological attribute knowledge network, the priority task events among the various task events of the monitoring and regulation tasks in the implantable monitoring and regulation information pool are determined.

[0035] If the priority task event does not have an associated priority task event, then the monitoring and control task corresponding to the priority task event shall be taken as the spatiotemporal domain monitoring and control task.

[0036] If the priority task event has associated priority task events that do not belong to the same data optimization execution strategy, and the correlation degree between the priority task event and the associated priority task event is less than or equal to a set correlation degree, then the monitoring and control task corresponding to the priority task event is taken as the spatiotemporal domain monitoring and control task.

[0037] Optionally, the step of generating target denoising indication text based on the correlation map, using the target liveness monitoring information as a denoising benchmark, and involving various spatiotemporal domain monitoring and control tasks, includes:

[0038] Based on the correlation map and long short-term memory model, the task chain characteristics of each spatiotemporal monitoring and control task are determined.

[0039] Using the target liveness monitoring information as the denoising benchmark, and based on the task chain characteristics of each spatiotemporal monitoring and control task, a target denoising instruction text with the highest timeliness weight is generated;

[0040] The task chain characteristics for determining each spatiotemporal monitoring and control task based on the correlation map and long short-term memory model include:

[0041] Based on the correlation map, the target denoising compliance feature with the highest correlation to the denoising benchmark of the target liveness monitoring information is determined from various spatiotemporal monitoring and control tasks.

[0042] The target denoising compliance feature is used as the denoising benchmark for the derived liveness monitoring information. Based on the correlation map, the target denoising compliance feature with the highest correlation to the denoising benchmark of the derived liveness monitoring information is determined from the remaining spatiotemporal monitoring and control tasks other than the target spatiotemporal monitoring and control task.

[0043] Jump to the step of using the target denoising compliance feature as the denoising benchmark for the derived liveness monitoring information, and based on the correlation map, determine the target denoising compliance feature with the highest correlation to the denoising benchmark for the derived liveness monitoring information from the remaining spatiotemporal monitoring and control tasks other than the target spatiotemporal monitoring and control task, until all spatiotemporal monitoring and control tasks have been traversed.

[0044] The traversal time sequence label of each target spatiotemporal domain monitoring and control task is used as the task chain feature of each spatiotemporal domain monitoring and control task.

[0045] The spatiotemporal domain monitoring and control task is a neural signal monitoring and control task; the step of generating the target denoising indication text with the highest timeliness weight based on the task chain features of each spatiotemporal domain monitoring and control task includes:

[0046] Based on the task chain characteristics of each spatiotemporal domain monitoring and control task, the neural signal monitoring and control elements between two consecutive spatiotemporal domain monitoring and control tasks are determined.

[0047] If the neural signal monitoring and control element represents a labeled monitoring and control task that has completed denoising optimization in the implanted monitoring and control information pool between the two consecutive spatiotemporal domain monitoring and control tasks, then the target denoising instruction text is generated based on the two consecutive spatiotemporal domain monitoring and control tasks and the labeled monitoring and control task that has completed denoising optimization.

[0048] If the neural signal monitoring and control elements indicate that there are no related task items between the two consecutive spatiotemporal domain monitoring and control tasks, then, taking the two consecutive spatiotemporal domain monitoring and control tasks as a reference, several spatiotemporal domain monitoring and control tasks are decomposed to obtain a first target spatiotemporal domain monitoring and control task set and a second target spatiotemporal domain monitoring and control task set; based on the first target spatiotemporal domain monitoring and control task set and the second target spatiotemporal domain monitoring and control task set, a first alternative denoising instruction text and a second alternative denoising instruction text are generated respectively.

[0049] Optionally, after sending the target denoising instruction text to the edge-side data optimization device, the method further includes:

[0050] Obtain the current device status characteristics of the edge-side data optimization device;

[0051] If the current device status features do not match the target denoising instruction text, and the number of spatiotemporal monitoring and control tasks to be processed is less than a set number, then the correlation between the spatiotemporal monitoring and control tasks to be processed is identified based on the cross-modal decision tree, and the denoising instruction update text is determined based on the correlation between the current device status features and the spatiotemporal monitoring and control tasks to be processed.

[0052] If the current device status features do not match the target denoising instruction text, and the number of spatiotemporal monitoring and control tasks to be processed is greater than the set number, then a new denoising instruction update text with the highest timeliness weight among all the spatiotemporal monitoring and control tasks to be processed will be generated based on the current device status features.

[0053] The denoising instruction update text is sent to the edge-side data optimization device so that the edge-side data optimization device can optimize the spatiotemporal domain monitoring and control task to be processed through the denoising instruction update text.

[0054] This application provides a data denoising system, including at least one processor and a memory; the memory stores computer execution instructions; the at least one processor executes the computer execution instructions stored in the memory, causing the at least one processor to perform the above-described method.

[0055] This application provides a readable storage medium on which a program or instruction is stored, and when the program or instruction is executed by a processor, it implements the steps of the above method.

[0056] This application first generates several data optimization execution strategies with denoising activation and denoising pause features. These strategies intelligently identify and address data noise characteristics under different spatiotemporal monitoring and control tasks, thereby improving the targeting and effectiveness of denoising processing. Secondly, it further determines the correlation map between these data optimization execution strategies, making the selection and execution of denoising strategies more scientific and reasonable, avoiding potential conflicts and misoperations during the denoising process. Finally, it generates target denoising instruction text based on target liveness monitoring information, covering various spatiotemporal monitoring and control tasks. This highly customized data optimization scheme is then distributed to edge-side data optimization devices, achieving precise preprocessing and optimization of the original data, significantly improving data accuracy and reliability. Attached Figure Description

[0057] Figure 1 This is a flowchart illustrating a method for denoising control data based on implantable medical devices, provided as an embodiment of this application.

[0058] Figure 2 This is a schematic diagram of the structure of a data denoising system provided in an embodiment of this application. Detailed Implementation

[0059] The technical solutions of the embodiments 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, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0060] The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, a first object can be one or more. Furthermore, in this application, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects have an "or" relationship.

[0061] Figure 1 A method for denoising control data based on implantable medical devices is shown, which is applied to a control data denoising system. The method includes the following steps 110-150.

[0062] Step 110: The control data denoising system acquires the target liveness monitoring information in the implanted monitoring and control information pool and several spatiotemporal monitoring and control tasks belonging to the implanted monitoring and control information pool, and there is overlap between the several spatiotemporal monitoring and control tasks.

[0063] The primary task of the regulatory data denoising system is to acquire target in vivo monitoring information and several related spatiotemporal monitoring and regulatory tasks from the implanted monitoring and regulatory information pool. Target in vivo monitoring information and spatiotemporal monitoring and regulatory tasks form the basis for the system's data optimization processing. The implanted monitoring and regulatory information pool is a database integrating various biological signals and regulatory parameters, recording the physiological state of the target in vivo and the operational status of the neural monitoring and regulatory system in real time. Spatiotemporal monitoring and regulatory tasks refer to specific monitoring and regulatory tasks performed on the target in vivo under different time and spatial conditions. These tasks often overlap and intersect, forming a complex data association network.

[0064] Step 120: The data denoising system generates several data optimization execution strategies with denoising activation and denoising pause features based on the target liveness monitoring information and the several spatiotemporal monitoring and control tasks.

[0065] After acquiring target liveness monitoring information and spatiotemporal monitoring and control tasks, the control data denoising system generates several data optimization execution strategies with denoising activation and denoising cessation features based on this information. Denoising activation features are those that can activate the denoising mechanism, effectively filtering and processing the data; they are typically closely related to the physiological state of the target liveness organism, the nature of the monitoring and control task, and the noise characteristics of the data. Denoising cessation features, on the other hand, are those features that are not suitable for denoising under the current conditions; their existence helps avoid errors and data loss during the denoising process.

[0066] Step 130: The data denoising system determines the correlation map corresponding to the several data optimization execution strategies based on the denoising activation features and denoising pause features corresponding to the several data optimization execution strategies.

[0067] Next, the data denoising system will optimize the execution strategies based on the generated data and determine the correlation graph between them. The correlation graph is a multi-dimensional data model that describes the interactions and dependencies between different execution strategies. Through the correlation graph, the data denoising system can more intuitively understand which data optimization execution strategies are mutually reinforcing and which may conflict, thus providing a strong basis for the subsequent generation of denoised instruction text.

[0068] Step 140: The data denoising system generates target denoising instruction text based on the correlation map, using the target liveness monitoring information as the denoising benchmark, and involving various spatiotemporal monitoring and control tasks.

[0069] After determining the correlation map of the data optimization execution strategy, the control data denoising system further generates target denoising instruction texts based on the target liveness monitoring information, covering various spatiotemporal domain monitoring and control tasks. The target denoising instruction text is a highly customized data optimization scheme that details how to select and execute denoising strategies under different spatiotemporal domain monitoring and control tasks to achieve the best denoising effect. The generation of the target denoising instruction text signifies that the control data denoising system has moved from the macro-level data perspective to the specific operational level, providing clear guidance for edge-side data optimization equipment.

[0070] Step 150: The control data denoising system sends the target denoising instruction text to the edge-side data optimization device, so that the edge-side data optimization device can optimize the control data of the several spatiotemporal domain monitoring and control tasks through the target denoising instruction text.

[0071] Finally, the data denoising system sends the generated target denoising instruction text to the edge-side data optimization device. The edge-side data optimization device is a crucial component of the implantable monitoring and control system, responsible for directly preprocessing and optimizing the collected raw data. Upon receiving the target denoising instruction text, the edge-side data optimization device optimizes the control data for several spatiotemporal monitoring and control tasks according to the guidance within the text. Through precise denoising processing and data filtering, the edge-side data optimization device significantly improves the accuracy and reliability of the data, providing a stable and trustworthy data source for subsequent medical data management operations.

[0072] In detail, denoising activation features refer to specific attributes or conditions that can trigger or activate denoising mechanisms during data denoising, thereby enabling effective filtering and processing of data. These features are usually closely related to the physiological state of the target living organism, the nature of the regulatory task, and the noise characteristics of the data. For example, in implantable neural monitoring and regulation, if the heart rate of the target living organism suddenly increases (e.g., exceeding the normal value by 20%), this change may be considered a denoising activation feature because it may indicate that the living organism is in a state of stress, at which point the noise level in the data may increase accordingly, requiring intervention from a denoising mechanism.

[0073] Noise reduction pause features refer to specific attributes or conditions under which denoising is not suitable. The existence of these features can prevent errors and data loss during the denoising process. For example, in implantable neural monitoring and modulation, if the target organism's body temperature suddenly drops (e.g., below 2°C), this change might be considered a noise reduction pause feature. Because a sharp drop in body temperature may indicate an abnormal state, performing denoising at this time could inadvertently delete important physiological information; therefore, the denoising operation should be temporarily stopped.

[0074] Data optimization execution strategies are a series of specific operational steps or methods formulated based on denoising activation and denoising pause features to achieve the best denoising effect. These strategies typically include selecting a suitable denoising algorithm, adjusting denoising parameters, and determining the timing of denoising. For example, in implantable neural monitoring and modulation, a data optimization execution strategy might be: when the target living organism's heart rate increases by more than 20% (denoising activation feature), a wavelet transform denoising algorithm is activated, and the denoising intensity is set to medium (specific parameters such as a threshold of 0.5), and denoising processing continues until the heart rate returns to normal; while when the body temperature drops below 2°C below the normal value (denoising pause feature), all denoising operations are paused to avoid accidentally deleting important data.

[0075] Correlation maps are multi-dimensional data models used to describe the interactions and dependencies between different data optimization strategies. They provide a visual understanding of which strategies are mutually reinforcing and which may conflict. For example, in implantable neural monitoring and modulation, a correlation map might show that when the target organism's heart rate increases (a denoising activation feature), initiating wavelet transform denoising and increasing sampling frequency are mutually reinforcing because they both improve data accuracy and reliability. Conversely, simultaneously initiating overly strong denoising and decreasing sampling frequency may conflict because they could lead to the loss or distortion of important data. Analysis of correlation maps allows for the more scientific formulation and execution of data optimization strategies.

[0076] It is worth mentioning that the technical solutions described in steps 120 to 140 are the core invention of the control data denoising system. It describes in detail how to generate data optimization execution strategies with denoising activation features and denoising pause features based on target liveness monitoring information and spatiotemporal monitoring and control tasks, and further generate target denoising instruction text based on these strategies.

[0077] In step 120, the regulatory data denoising system generates several data optimization execution strategies with denoising activation and denoising pausing features based on the acquired target liveness monitoring information and several spatiotemporal monitoring and regulation tasks. The key to this step is that the regulatory data denoising system intelligently identifies when the denoising mechanism needs to be activated (denoising activation features) and when denoising processing should be paused (denoising pausing features) according to the real-time physiological state of the target liveness and the specific requirements of the regulation tasks. For example, when the target liveness shows abnormal fluctuations in heart rate or EEG signals, the regulatory data denoising system may identify this as a denoising activation feature and generate a corresponding denoising strategy; while when the liveness is in deep sleep or resting, the regulatory data denoising system may identify this as a denoising pausing feature to avoid misprocessing normal physiological signals.

[0078] In step 130, the data denoising system optimizes the execution strategies based on the generated data and their corresponding denoising activation and denoising cessation features, determining the correlation map between these strategies. The purpose of this step is to reveal the interactions and dependencies between different denoising strategies, in order to better understand and optimize the entire denoising process. For example, the data denoising system may find that some denoising strategies are mutually reinforcing under certain conditions and can be used simultaneously to improve the denoising effect; while other strategies may have conflicting or competitive relationships, requiring trade-offs and choices when using them.

[0079] In step 140, the data denoising system generates target denoising instruction text based on the correlation map, using the target liveness monitoring information as the denoising benchmark, and covering various spatiotemporal monitoring and control tasks. This step transforms the preceding analysis results into specific, executable denoising instructions. The target denoising instruction text details how to select and execute denoising strategies under different spatiotemporal monitoring and control tasks to achieve the best denoising effect. For example, the text might indicate which denoising algorithm should be prioritized under a specific time period and physiological state, or which denoising parameters need to be adjusted to obtain better data quality. Through such instruction text, the data denoising system can provide clear guidance to edge-side data optimization equipment, ensuring the accuracy and effectiveness of the data denoising process.

[0080] For example, the target denoising instruction text sent to the edge-side data optimization device can be as follows.

[0081] 1. Specific noise reduction strategies and execution conditions

[0082] 1.1 Denoising Strategy 1: Adaptive Denoising Based on Wavelet Transform

[0083] 1.1.1 Execution conditions:

[0084] The target living subject's heart rate exceeds 20% of the normal value for more than 5 minutes.

[0085] Meanwhile, the target living subject is not in a deep sleep state (judged based on EEG signals).

[0086] 1.1.2 Operating Procedures:

[0087] Start the wavelet transform denoising algorithm.

[0088] The noise reduction intensity is dynamically adjusted based on real-time heart rate data, with an initial threshold set to 0.5.

[0089] Continue noise reduction processing until the heart rate returns to normal or the noise reduction effect reaches the preset standard.

[0090] 1.2 Denoising Strategy Two: Targeted Filtering Based on Frequency Domain Characteristics

[0091] 1.2.1 Execution conditions:

[0092] The target living organism's EEG signals showed abnormal fluctuations in specific frequency bands (such as alpha waves and beta waves), with the fluctuation amplitude exceeding 30% of the normal range.

[0093] The target living organism is either awake or in a light sleep state.

[0094] 1.2.2 Operating Procedures:

[0095] Frequency domain analysis of EEG signals was performed to determine the specific frequency bands of abnormal fluctuations.

[0096] Targeted filtering methods are applied to denoise this frequency band.

[0097] Adjust the filtering parameters as needed based on the noise reduction effect until abnormal fluctuations are eliminated or reduced to the normal range.

[0098] 1.3 Noise Reduction Strategy Three: Noise Reduction Pause Based on Physiological State Judgment

[0099] 1.3.1 Execution conditions:

[0100] The target living organism's body temperature is 2°C below normal and lasts for more than 10 minutes.

[0101] Alternatively, the target living organism is in a deep sleep state, and its physiological signals (such as heart rate and respiration) are stable without abnormal fluctuations.

[0102] 1.3.2 Operating Procedures:

[0103] Immediately pause all noise reduction operations.

[0104] Record the physiological state and noise reduction progress during pauses.

[0105] Noise reduction treatment should be considered after the physiological state returns to normal or after deep sleep ends.

[0106] 2. Strategy relevance and execution order

[0107] 2.1 Strategy Relevance Analysis:

[0108] Denoising strategy one and strategy two may meet the execution conditions at the same time under certain conditions. In this case, priority should be given to judging based on the main physiological abnormalities of the target living organism.

[0109] Denoising strategy three has the highest priority. Once the execution conditions are met, all other denoising operations should be paused immediately to avoid unnecessary interference or risk to the target living organism.

[0110] 2.2 Execution sequence guidance:

[0111] 2.2.1 When both denoising strategy one and strategy two meet the execution conditions:

[0112] Prioritize strategy one, as abnormal heart rate may more directly reflect changes in the physiological state of the target living organism.

[0113] Once the heart rate returns to normal, assess whether strategy two needs to be implemented.

[0114] 2.2.2 When the denoising strategy three meets the execution conditions:

[0115] Immediately suspend the execution of all other noise reduction strategies.

[0116] Record the current physiological state and noise reduction progress for reference during subsequent recovery processing.

[0117] 2.3 Strategies for Handling Special Circumstances:

[0118] If the physiological state of the target living organism changes drastically during the execution of the denoising strategy (such as a sudden drop in heart rate or a sharp rise in body temperature), the current denoising operation should be stopped immediately, and emergency medical intervention should be considered.

[0119] If the data quality does not meet the expected standards after implementing the denoising strategy, the selection and implementation conditions of the denoising strategy should be re-evaluated, and the strategy should be adjusted or professional assistance should be sought if necessary.

[0120] Furthermore, it is important to note that the embodiments of this application do not pertain to disease diagnosis and treatment methods, but rather focus on data denoising and optimization at the data level. The core objective of these embodiments is to optimize implantable monitoring and control information, improving the accuracy and reliability of the data. It primarily focuses on the generation and execution of data denoising strategies, rather than directly diagnosing or treating diseases. Moreover, these embodiments optimize the monitoring information of the target living organism by regulating the data denoising system and utilizing algorithms and strategies, without involving the operation of any medical devices or the execution of any medical actions, nor directly affecting the human body. Furthermore, these embodiments are a data processing method, and their output (i.e., the denoised data) is only used as a reference for medical data management, medical analysis, or decision-making, and does not directly constitute a diagnosis or treatment. Finally, these embodiments are widely used in the field of medical data management, particularly in implantable monitoring and control systems, to improve the accuracy and reliability of data. Its value lies in providing a stable and reliable data source for medical data management.

[0121] In summary, this application addresses the challenges of applying implantable neural monitoring and control technology in modern medicine, particularly the data noise problem caused by the complexity and variability of the biological internal environment and various interference factors during the monitoring and control process. It proposes an innovative technical solution: by obtaining target in vivo monitoring information and several overlapping spatiotemporal monitoring and control tasks from the implantable monitoring and control information pool as a starting point, it achieves comprehensive optimization of data denoising processing.

[0122] Specifically, firstly, by generating several data optimization execution strategies with denoising activation and denoising pausing features, the system can intelligently identify and address data noise characteristics under different spatiotemporal monitoring and control tasks, thereby improving the targeting and effectiveness of denoising processing. Secondly, the correlation map between these data optimization execution strategies was further determined, making the selection and execution of denoising strategies more scientific and reasonable, avoiding potential conflicts and misoperations during the denoising process. Finally, a target denoising instruction text was generated, using the target liveness monitoring information as the denoising benchmark and involving various spatiotemporal monitoring and control tasks. This highly customized data optimization scheme was then distributed to edge-side data optimization devices, achieving precise preprocessing and optimization of the raw data, significantly improving the accuracy and reliability of the data. Thus, this not only effectively solves the problem of data noise in implantable neural monitoring and control technology but also provides a stable and reliable data source for subsequent medical data management and analysis.

[0123] In some optional embodiments, the step of generating several data optimization execution strategies with denoising activation features and denoising pause features based on the target liveness monitoring information and the several spatiotemporal monitoring and control tasks includes: mapping the several spatiotemporal monitoring and control tasks into several first task knowledge graphs containing denoising activation features and denoising pause features according to the monitoring expectation mode of the spatiotemporal monitoring and control tasks; mapping the target liveness monitoring information into a second task knowledge graph containing the same denoising activation features and denoising pause features; and generating the several data optimization execution strategies based on the several first task knowledge graphs and the second task knowledge graphs.

[0124] It is understandable that, in order to generate several data optimization execution strategies with denoising activation and denoising pause features based on the target liveness monitoring information and the several spatiotemporal monitoring and control tasks, the system first needs to map the several spatiotemporal monitoring and control tasks into several first task knowledge graphs containing denoising activation and denoising pause features, according to the monitoring expectation pattern of the spatiotemporal monitoring and control tasks. This step is crucial because it provides the system with a structured framework for understanding and handling the specific needs and constraints in different spatiotemporal monitoring and control tasks.

[0125] When constructing the knowledge graph for the first task, the system conducts in-depth analysis of each spatiotemporal monitoring and control task, identifying its unique monitoring expectation patterns. These patterns may include specific physiological indicator monitoring, specific time windows, specific spatial locations, etc. Based on these expectation patterns, the system further extracts key features related to denoising, namely denoising activation features and denoising cessation features. Denoising activation features refer to those features that indicate whether the system should activate the denoising mechanism under the current spatiotemporal conditions. They are usually closely related to the physiological state of the target living organism, the nature of the control task, and the noise characteristics of the data. Denoising cessation features, on the other hand, refer to those features that are not suitable for denoising under the current conditions. Their existence can avoid errors and data loss during the denoising process.

[0126] After constructing the first task knowledge graph, the system next needs to map the target liveness monitoring information into a second task knowledge graph containing the same denoising activation and denoising pause features. The purpose of this step is to align the real-time monitoring information of the target liveness with the expected patterns of the spatiotemporal monitoring and control tasks, in order to generate more accurate and effective data optimization execution strategies. When constructing the second task knowledge graph, the system utilizes advanced machine learning algorithms and pattern recognition technology to extract key features related to denoising from the target liveness monitoring information and match and map them with the features in the first task knowledge graph.

[0127] Once several first-task knowledge graphs and second-task knowledge graphs are available, the system can generate several data optimization execution strategies based on these two graphs. This step is the core of the entire process, involving complex strategy optimization and decision-making. The system will perform comprehensive analysis and reasoning based on the monitoring expectation patterns and denoising features in the first-task knowledge graph, and the real-time monitoring information and denoising features in the second-task knowledge graph, to determine which denoising strategies should be adopted under the current spatiotemporal conditions to achieve the best denoising effect.

[0128] Specifically, the system evaluates the potential impact of different denoising strategies on improving data accuracy and reliability, and considers the feasibility and cost of strategy implementation. In this process, the system may utilize advanced optimization algorithms, such as genetic algorithms and particle swarm optimization, to search for and determine the optimal data optimization execution strategy. Simultaneously, the system also considers the interactions and dependencies between different strategies to ensure that the generated data optimization execution strategies are mutually reinforcing rather than conflicting.

[0129] When generating data optimization execution strategies, the system pays special attention to the application of denoising activation and denoising pause features. For denoising activation features, the system ensures that the denoising mechanism is activated promptly when specific conditions are met to effectively filter and process the data. For denoising pause features, the system ensures that denoising operations are paused promptly when current conditions are unsuitable to avoid misoperation and data loss.

[0130] When discussing specific examples of the first and second task knowledge graphs, a scenario can be constructed based on the authorized access data to illustrate these two concepts.

[0131] First Task Knowledge Graph

[0132] Definition: The first task knowledge graph is constructed for several spatiotemporal monitoring and control tasks. It includes the monitoring expectation patterns, key features (including denoised activation features and denoised pause features) of these tasks, and the relationships between them.

[0133] Example: In the field of neural monitoring and regulation, three patients participated in different spatiotemporal monitoring and regulation tasks.

[0134] Task MA: Real-time monitoring of cortical activity in patient A to assess responses to specific stimuli. The expected monitoring pattern for this task is to capture changes in electrical signals in the cortex before and after stimulation. Denoising activation features may include sudden changes in signal intensity or the appearance of specific frequency components, while denoising cessation features may be small fluctuations in signal around a stable baseline.

[0135] Task MB: Long-term monitoring of spinal nerve conduction in patient B to study the process of nerve regeneration. The expected monitoring pattern for this task is to track changes in nerve conduction velocity. Denoising activation features may be significant increases or decreases in conduction velocity, while denoising cessation features may be due to transient signal interference caused by the patient's daily activities.

[0136] Task MC: Periodically monitor the autonomic nervous system function of patient C to assess autonomic regulation ability. The expected monitoring pattern for this task is to analyze indicators such as heart rate variability. Denoising activation features may indicate abnormal increases or decreases in heart rate variability, while denoising cessation features may indicate normal physiological rhythm fluctuations.

[0137] Integrating the expected monitoring patterns, denoising activation features, and denoising pause features of these tasks, along with the relationships between them, forms the first task knowledge graph. This graph helps the system understand the needs and constraints of different monitoring tasks, providing a foundation for subsequent data optimization execution strategies.

[0138] Second Task Knowledge Graph

[0139] Definition: The second task knowledge graph is constructed based on the target liveness monitoring information. It also includes denoised activation features and denoised pause features, but these features need to match and correspond to the features in the first task knowledge graph.

[0140] Example: Continuing with the medical scenario above, the system is receiving real-time monitoring data on the cerebral cortex activity of patient A. This data constitutes the target in vivo monitoring information. To denoise this data, the system needs to construct a second task knowledge graph, where the denoised activation features and denoised pausing features should match the features of task MA in the first task knowledge graph.

[0141] For example, in the second task knowledge graph, denoising activation features may include identifying patterns similar to sudden changes in signal strength or the appearance of specific frequency components as defined in task MA; while denoising pausing features may be used to exclude signal interference that is similar to small fluctuations near a stable baseline defined in task MA.

[0142] Through this matching and correspondence process, the system can perform precise noise reduction on the target liveness monitoring information based on the second task knowledge graph, thereby improving the accuracy and reliability of the data.

[0143] Therefore, the first and second task knowledge graphs represent different levels of knowledge representation in the medical context: the former focuses on the needs and constraints of the monitoring task itself; the latter focuses on how to apply these needs and constraints to specific target in vivo monitoring information. The collaborative work of these two graphs provides strong support for denoising implantable neural monitoring and modulation data.

[0144] It is evident that by constructing the first and second task knowledge graphs and using them as the basis for generating data optimization execution strategies, the system can more accurately and effectively handle noise issues in the data collected by the implantable monitoring and control system, thereby improving the accuracy and reliability of the data and providing a more stable and reliable data source for subsequent medical data management and analysis.

[0145] In some other preferred embodiments, determining the correlation map corresponding to the plurality of data optimization execution strategies based on the denoising activation features and denoising pausing features corresponding to the plurality of data optimization execution strategies includes: obtaining the denoising activation features and denoising pausing features corresponding to each data optimization execution strategy; determining potential noisy image descriptors based on the denoising activation features and denoising pausing features corresponding to each data optimization execution strategy; determining the correlation degree between the first strategy identifier of each data optimization execution strategy and the second strategy identifier of the remaining data optimization execution strategies based on the denoising activation features and denoising pausing features corresponding to each data optimization execution strategy; and mapping the potential noisy image descriptors to the correlation map according to the determined correlation degree.

[0146] It is understandable that determining the correlation maps corresponding to the aforementioned data optimization execution strategies based on their respective denoising activation and denoising pausing features is a complex process involving feature extraction, correlation analysis, and map construction. This process aims to provide strong support for subsequent data optimization by deeply analyzing the intrinsic relationships between the data optimization execution strategies.

[0147] First, the system needs to acquire the denoising activation features and denoising pause features corresponding to each data optimization execution strategy. These features are extracted from the previously constructed first and second task knowledge graphs, representing the unique behaviors and constraints of different strategies when processing noise. The denoising activation features are typically related to the conditions that activate the denoising mechanism in the strategy, while the denoising pause features are related to the conditions that pause the denoising operation. By extracting these features, the system can quantify and describe the denoising behavior of different strategies.

[0148] Next, based on the denoising activation features and denoising pause features corresponding to each data optimization execution strategy, the system needs to determine latent noise image descriptors. A latent noise image descriptor is an abstract representation used to describe noise patterns that may occur under a specific strategy. These descriptors are obtained by analyzing the interaction between denoising features in the strategy and target liveness monitoring information. They can help the system understand the potential effects and limitations of different strategies in processing noise.

[0149] To construct latent noise image descriptors, the system may employ advanced machine learning algorithms, such as deep learning networks or clustering algorithms. These algorithms can learn noise patterns and features from large amounts of monitoring data and abstract them into descriptor forms. In this way, the system can obtain a series of latent noise image descriptors associated with different strategies, supporting subsequent analysis and decision-making.

[0150] After obtaining the latent noise image descriptors, the system needs to further determine the correlation between the first strategy identifier of each data optimization execution strategy and the second strategy identifier of the remaining data optimization execution strategies. Correlation is a quantitative metric used to measure the similarity and complementarity of different strategies in denoising behavior. By calculating the correlation, the system can discover which strategies have similar effects in handling noise, which strategies may be complementary, and which strategies may conflict with each other.

[0151] To calculate the correlation, the system may employ similarity metrics such as cosine similarity, Pearson correlation coefficient, or mutual information. These methods can quantitatively compare the denoised features between different strategies and obtain a numerical value representing the similarity. Through this approach, the system can obtain a correlation matrix, where each element represents the degree of similarity and complementarity between a pair of strategies.

[0152] Finally, based on the determined correlation, the system needs to map the latent noisy image descriptors into the correlation map. The correlation map is a visual representation that shows the inherent connections and interactions between different strategies in their denoising behavior. In the correlation map, each node represents a data optimization execution strategy, and the edges between nodes represent the correlation between the strategies. The weights and colors of the edges can be used to represent the strength and directionality of the correlation.

[0153] By constructing correlation maps, the system can visually demonstrate the similarities and complementarities between different strategies, as well as their potential effects and limitations in noise processing. This helps physicians or researchers better understand the behavior and effects of different strategies and provides strong support for subsequent data optimization. For example, they can select strategies with similar effects for combined use based on the correlation maps to improve the efficiency and accuracy of noise denoising; or they can choose complementary strategies for alternating use to address different types of noise.

[0154] For example, regarding latently noisy image descriptors, there are three data optimization execution strategies A, B, and C, each corresponding to a set of denoising activation features and denoising pausing features. Furthermore, each strategy may have only one primary denoising activation feature and one primary denoising pausing feature, both represented numerically.

[0155] Strategy A: Denoising activation feature = 0.8, Denoising pause feature = 0.2;

[0156] Strategy B: Denoising activation feature = 0.6, Denoising pause feature = 0.4;

[0157] Strategy C: Denoising activation feature = 0.9, Denoising pause feature = 0.1.

[0158] Latent noise image descriptors are the result of further abstraction and quantization of these features. In this exemplary scenario, the latent noise image descriptor can be defined as the difference between the denoising activation features and the denoising pause features, i.e.:

[0159] The potential noisy image descriptor for strategy A is 0.8 - 0.2 = 0.6;

[0160] The potential noisy image descriptor for strategy B is 0.6 - 0.4 = 0.2;

[0161] The potential noisy image descriptor for strategy C is 0.9 - 0.1 = 0.8.

[0162] These descriptors reflect the potential effectiveness and tendency of different strategies in handling noise. For example, strategy C has the highest denoising activation feature and the lowest denoising pausing feature, thus its potential noisy image descriptor is the highest, indicating that it may have the strongest effect in handling noise.

[0163] Next, the correlation between strategies is calculated based on these latent noisy image descriptors, and a correlation map is constructed. The correlation can be obtained by calculating the similarity between descriptors. In this exemplary scenario, the simple absolute difference is used as the measure of similarity:

[0164] The correlation between strategy A and strategy B is |0.6 - 0.2| = 0.4;

[0165] The correlation between strategy A and strategy C is |0.6 - 0.8| = 0.2;

[0166] The correlation between strategy B and strategy C is |0.2 - 0.8| = 0.6.

[0167] Based on these correlations, a correlation graph can be constructed. In this graph, each policy is a node, and the edges between nodes represent the correlation between policies. The weight and color of the edges can be used to represent the strength and direction of the correlation. For example, different colors can be used to represent different ranges of correlation, and the thickness of the edges can be used to represent the strength of the correlation.

[0168] In this exemplary scenario, the correlation map might look like this:

[0169] There is a medium-thickness blue border between strategy A and strategy B, indicating that their correlation is 0.4;

[0170] There is a thin blue border between strategy A and strategy C, indicating that their correlation is 0.2;

[0171] There is a thick red edge between strategy B and strategy C, indicating that their correlation is 0.6.

[0172] This correlation map visually illustrates the similarities and complementarities between different strategies. For example, the high correlation between strategy B and strategy C suggests they may have similar effects or complementary behaviors when dealing with noise. Conversely, the low correlation between strategy A and strategy C indicates they may have significant differences in noise handling.

[0173] Therefore, this embodiment can provide in-depth analysis of the intrinsic connections and interactions between different strategies, and provide strong support for subsequent optimization of regulatory data.

[0174] In some alternative embodiments, obtaining the denoising activation features and denoising pause features corresponding to each data optimization execution strategy includes: obtaining, from the plurality of data optimization execution strategies, a target data optimization execution strategy that is related to the historical execution strategies in the implanted monitoring and control information pool in the monitoring and control dimension; and determining the denoising activation features and denoising pause features corresponding to the target data optimization execution strategy based on the denoising activation features and denoising pause features of the historical execution strategies.

[0175] In practical applications, this embodiment aims to identify target strategies that are related to specific historical execution strategies in terms of monitoring and control from a large number of data optimization execution strategies, and further determine the denoising activation characteristics and denoising pause characteristics of these target strategies.

[0176] Among them, historical execution strategies refer to strategies that have been executed in past monitoring and control tasks, which include experience and knowledge accumulated in practical applications; monitoring and control dimensions refer to multiple aspects used to describe and evaluate monitoring and control tasks, such as the nature of the monitoring target, the conditions of the monitoring environment, and the characteristics of the monitoring data; target data optimization execution strategies refer to strategies that have some connection or similarity with historical execution strategies in the monitoring and control dimensions, and these may be strategies that need to be considered or selected in the current task.

[0177] Based on this, from several data optimization execution strategies, the regulatory data denoising system needs to select target data optimization execution strategies that are related to historical execution strategies in the embedded monitoring and regulation information pool in terms of monitoring and regulation dimensions. This process is a strategy screening process, which requires the regulatory data denoising system to conduct in-depth analysis and comparison of existing data optimization execution strategies to identify those strategies that are similar to or related to historical execution strategies in specific dimensions.

[0178] To achieve this goal, data denoising systems can employ various methods. For example, they can utilize classification or clustering algorithms from machine learning to automatically classify or cluster data optimization execution strategies, identifying strategies with similar characteristics. Specifically, a data denoising system can use historical execution strategies as training samples, extracting features from their monitoring and control dimensions, and training a classifier or clusterer. Then, the system can use the remaining data optimization execution strategies as input, employing the trained model for prediction or clustering to find strategies similar to the target historical execution strategy.

[0179] Alternatively, data denoising systems can employ rule-based methods for strategy selection. For instance, a data denoising system can define a set of rules or conditions to describe the characteristics or constraints of historical execution strategies across the monitoring and control dimensions. Then, the system can iterate through all data optimization execution strategies, checking whether they satisfy these rules or conditions, thereby selecting strategies that are related to the target historical execution strategy.

[0180] Once the data denoising system has identified the target data optimization strategies, the next step is to determine the corresponding denoising activation and denoising cessation features for these target strategies based on the denoising activation and denoising cessation features of historical execution strategies. This process is a feature mapping and feature determination process, requiring the data denoising system to apply the feature knowledge of historical execution strategies to the target strategies.

[0181] To achieve this goal, various methods can be employed in the data denoising system. One direct approach is that if the historical execution strategy and the target strategy are very similar or almost identical in terms of monitoring and control dimensions, the data denoising system can directly assign the denoising activation and denoising pause features of the historical execution strategy to the target strategy. This method is suitable for historical strategies that are very close to the target strategy and can ensure the accuracy and effectiveness of feature mapping.

[0182] However, in many cases, there may be differences or changes between the historical execution strategy and the target strategy. In such cases, the data denoising system needs to employ more complex methods for feature mapping and determination. For example, the data denoising system can utilize regression or interpolation algorithms from machine learning to predict or calculate the features of the target strategy based on the characteristics of the historical execution strategy and the differences between the target strategy. Specifically, the data denoising system can use the features of the historical execution strategy as input and the differences in the target strategy as output to train a regression or interpolation model. Then, the data denoising system can use this model to predict or calculate the denoised activation features and denoised pausing features of the target strategy.

[0183] Alternatively, data denoising systems can employ rule-based or template-based methods for feature mapping and determination. For example, a data denoising system can define a set of rules or templates to describe how to derive the characteristics of the target policy from the characteristics of historical execution policies. These rules or templates can be developed based on domain knowledge, experience, or expert opinions. The data denoising system can then apply these rules or templates to derive the denoised activation and denoised pausing features of the target policy.

[0184] Therefore, obtaining the denoising activation and denoising pause features corresponding to each data optimization execution strategy is a comprehensive process involving strategy selection, feature mapping, and feature determination. This process requires the data denoising system to conduct in-depth analysis and comparison of existing data optimization execution strategies to identify strategies similar to the target's historical execution strategies, and further apply the feature knowledge of historical strategies to the target strategy. In this way, the data denoising system can provide accurate and effective strategy feature information for subsequent data optimization, thereby supporting more precise data denoising optimization.

[0185] In other possible embodiments, obtaining the denoising activation features and denoising pause features corresponding to each data optimization execution strategy includes: determining the target feature space by referring to the topological labels corresponding to the target liveness monitoring information in the plurality of data optimization execution strategies; obtaining the policy node variables of each data optimization execution strategy other than the topological labels corresponding to the target liveness monitoring information; and determining the denoising activation features and denoising pause features of each data optimization execution strategy in the target feature space based on the policy node variables of each data optimization execution strategy.

[0186] It is important to note that the process of obtaining the denoising activation features and denoising pause features corresponding to each data optimization execution strategy is a comprehensive process involving topological label reference, target feature space determination, strategy node variable acquisition, and feature determination. The aim is to extract features related to the target liveness monitoring information from numerous data optimization execution strategies, thereby providing accurate strategy feature information for subsequent regulation data optimization.

[0187] In this embodiment, topology labels are identifiers used to describe and distinguish different data optimization execution strategies. They reflect the position or attributes of a strategy in a certain topology or relational network. The target feature space refers to the space composed of features related to the target liveness monitoring information. It is the space in which the data denoising system wants to extract and determine the denoising activation features and denoising cessation features. The strategy node variables refer to variables that describe the data optimization execution strategy in a specific dimension or aspect. These can be the strategy's parameters, configuration, state, etc.

[0188] Furthermore, using the topological labels corresponding to the target liveness monitoring information in several data optimization execution strategies as a reference, the data denoising system first needs to determine the target feature space. This process is a feature space localization process, which requires the data denoising system to find features related to the target liveness monitoring information from numerous data optimization execution strategies and determine the space constituted by these features.

[0189] To achieve this goal, various methods can be employed in the data denoising system. One possible approach is to utilize feature selection algorithms from machine learning to extract features relevant to the target liveness detection information from all data optimization execution strategies. Specifically, the system can use the target liveness detection information as a label or response variable and the features of the data optimization execution strategies as input variables to train a feature selection model. Then, the system can use this model to select the features most relevant to the target liveness detection information, thereby determining the target feature space.

[0190] Another possible approach is to allow the data denoising system to determine the target feature space based on domain knowledge or expert experience. For example, the data denoising system can select features that have a significant impact or relevance to the target liveness monitoring information based on its nature and requirements, thereby constructing the target feature space.

[0191] Once the data denoising system has determined the target feature space, the next step is to obtain the policy node variables for each data optimization execution strategy, excluding the topological labels corresponding to the target liveness monitoring information. This process is a policy variable extraction process, requiring the data denoising system to extract variables describing the strategy in specific dimensions or aspects from the remaining data optimization execution strategies.

[0192] To achieve this goal, various methods can be employed to regulate data denoising systems. One possible approach is for the system to directly access the storage or representation of the data optimization execution strategy and extract strategy node variables. These variables can be strategy parameters, configurations, states, etc., describing the strategy's characteristics or attributes in specific aspects.

[0193] Another possible approach is for the data denoising system to utilize data mining or information extraction techniques to extract policy node variables from relevant data sources. For example, the data denoising system can extract variables describing policy characteristics from documents, logs, or metadata related to data optimization execution strategies.

[0194] Once the data denoising system has acquired the policy node variables for each data optimization execution strategy, the next step is to determine the denoising activation and denoising pausing features of each strategy in the target feature space based on these variables. This process is a feature mapping and determination process, requiring the data denoising system to map the policy node variables to the target feature space and determine the corresponding denoising activation and denoising pausing features.

[0195] To achieve this goal, various methods can be employed in the data denoising system. One possible approach is to utilize regression or classification algorithms from machine learning to predict or classify the denoised activation and denoised stationary features in the target feature space based on policy node variables. Specifically, the system can train a regression or classification model by using the policy node variables as input and the denoised activation and stationary features as outputs. Then, the system can use this model to predict or classify the denoised activation and stationary features of each data optimization execution policy in the target feature space.

[0196] Another possible approach is for the data denoising system to determine denoising activation and denoising pausing features based on rules or templates. For example, the system could define a set of rules or templates describing how to compute or derive denoising activation and denoising pausing features in the target feature space based on policy node variables. These rules or templates could be developed based on domain knowledge, experience, or expert opinions. The system could then apply these rules or templates to determine the denoising activation and denoising pausing features of each data optimization execution policy in the target feature space.

[0197] To illustrate this process more concretely, consider a data denoising system with three data optimization execution strategies, A, B, and C, each corresponding to different topological labels and strategy node variables. The goal of the data denoising system is to determine the denoising activation and denoising pausing features of these three strategies in the target feature space.

[0198] First, the data denoising system uses the topological labels corresponding to the target liveness monitoring information as a reference to determine the target feature space as {X1, X2, X3}, where X1, X2, and X3 are features related to the target liveness monitoring information.

[0199] Then, the data denoising system obtains the policy node variables for policies A, B, and C. Assume the node variables for policy A are {V1=0.5, V2=0.7}, for policy B are {V1=0.6, V2=0.8}, and for policy C are {V1=0.4, V2=0.6}.

[0200] Finally, the data denoising system determines the denoising activation features and denoising pause features of each strategy in the target feature space based on the strategy node variables. Assuming the data denoising system uses a regression algorithm for feature mapping and determination, the following results are obtained: Strategy A has denoising activation features of {X1=0.6, X2=0.7, X3=0.5} and denoising pause features of {X1=0.4, X2=0.5, X3=0.3} in the target feature space; Strategy B has denoising activation features of {X1=0.7, X2=0.8, X3=0.6} and denoising pause features of {X1=0.5, X2=0.6, X3=0.4} in the target feature space; Strategy C has denoising activation features of {X1=0.5, X2=0.6, X3=0.4} in the target feature space and denoising pause features of {X1=0.3, X2=0.4, X3=0.2}.

[0201] This demonstrates that different data optimization strategies exhibit varying denoising activation and denoising pausing characteristics in the target feature space. These characteristics reflect the potential effectiveness and tendency of the strategies in handling noise, providing accurate strategy feature information for subsequent data optimization adjustments.

[0202] It is understandable that the process of obtaining the denoising activation features and denoising pause features corresponding to each data optimization execution strategy is a comprehensive process involving topological label reference, target feature space determination, strategy node variable acquisition, and feature mapping and determination. This process requires the regulatory data denoising system to extract features related to the target liveness monitoring information from numerous data optimization execution strategies and determine the denoising activation features and denoising pause features of these features in the target feature space. In this way, the regulatory data denoising system can provide accurate and effective strategy feature information for subsequent regulatory data optimization, thereby supporting more precise regulatory data denoising optimization.

[0203] In some alternative design approaches, mapping the latent noisy image descriptor to the correlation map based on the determined correlation degree includes: mining target denoising activation features and target denoising pause features from the latent noisy image descriptor; if the target denoising activation features and the target denoising pause features correspond to the same data optimization execution strategy, then determining the graph semantic unit corresponding to the target denoising activation features and the target denoising pause features as a preset semantic code; if the target denoising activation features and the target denoising pause features correspond to different data optimization execution strategies, then taking the cosine similarity between the target denoising pause features and the target denoising activation features from the determined correlation degree as the graph semantic unit corresponding to the target denoising activation features and the target denoising pause features, to obtain the correlation map.

[0204] Based on this design concept, the process of mapping latent noisy image descriptors into a correlation map according to the determined correlation degree is a comprehensive task involving feature mining, semantic unit determination, and map construction. This process aims to extract the target denoising activation features and target denoising pause features from the latent noisy image descriptors by analyzing and processing them, and then map them into a correlation map that reflects the relationships between these features based on the correlation degree between them.

[0205] First, controlling the data denoising system requires extracting target denoising activation features and target denoising pausing features from the latent noisy image descriptor. This is a feature extraction process that requires the control system to identify features related to denoising activation and denoising pausing from complex image descriptors. These features can be image texture, color, shape, etc., and they play an important indicative role in subsequent noise processing and image optimization.

[0206] After identifying the target denoising activation feature and the target denoising pausing feature, the data denoising system needs to further determine whether these two features correspond to the same data optimization execution strategy. If they correspond to the same data optimization execution strategy, the data denoising system can determine that the graph semantic unit corresponding to these two features is a preset semantic code. In this embodiment, the preset semantic code is a predefined code used to represent the specific relationship between the target denoising activation feature and the target denoising pausing feature under the same data optimization execution strategy. By assigning the same preset semantic code to these two features, the data denoising system can clearly represent their association in the correlation graph.

[0207] However, if the target denoising activation feature and the target denoising paused feature correspond to different data optimization execution strategies, then the data denoising system cannot simply assign them the same preset semantic encoding. In this case, the data denoising system needs to calculate the cosine similarity between the target denoising paused feature and the target denoising activation feature from the determined correlation. Cosine similarity is an index that measures the degree of similarity between two vectors; the closer its value is to 1, the more similar the two vectors are. By calculating the cosine similarity between the target denoising paused feature and the target denoising activation feature, the data denoising system can obtain a numerical value that reflects the degree of similarity between them.

[0208] After obtaining the cosine similarity, the modulating data denoising system can use this value as the corresponding spectral semantic unit for the target denoising activation feature and the target denoising paused feature. Thus, in the correlation map, the modulating data denoising system can represent the degree of similarity between these two features using spectral semantic units. When the modulating data denoising system plots these two features in the correlation map, it can adjust their positions in the map based on their cosine similarity, placing similar features closer together and dissimilar features further apart.

[0209] To illustrate the process more specifically, consider a data denoising system with two potential noisy image descriptors, Ua and Ub. From these descriptors, the system extracts target denoising activation features a1 and b1, as well as target denoising pause features a2 and b2. Further analysis reveals that features a1 and a2 correspond to the same data optimization execution strategy, while features b1 and b2 correspond to different data optimization execution strategies.

[0210] For features a1 and a2, since they correspond to the same data optimization execution strategy, the data denoising system can assign them the same preset semantic code, such as "001". In the correlation map, the data denoising system can plot features a1 and a2 in the same or similar positions and label them with "001".

[0211] For features b1 and b2, since they correspond to different data optimization execution strategies, the data denoising system needs to calculate their cosine similarity. Suppose that the cosine similarity between b1 and b2, calculated by the data denoising system, is 0.7. Then, in the correlation map, the data denoising system can plot features b1 and b2 at relatively far positions and label the connection or relationship between them with "0.7". In this way, by observing the correlation map, the data denoising system can clearly see the similarity between features b1 and b2 and their differences from features a1 and a2.

[0212] Through the above process, the modulating data denoising system can map the target denoising activation features and target denoising pause features in the latent noisy image descriptor into a correlation map. This map not only helps the modulating data denoising system intuitively understand the relationships between features, but also provides important reference information for subsequent noise processing and image optimization. For example, by observing the correlation map, the modulating data denoising system can identify which features have high similarity, thus allowing it to consider merging them or performing similar processing; conversely, it can also identify which features have low similarity, thus allowing it to consider differentiated processing or optimization.

[0213] In summary, mapping potentially noisy image descriptors to a correlation map based on the determined correlation degree is a crucial process. It requires the control data denoising system to extract useful feature information from complex image descriptors and construct a correlation map reflecting the relationships between these features. This process not only helps the control data denoising system better understand the feature information in the image descriptors but also provides strong support for subsequent noise processing and image optimization.

[0214] In some other possible embodiments, the step of acquiring the target liveness monitoring information and several spatiotemporal monitoring and control tasks belonging to the implantable monitoring and control information pool includes: acquiring historical execution strategies created by a denoising strategy planning system in the implantable monitoring and control information pool; taking the execution strategies in the implantable monitoring and control information pool after removing the historical execution strategies as pending execution strategies; decomposing the implantable monitoring and control information pool into several monitoring and control information sets based on the information involvement weight of the pending execution strategies; determining the several spatiotemporal monitoring and control tasks according to the pending execution strategies in the target monitoring and control information set; and using the target monitoring and control information set to determine the target liveness monitoring information.

[0215] It can be understood that acquiring target liveness monitoring information from the implanted monitoring and control information pool, and several spatiotemporal monitoring and control tasks belonging to this pool, is a comprehensive process involving data processing and task planning. This process aims to extract useful information from the complex implanted monitoring and control information pool and, based on this information, plan a series of specific spatiotemporal monitoring and control tasks.

[0216] First, the regulatory data denoising system needs to acquire historical execution strategies created by the denoising strategy planning system from the embedded monitoring and control information pool. This process is a data retrieval process, requiring the regulatory data denoising system to extract data from the embedded monitoring and control information pool that has already undergone denoising and has been identified by the system as historical execution strategies. These historical execution strategies are important information that the regulatory data denoising system needs to refer to and rely on in subsequent processing; they represent the system's past experience and knowledge in processing embedded monitoring and control information.

[0217] After acquiring historical execution strategies, the regulatory data denoising system needs to remove those historical strategies from the embedded monitoring and regulation information pool as pending execution strategies. This process is a data filtering process, requiring the regulatory data denoising system to remove processed historical execution strategies from the embedded monitoring and regulation information pool, retaining those that have not yet been processed or require further processing. These pending execution strategies are the focus of the regulatory data denoising system in subsequent processing.

[0218] Next, the control data denoising system needs to decompose the embedded monitoring and control information pool into several monitoring and control information sets based on the information involvement weight of the execution strategy to be processed. This process is a data decomposition and classification process, requiring the control data denoising system to decompose the data in the embedded monitoring and control information pool into several relatively independent and somewhat related monitoring and control information sets according to the information involvement weight of the execution strategy to be processed. The information involvement weight is an indicator that measures the degree of involvement or importance of the execution strategy in the information pool. The larger the value, the higher the degree of involvement or importance of the execution strategy in the information pool. By considering the information involvement weight, the control data denoising system can ensure that, during the decomposition process, execution strategies with high involvement or importance are assigned to the same monitoring and control information set for subsequent processing and analysis.

[0219] After breaking down the data into several monitoring and control information sets, the control data denoising system needs to determine several spatiotemporal monitoring and control tasks based on the execution strategies to be processed within the target monitoring and control information sets. This process is a task planning and generation process, requiring the control data denoising system to plan a series of specific spatiotemporal monitoring and control tasks according to the execution strategies to be processed within the target monitoring and control information sets. These tasks are the specific operations or instructions that need to be executed in subsequent actual operations, and they aim to achieve the effective acquisition and processing of target liveness monitoring information. When determining spatiotemporal monitoring and control tasks, the control data denoising system needs to consider the spatiotemporal characteristics of the tasks, that is, the tasks need to be executed within a specific time and space range. By considering the spatiotemporal characteristics, the control data denoising system can ensure that the planned tasks can be effectively executed in actual operations.

[0220] Finally, the regulatory data denoising system needs to utilize the target monitoring and regulation information set to determine the target liveness monitoring information. This process is one of information extraction and determination, requiring the regulatory data denoising system to extract the target liveness monitoring information based on the data and information in the target monitoring and regulation information set. The target liveness monitoring information is the final result that the regulatory data denoising system needs to obtain in this process; it represents useful information related to the target liveness within the implantable monitoring and regulation information pool. By extracting the target liveness monitoring information, the regulatory data denoising system can effectively utilize and process implantable monitoring and regulation information, providing strong support for subsequent medical diagnosis, treatment, or health management.

[0221] For example, an embedded monitoring and control information pool contains 1000 execution strategies, of which 500 are historical execution strategies created by a denoising strategy planning system, and the remaining 500 are execution strategies to be processed. The control data denoising system can decompose the embedded monitoring and control information pool into 10 monitoring and control information sets, each containing 50 execution strategies, based on the information involvement weights of these 500 pending execution strategies. Then, the control data denoising system can plan 10 spatiotemporal monitoring and control tasks based on the pending execution strategies within these 10 monitoring and control information sets, with each task processing a specific monitoring and control information set. Finally, the control data denoising system can extract monitoring information related to the target liveness from these 10 monitoring and control information sets as the final target liveness monitoring information.

[0222] Through the above process, the control data denoising system can effectively acquire and process target liveness monitoring information and several spatiotemporal monitoring and control tasks from the implanted monitoring and control information pool. This process not only considers the involvement weight and spatiotemporal characteristics of the information, but also ensures that the planned tasks can be effectively executed in actual operation. Simultaneously, by extracting target liveness monitoring information, the control data denoising system can provide strong support for subsequent data denoising optimization.

[0223] In some other possible embodiments, the step of acquiring the target live monitoring information in the implantable monitoring and control information pool and a plurality of spatiotemporal monitoring and control tasks belonging to the implantable monitoring and control information pool includes: performing knowledge mining on the live physiological attribute data corresponding to the implantable monitoring and control information pool to obtain a live physiological attribute knowledge relationship network; based on the live physiological attribute knowledge relationship network, determining the priority task event among each task event of the monitoring and control tasks in the implantable monitoring and control information pool; if the priority task event does not have an associated priority task event, then the monitoring and control task corresponding to the priority task event is taken as the spatiotemporal monitoring and control task; if the priority task event has an associated priority task event that does not belong to the same data optimization execution strategy, and the correlation degree between the priority task event and the associated priority task event is less than or equal to a set correlation degree, then the monitoring and control task corresponding to the priority task event is taken as the spatiotemporal monitoring and control task.

[0224] It is understood that this embodiment requires the data denoising system to not only deeply understand the structure and content of the embedded monitoring and control information pool, but also to accurately identify and extract key information to support subsequent data denoising processing. The following is a detailed description of this technical solution:

[0225] First, the regulatory data denoising system needs to perform knowledge mining on the live physiological attribute data corresponding to the implanted monitoring and regulation information pool to obtain a knowledge network of live physiological attributes. This knowledge network is a complex network containing various knowledge about live physiological attributes and the relationships between them. By performing knowledge mining on the live physiological attribute data, the regulatory data denoising system can reveal the implicit knowledge and patterns behind this data, providing strong support for subsequent monitoring and regulation tasks. In this process, the regulatory data denoising system may use various data mining and knowledge discovery techniques and methods, such as association rule mining, cluster analysis, and classification prediction, to extract useful information and knowledge from the live physiological attribute data.

[0226] Secondly, based on the knowledge network of in vivo physiological attributes, the regulatory data denoising system can identify priority task events among the various task events of the monitoring and regulation tasks within the implantable monitoring and regulation information pool. Priority task events refer to events with high priority and importance in the monitoring and regulation tasks. These events typically require priority processing and execution to ensure the smooth progress and completion of the monitoring and regulation tasks. In determining priority task events, the regulatory data denoising system needs to comprehensively consider multiple factors, such as the urgency, importance, and correlation with other events. By comprehensively considering these factors, the regulatory data denoising system can accurately identify priority task events in the implantable monitoring and regulation information pool.

[0227] Then, the data denoising system needs to further judge and process the priority task events. If a priority task event has no associated priority task events, the data denoising system can directly treat the corresponding monitoring and control task as a spatiotemporal domain monitoring and control task. This is because if a priority task event is not associated with other priority task events, then its corresponding monitoring and control task is a relatively independent task that can be directly executed as a spatiotemporal domain monitoring and control task.

[0228] However, if a priority task event has associated priority task events that do not belong to the same data optimization execution strategy, and the correlation between the priority task event and its associated priority task events is less than or equal to a set correlation degree, then the data denoising system also needs to treat the monitoring and control task corresponding to the priority task event as a spatiotemporal domain monitoring and control task. This is because, although the priority task event is related to other priority task events, this relationship is not strong, or they do not belong to the same data optimization execution strategy. Therefore, the data denoising system can execute the monitoring and control task corresponding to the priority task event as an independent spatiotemporal domain monitoring and control task.

[0229] In this process, setting the correlation degree is an important parameter, used to measure the degree of correlation between priority task events and their associated priority task events. The specific value of the correlation degree can be set and adjusted according to the actual application scenario and requirements. For example, in some application scenarios, the correlation degree of the data denoising system can be set to 0.5, meaning that if the correlation degree between a priority task event and its associated priority task event is less than or equal to 0.5, then the monitoring and control task corresponding to the priority task event will be treated as a spatiotemporal domain monitoring and control task.

[0230] Through the above process, the regulatory data denoising system can obtain target live monitoring information and several spatiotemporal monitoring and regulatory tasks from the implanted monitoring and regulatory information pool. These spatiotemporal monitoring and regulatory tasks are obtained based on the knowledge network of live physiological attributes and the judgment and processing of priority task events. They have clear spatiotemporal characteristics and priorities, which can provide strong support and guidance for subsequent regulatory data denoising processing.

[0231] In other possible embodiments, the step of generating target denoising instruction text based on the correlation map, using the target liveness monitoring information as the denoising benchmark and involving various spatiotemporal domain monitoring and control tasks, includes: determining the task chain characteristics of each spatiotemporal domain monitoring and control task based on the correlation map and a long short-term memory model; and generating target denoising instruction text with the highest timeliness weight based on the target liveness monitoring information as the denoising benchmark and the task chain characteristics of each spatiotemporal domain monitoring and control task.

[0232] In this embodiment, the regulatory data denoising system needs to determine the task chain features of each spatiotemporal monitoring and regulation task based on correlation maps and long short-term memory (LSM) models. This process is a feature extraction process, requiring the regulatory data denoising system to extract key information related to each spatiotemporal monitoring and regulation task from the correlation maps and to model and analyze this information using LSM models. LSM models are a special type of recurrent neural network that can effectively process sequential data and capture long-term dependencies in the data. By utilizing LSM models, the regulatory data denoising system can accurately extract and represent the task chain features of each spatiotemporal monitoring and regulation task.

[0233] When extracting task chain features, the regulatory data denoising system needs to consider the correlation and temporal sequence between various spatiotemporal monitoring and regulation tasks. Correlation refers to the potential mutual influence and dependencies between different tasks, while temporal sequence refers to the possible order and time intervals between tasks during execution. By considering correlation and temporal sequence, the regulatory data denoising system can ensure that the extracted task chain features comprehensively and accurately reflect the characteristics and relationships of each spatiotemporal monitoring and regulation task.

[0234] Secondly, the regulatory data denoising system needs to use the target liveness monitoring information as the denoising benchmark and, based on the task chain characteristics of each spatiotemporal domain monitoring and regulation task, generate the target denoised instruction text with the highest timeliness weight. This process is a text generation process, requiring the regulatory data denoising system to use the target liveness monitoring information as the benchmark and, according to the task chain characteristics of each spatiotemporal domain monitoring and regulation task, generate the target denoised instruction text with the highest timeliness weight. Timeliness weight is an indicator that measures the timeliness of text; the higher its value, the higher the timeliness of the text.

[0235] When generating target denoising instruction text, the regulatory data denoising system needs to consider the timeliness and accuracy of the text. Timeliness refers to whether the information contained in the text is up-to-date and relevant, while accuracy refers to whether the information contained in the text is accurate and reliable. To ensure the timeliness and accuracy of the text, the regulatory data denoising system needs to conduct in-depth analysis and processing of the task chain characteristics of each spatiotemporal monitoring and regulation task, extract the most timely and accurate information, and integrate it into the generation process of the target denoising instruction text.

[0236] For example, the regulatory data denoising system has three spatiotemporal monitoring and regulation tasks: task MA, task MB, and task MC. Through analysis based on correlation maps and long short-term memory models, the regulatory data denoising system can extract the task chain features of these three tasks, namely features FA, features FB, and features FC. Simultaneously, the regulatory data denoising system also obtains target liveness monitoring information, which includes the latest monitoring data on the target liveness.

[0237] Next, the data denoising system needs to use the target liveness monitoring information as the denoising benchmark and generate a target denoising instruction text based on features FA, FB, and FC. In this process, the system considers the timeliness and accuracy of the features, as well as their correlation with the target liveness monitoring information. Through comprehensive analysis and processing, the system can generate a target denoising instruction text with the highest timeliness weight, containing the latest instructions and suggestions on how to denoise the target liveness.

[0238] The target denoising instruction text can state: "Based on the latest target liveness monitoring information, the data denoising system has found that feature FA of task MA has the greatest impact on target liveness denoising. Therefore, the data denoising system recommends prioritizing task MA and paying special attention to changes in feature FA. Meanwhile, feature FB of task MB and feature FC of task MC also have some impact on denoising, but their timeliness is relatively low. When executing tasks, please be sure to use the target liveness monitoring information as a benchmark to ensure the accuracy and effectiveness of denoising."

[0239] Through this process, the control data denoising system can generate a target denoising instruction text that uses the target liveness monitoring information as the denoising benchmark and covers various spatiotemporal domain monitoring and control tasks. This text not only has the highest timeliness weight but also contains the latest instructions and suggestions on how to denoise the target liveness. It helps the control data denoising system better understand and execute various spatiotemporal domain monitoring and control tasks, ensuring that the denoising processing of the control data achieves the best results.

[0240] In some alternative embodiments, determining the task chain features of each spatiotemporal monitoring and control task based on the correlation map and the long short-term memory model includes: determining, based on the correlation map, the target denoising feature with the highest correlation to the denoising benchmark of the target liveness monitoring information from each spatiotemporal monitoring and control task; using the target denoising feature as the denoising benchmark of the derived liveness monitoring information, and based on the correlation map, determining the target denoising feature with the highest correlation to the denoising benchmark of the derived liveness monitoring information from the remaining spatiotemporal monitoring and control tasks other than the target spatiotemporal monitoring and control task; proceeding to using the target denoising feature as the denoising benchmark of the derived liveness monitoring information, and based on the correlation map, determining the target denoising feature with the highest correlation to the denoising benchmark of the derived liveness monitoring information from the remaining spatiotemporal monitoring and control tasks other than the target spatiotemporal monitoring and control task, until all spatiotemporal monitoring and control tasks have been traversed; and using the traversal time sequence label of each target spatiotemporal monitoring and control task as the task chain feature of each spatiotemporal monitoring and control task.

[0241] In this embodiment, determining the task chain characteristics of each spatiotemporal monitoring and control task based on the correlation map and long short-term memory model aims to extract key features that can effectively characterize the task chain by deeply analyzing and learning the intrinsic relationships between each task. This process not only relies on the direct correlation between tasks but also involves iteratively mining deeper-level correlation patterns, ultimately forming a feature set that can comprehensively reflect the dynamic characteristics of the task chain.

[0242] To reiterate, a correlation graph is a data structure used to represent the relationships between different entities (in this case, spatiotemporal monitoring and control tasks). In the graph, nodes represent individual tasks, while edges represent the correlation strength between tasks. This structure allows control data denoising systems to intuitively understand and analyze the complex relationships between tasks, providing a foundation for subsequent feature extraction.

[0243] Next, one of the core steps of this embodiment is to identify the target denoising achievement feature most relevant to the denoising benchmark of the target liveness monitoring information from the correlation map. This target liveness monitoring information may represent a specific monitoring target or indicator, while the denoising benchmark is a standard for evaluating the quality of monitoring data. Through the correlation map, the data denoising system can quantify the correlation between each task and this denoising benchmark, and select the task with the highest correlation as the initial target denoising achievement feature.

[0244] Subsequently, the technical solution employs an iterative approach, treating the currently selected target denoising compliant feature as a new "denoising benchmark for derived liveness monitoring information," and searching among the remaining tasks for the next target denoising compliant feature most relevant to this new benchmark. This process is repeated continuously, with each iteration updating the denoising benchmark and selecting the most relevant task from the remaining tasks, until all tasks have been traversed. This iterative method simulates the propagation and accumulation effects of information in the task chain, enabling the finally extracted features to capture the dynamic changes and long-term dependencies of the task chain.

[0245] In this application, the application of the Long Short-Term Memory (LSTM) model is particularly important. LSTM is a special type of recurrent neural network, especially suitable for processing and predicting long-term dependencies in time series data. In this technical solution, the LSTM model is used to assist in analyzing and predicting the time series correlations between various tasks in a task chain. By using the LSTM model, the data denoising system can be better understood and simulated to better understand the evolution of task chain features over time, which is crucial for capturing the dynamic characteristics of the task chain.

[0246] Specifically, when processing task chain data, the LSTM model considers the historical information of each task and their interactions, thereby generating a series of feature vectors that reflect the dynamic state of the task chain. These feature vectors not only contain the direct correlations between tasks but also implicitly contain indirect correlation patterns formed over time. This feature representation, which integrates direct and indirect correlations, provides a richer description of the task chain characteristics for regulating data denoising systems.

[0247] Finally, the traversal time sequence label of each target spatiotemporal domain monitoring and control task is used as its task chain feature. In this embodiment, the "traversal time sequence label" records the order and time points at which tasks are selected as target denoising achievement features during the iteration process. Through these labels, the control data denoising system can clearly see how the importance and influence of each task in the task chain changes over time, which is of great significance for understanding the dynamic behavior of the task chain and optimizing control strategies.

[0248] For example, a data denoising system has a task chain containing five spatiotemporal monitoring and control tasks, labeled T1, T2, T3, T4, and T5. Based on correlation analysis, the system finds that T1 has the highest correlation with the initial denoising benchmark (e.g., a specific environmental indicator), therefore T1 is selected as the first target denoising feature. Next, the system uses T1 as the new benchmark and finds that T3 has the highest correlation with T1, making T3 the second target denoising feature. Following this logic, the system iterates, eventually obtaining a task chain with the traversal order T1→T3→T5→T2→T4. This traversal order constitutes the task chain feature for each task, reflecting their importance and dynamic relationships within the task chain.

[0249] Thus, by combining correlation mapping and long short-term memory models, in-depth mining of the characteristics of spatiotemporal monitoring and control task chains was achieved. This method not only considers the direct correlation between tasks but also captures the dynamic changes and long-term dependencies in the task chain through iterative methods, providing a powerful tool for task chain analysis and optimization.

[0250] In a preferred embodiment, the spatiotemporal domain monitoring and control task is a neural signal monitoring and control task; the step of generating the target denoising instruction text with the highest timeliness weight based on the task chain characteristics of each spatiotemporal domain monitoring and control task includes: determining the neural signal monitoring and control elements between two consecutive spatiotemporal domain monitoring and control tasks according to the task chain characteristics of each spatiotemporal domain monitoring and control task; if the neural signal monitoring and control elements represent that there is a labeled monitoring and control task that has completed denoising optimization in the implanted monitoring and control information pool between the two consecutive spatiotemporal domain monitoring and control tasks, then based on the two consecutive spatiotemporal domain monitoring and control tasks... The control task and the marker monitoring and control task that has completed denoising optimization generate the target denoising instruction text; if the neural signal monitoring and control element indicates that there are no related task items between the two consecutive spatiotemporal domain monitoring and control tasks, then, with the two consecutive spatiotemporal domain monitoring and control tasks as a reference, several spatiotemporal domain monitoring and control tasks are decomposed to obtain a first target spatiotemporal domain monitoring and control task set and a second target spatiotemporal domain monitoring and control task set; based on the first target spatiotemporal domain monitoring and control task set and the second target spatiotemporal domain monitoring and control task set, a first alternative denoising instruction text and a second alternative denoising instruction text are generated respectively.

[0251] In this application scenario, the data denoising system focuses on a specific spatiotemporal monitoring and control task: neural signal monitoring and control. The core of this task lies in the precise monitoring and control of neural signals to achieve a deep understanding and effective intervention of nervous system activity. Based on this, this embodiment aims to generate the target denoised instruction text with the highest timeliness weight based on the task chain characteristics of each neural signal monitoring and control task. This process requires not only a deep understanding of the task chain characteristics by the data denoising system, but also the ability to accurately identify and utilize these characteristics to guide the generation of the denoised instruction text.

[0252] In this embodiment, the neural signal monitoring and regulation elements refer to a series of key parameters and characteristics that form the basis of the monitoring task, which together determine the nature and execution method of the monitoring task. These elements may include the target area to be monitored, the type of signal, the frequency of acquisition, the processing algorithm, etc., which together constitute a complete description of the monitoring task.

[0253] Next, the core step of this embodiment is to determine the neural signal monitoring and regulation elements between two consecutive tasks based on the task chain characteristics of each neural signal monitoring and regulation task. The key to this step is identifying the correlations and differences between tasks, thereby extracting feature elements that can characterize the relationships between tasks. For example, if two consecutive tasks monitor neural signals in the same brain region but use different processing algorithms, then the "processing algorithm" is an important monitoring and regulation element between these two tasks.

[0254] Once the neural signal monitoring and regulation elements between tasks are identified, the regulatory data denoising system can further analyze the information contained in these elements. If the elements represent a labeled monitoring and regulation task that has undergone denoising optimization within an embedded monitoring and regulation information pool between two consecutive tasks, this means that there is a task between these two tasks that has been optimized and has achieved high-quality denoising. In this case, the regulatory data denoising system can directly generate target denoising instruction text based on these two tasks and the denoising-optimized task. This text will contain methods and strategies on how to replicate or borrow from the denoising-optimized task to guide the execution of subsequent tasks.

[0255] For example, tasks R1 and R2 are two consecutive neural signal monitoring and modulation tasks, and the monitoring and modulation elements between them indicate the existence of a task R_opt that has already undergone denoising optimization. In this case, the modulated data denoising system can generate target denoising instruction text such as: "When performing tasks R1 and R2, it is recommended to refer to the methods and strategies of the denoising-optimized task R_opt to ensure the accuracy and reliability of the data." Such text directly provides specific guidance on how to perform effective denoising.

[0256] However, if the neural signal monitoring and regulation elements represent two consecutive tasks without any related task items, this means that the two tasks are not directly related in terms of monitoring objectives, methods, or strategies. In this case, the regulation data denoising system needs to adopt a different strategy to generate denoised instruction text. Specifically, the regulation data denoising system can use these two consecutive tasks as reference points to decompose the entire neural signal monitoring and regulation task set into two subsets: a first target spatiotemporal domain monitoring and regulation task set and a second target spatiotemporal domain monitoring and regulation task set. These two subsets respectively contain other tasks related to or similar to the first and second tasks.

[0257] After obtaining these two subsets, the regulatory data denoising system can generate a first alternative denoising instruction text and a second alternative denoising instruction text based on them, respectively. These two texts will provide suggestions and guidance on how to perform denoising, tailored to the task characteristics of each subset. For example, if the tasks in the first target spatiotemporal domain monitoring and regulation task set primarily focus on neural signals in a specific brain region, the first alternative denoising instruction text might emphasize denoising techniques and strategies specific to that brain region; while if the tasks in the second target spatiotemporal domain monitoring and regulation task set involve different types of neural signals, the second alternative denoising instruction text might provide suggestions on how to handle denoising for these different types of signals.

[0258] Thus, by deeply analyzing the task chain characteristics of neural signal monitoring and regulation tasks, effective generation of denoising instruction text for targets with the highest timeliness weight was achieved. This process not only considers the direct correlation between tasks but also addresses situations where tasks are not directly related by decomposing the task set and generating alternative texts. This method provides a flexible and comprehensive strategy for denoising regulation data systems to guide denoising work in neural signal monitoring and regulation tasks, thereby improving data accuracy and reliability.

[0259] In some preferred embodiments, after the target denoising instruction text is sent to the edge-side data optimization device, the method further includes: obtaining the current device status characteristics of the edge-side data optimization device; if the current device status characteristics do not match the target denoising instruction text, and the number of spatiotemporal monitoring and control tasks to be processed is less than a set number, then identifying the correlation between the spatiotemporal monitoring and control tasks to be processed based on a cross-modal decision tree, and determining the denoising instruction update text based on the correlation between the current device status characteristics and the spatiotemporal monitoring and control tasks to be processed; if the current device status characteristics do not match the target denoising instruction text, and the number of spatiotemporal monitoring and control tasks to be processed is greater than a set number, then generating again the denoising instruction update text with the largest timeliness weight involving each spatiotemporal monitoring and control task to be processed, based on the current device status characteristics; sending the denoising instruction update text to the edge-side data optimization device, so that the edge-side data optimization device can optimize the spatiotemporal monitoring and control tasks to be processed through the denoising instruction update text.

[0260] Understandably, after the target denoising instruction text is sent to the edge-side data optimization device, a series of follow-up operations are performed to ensure that the denoising instruction text matches the current state of the device and to make corresponding adjustments for different situations. This process not only demonstrates the flexibility and adaptability of the technology but also further improves the effectiveness and efficiency of data optimization.

[0261] In edge data optimization applications, the current device status characteristics refer to a set of parameters, including the device's operating status, resource usage, and processing capacity at a given moment. These characteristics collectively describe the device's overall state at the current time and are crucial for determining whether the device can perform a specific task.

[0262] Next, the technical solution considers the situation where the current device status characteristics do not match the target denoising instruction text. When this mismatch occurs, if the number of pending spatiotemporal monitoring and control tasks is less than a set number, the system will adopt a strategy based on a cross-modal decision tree to determine the denoising instruction update text.

[0263] Furthermore, cross-modal decision trees are a special type of decision tree model capable of handling data from different modalities (i.e., different sources or types) and establishing effective correlations between these data. In this technical solution, cross-modal decision trees are used to identify the correlation between the spatiotemporal monitoring and control tasks to be processed. This correlation may be based on factors such as the similarity between tasks, their spatiotemporal proximity, or their degree of sharing of equipment resources. Through the analysis of cross-modal decision trees, the control data denoising system can obtain a quantitative index of task correlation, which will serve as an important basis for subsequently generating denoised instruction update text.

[0264] After determining the correlation between tasks, a denoising instruction update text is generated based on the current device state characteristics. This update text will contain specific guidance on how to adjust the denoising strategy according to the device's current state and the correlation between tasks. For example, if the device's current processing capacity is low, but the correlation between tasks is high, the update text might suggest merging highly correlated tasks to reduce the device's load and improve processing efficiency.

[0265] On the other hand, if the current device status characteristics do not match the target denoising instruction text, and the number of pending spatiotemporal monitoring and control tasks exceeds the set number, then the technical solution will adopt another strategy. In this case, the system will generate a new denoising instruction text based on the current device status characteristics, with the highest timeliness weight among all pending spatiotemporal monitoring and control tasks.

[0266] In this application's embodiments, timeliness weight is a key concept. It refers to a weight value assigned to each task based on its importance and urgency. Tasks with higher timeliness weights mean they have a greater impact on the overall system performance or data accuracy, and therefore require more attention in resource allocation and processing priority.

[0267] The process of generating this denoising instruction update text takes into account the current state of the device, the correlation between tasks, and the timeliness weight of each task. Through an optimized algorithm, the system can calculate how to allocate resources and process them in a given device state to maximize overall data optimization. This update text will contain detailed instructions on how to perform the denoising operation according to the calculated optimal strategy.

[0268] Finally, both the denoising instruction update text generated through cross-modal decision trees and the denoising instruction update text regenerated based on timeliness weights are sent to the edge-side data optimization device. Upon receiving this update text, the device will optimize the spatiotemporal monitoring and control tasks according to the instructions in the text. In this way, by continuously adjusting the denoising strategy based on device status and task characteristics, the control data denoising system can ensure the efficiency and accuracy of the data optimization process, thereby improving the overall system performance and stability.

[0269] For example, the edge-side data optimization device currently has low processing capacity, while there are five spatiotemporal monitoring and control tasks to be processed, and these tasks are highly correlated. In this case, the system may use cross-modal decision tree analysis to determine the correlation between tasks and generate a denoised instruction update text based on the device's current state. This text may suggest merging highly correlated tasks or prioritizing tasks with high timeliness weights to reduce the device's burden and improve processing efficiency. If the number of tasks to be processed exceeds a set limit, the system may recalculate the timeliness weight of each task and generate a new denoised instruction update text to ensure optimal data optimization results with limited resources.

[0270] Based on the above, in some independent embodiments, after the target denoising instruction text is sent to the edge-side data optimization device so that the edge-side data optimization device can optimize the control data of the plurality of spatiotemporal domain monitoring and control tasks through the target denoising instruction text, the method further includes: receiving the denoised neural control data returned by the edge-side data optimization device; and setting access permissions for the denoised neural control data.

[0271] It is understandable that the above-mentioned access permission settings not only ensure data security and privacy, but also make it possible to achieve more refined data management.

[0272] Denoising-reduced neuromodulation data refers to neuromodulation-related data that has been processed using specific algorithms or methods to remove noise or interference signals. This data may include neuronal electrical activity recordings, electroencephalogram (EEG) signals, or response data generated through certain neuromodulation techniques (such as deep brain stimulation or transcranial magnetic stimulation). The purpose of denoising is to improve the signal-to-noise ratio of the data, making subsequent analysis or processing more accurate and reliable.

[0273] After the edge-side data optimization device optimizes the control data for the spatiotemporal domain monitoring and control task, it sends the optimized, denoised neural control data back to the central server or a designated data processing node. This process is achieved through network transmission and may involve data encryption and compression to ensure the security and efficiency of the transmission.

[0274] After receiving the denoised neuromodulation data, the next crucial step is to set access permissions for this data. Access permission settings are a data security management measure that defines the access capabilities and operational permissions of different users or user groups to data resources. By setting appropriate access permissions, the leakage and misuse of sensitive data can be prevented, ensuring data security and privacy.

[0275] When setting access permissions, the first step is to determine the sensitivity and importance of the data. For highly sensitive or important data, stricter access permissions should be set, such as allowing access only to specific users or user groups, with detailed access logging and auditing. For relatively less sensitive or important data, more lenient access permissions can be set to facilitate data analysis and processing by a wider range of users.

[0276] Specifically, access permission settings may include the following aspects:

[0277] (1) User authentication: Ensure that the user accessing the data is legitimate and has the appropriate access rights. This is usually achieved through username and password, two-factor authentication, or other authentication mechanisms;

[0278] (2) Access Control List (ACL): Defines the access permissions of different users or user groups to data resources. ACL can specify the types of operations that users can perform (such as read, write, execute, etc.) and the objects of the operations (such as specific data files, database tables, etc.);

[0279] (3) Role-Based Access Control (RBAC): Users are assigned to different roles, and corresponding access permissions are defined for each role. In this way, users gain access to data resources through their assigned roles. RBAC simplifies access management, making it easier to allocate and adjust permissions when there are many users.

[0280] (4) Data encryption: For highly sensitive data, encryption can be performed during storage or transmission to ensure that even if the data is accessed by an unauthorized user, its true content cannot be easily obtained;

[0281] (5) Access auditing and monitoring: Record user access behavior to data resources and monitor it in real time. This helps to discover potential security threats or violations and take appropriate measures in a timely manner.

[0282] In practical applications, setting access permissions may involve multiple considerations and trade-offs. For example, in the medical field, neuromodulation data may contain patients' personal privacy information, thus requiring strict access permissions to protect patients' privacy rights. At the same time, it is necessary to ensure the availability and shareability of the data while protecting patient privacy.

[0283] To achieve this goal, a role-based access control strategy can be adopted. First, define different roles, such as data administrators, researchers, and doctors, and assign corresponding access permissions to each role. For example, a data administrator might have the highest level of access, managing data storage, backup, and recovery; a researcher might have permission to read and analyze data, but not modify or delete it; and a doctor might have permission to view specific patient data to develop personalized treatment plans.

[0284] Secondly, an authentication mechanism ensures that users accessing the data are legitimate. Users must log in to the system using a username and password or other authentication credentials and be authenticated before accessing the data. This prevents unauthorized users from accessing sensitive data.

[0285] Then, access control lists (ACLs) are used to further refine the data access permissions for each role. For example, an ACL can be set up for researchers, allowing them to access specific datasets but not sensitive data containing patients' personal privacy information. Meanwhile, a different ACL can be set up for physicians, allowing them to view specific patients' neuromodulation data for diagnosis and treatment.

[0286] Finally, user access behavior is tracked through access auditing and monitoring. The system can record information such as the time of each access, user identity, accessed data, and performed operations, and monitor this information in real time. This helps to identify potential security threats or violations and allows for timely measures to protect data security and privacy.

[0287] Therefore, setting access permissions is a crucial step after receiving the denoised neuromodulation data from the edge-side data optimization device. Appropriate access permission settings ensure data security and privacy while enabling data sharing and utilization. This not only helps protect patients' privacy but also promotes research and development in the medical field, providing better data support and safeguards for the application of neuromodulation technologies.

[0288] In other, standalone embodiments, the method further includes: performing data sharing processing on the denoised neuromodulation data in response to a data sharing request.

[0289] It is understandable that, for implantable neural monitoring and modulation applications, the embodiments of this application are not limited to data denoising, optimization, and access permission settings, but also extend to the important aspect of data sharing. This means that, while ensuring data security, privacy, and compliance, the method can actively respond to data sharing requests and properly process the denoised neural modulation data for data sharing.

[0290] In implantable neuromonitoring and modulation applications, data sharing is of paramount importance. Implantable neuromonitoring and modulation technology, as a cutting-edge medical approach, interacts directly with target neural tissue through implanted electrodes and other devices to record, stimulate, or modulate neural signals. However, due to the complexity and specialization of this technology, a single research institution or medical institution often cannot independently complete all related research and application work. Therefore, data sharing has become a crucial pathway to promote the development of implantable neuromonitoring and modulation technology. By sharing data, different research institutions can jointly analyze the effects of neuromodulation, optimize algorithms, and improve the accuracy and effectiveness of the technology. Medical institutions can also share data to compare the effects of different implantable neuromonitoring and modulation devices, providing patients with more personalized treatment plans.

[0291] However, data sharing is not without risk. Because implantable neuromonitoring data involves patient privacy and medical information, strict adherence to relevant laws, regulations, and privacy policies is essential when sharing it. This necessitates that the data denoising system implement a series of measures during the data sharing process to ensure data security and privacy.

[0292] In response to data sharing requests, the data denoising system first conducts rigorous identity verification and access control audits on the requesting party. This process aims to ensure that only legitimate and compliant requesters can obtain data access permissions, thereby preventing data leakage or misuse. Identity verification may include verifying the requester's identity information, institutional qualifications, and legitimate authorizations related to research or business. Access control audits involve assessing the requester's purpose, scope, method of data use, and security measures to ensure that data sharing complies with relevant laws, regulations, and privacy policies. This step is crucial because it lays a solid foundation for the security of data sharing.

[0293] Once the requester passes authentication and authorization verification, the data denoising system further refines the denoised neuromodulation data to meet data sharing needs. This may include data format conversion, anonymization, encrypted transmission, and access control. Data format conversion ensures that the shared data can be easily read and used by the requester, regardless of the data processing or analysis tools they use. Anonymization protects patient privacy by removing or replacing personal identifiers, preventing direct association of shared data with specific individuals and reducing the risk of privacy breaches. Encrypted transmission ensures data security during sharing, preventing theft, tampering, or malicious use. Access control restricts access to the shared data, allowing only authorized users to access and use it, further enhancing data security.

[0294] In the data sharing process, the data denoising system places particular emphasis on data traceability and auditability. This means that all data sharing activities are recorded in detail, including the sharing time, sharing recipients, shared content, and usage methods. This information is crucial for monitoring data usage, preventing data misuse, and meeting relevant legal and regulatory requirements. By retaining these data sharing records, the data denoising system can trace the flow and usage of data when necessary, ensuring the legal and compliant use of data.

[0295] Furthermore, the data denoising system offers flexible and diverse data sharing strategy configuration options. Administrators can easily configure different data sharing strategies based on various data sharing scenarios and needs, such as setting data sharing periods, limiting the number of times data can be shared, and defining the scope of data sharing. These strategy configurations aim to balance the needs of data sharing with the protection of data security and privacy, ensuring that data sharing meets research or business requirements without disclosing patients' personal privacy or sensitive information.

[0296] In practical applications, the data-sharing function of denoising control systems can be widely used in medical, scientific research, and commercial fields. In the medical field, multiple medical institutions can share denoised neuromodulation data to jointly study treatment methods for neurological diseases, improving diagnostic accuracy and treatment effectiveness. Research institutions can also use shared data to validate and improve neuromonitoring and modulation algorithms, driving progress in neuroscience and technology. In the commercial field, companies can share data to develop new neuromodulation products or services, meeting market demands and driving business growth. Simultaneously, data sharing can also promote cooperation and exchange between different institutions, jointly advancing the development and application of implantable neuromonitoring and modulation technologies.

[0297] However, data sharing is not without challenges. One of the biggest challenges is achieving effective data sharing while protecting personal privacy and data security. To address this challenge, the Data Denoising System employs various technical means and legal measures to ensure the security and compliance of data sharing. In addition to the aforementioned technical means such as identity verification, authorization auditing, anonymization, encrypted transmission, and access control, the system strictly adheres to relevant laws, regulations, and privacy policies to ensure the legality and compliance of data sharing. Furthermore, the system actively collaborates and communicates with relevant regulatory agencies to ensure that data sharing activities comply with regulatory requirements and obtain the necessary authorizations.

[0298] Therefore, by responding to data sharing requests and performing refined data sharing processing on the denoised neural modulation data, the modulation data denoising system can achieve effective data sharing and utilization while protecting personal privacy and data security.

[0299] Furthermore, Figure 2 This is a schematic diagram of the structure of a data denoising system 200 provided in an embodiment of this application. Figure 2 The control data denoising system 200 shown includes a processor 210, which can call and run computer programs from memory to implement the methods in the embodiments of this application.

[0300] Optionally, such as Figure 2 As shown, the data denoising system 200 may further include a memory 230. The processor 210 can retrieve and run computer programs from the memory 230 to implement the methods described in this embodiment.

[0301] The memory 230 can be a separate device independent of the processor 210, or it can be integrated into the processor 210.

[0302] Optionally, such as Figure 2 As shown, the data noise reduction system 200 may also include a transceiver 220. The processor 210 can control the transceiver 220 to interact with other devices. Specifically, it can send information or data to other devices or receive information or data sent by other devices.

[0303] Optionally, the data denoising system 200 can implement the corresponding processes of the storage engine or components (such as processing modules) in the storage engine or the device with the storage engine deployed in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0304] It should be understood that the processor in this application embodiment may be an integrated circuit chip with signal processing capabilities.

[0305] It is understood that the memory in the embodiments of this application may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. It should be noted that the memory in the systems and methods described herein is intended to include, but is not limited to, suitable types of memory.

[0306] Based on the above, a readable storage medium is provided, on which a program or instructions are stored, and when the program or instructions are executed by a processor, the steps of the above method are implemented.

[0307] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0308] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0309] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of protection of this application, and all of these forms are within the protection scope of this application.

Claims

1. A method for denoising control data based on implantable medical devices, characterized in that, include: The data denoising system acquires target liveness monitoring information from the implanted monitoring and control information pool and several spatiotemporal monitoring and control tasks belonging to the implanted monitoring and control information pool. There is overlap between the several spatiotemporal monitoring and control tasks. The spatiotemporal monitoring and control tasks refer to the monitoring and control tasks performed on the target liveness under different time and space conditions. The data denoising system generates several data optimization execution strategies with denoising activation and denoising pause features based on the target liveness monitoring information and the several spatiotemporal monitoring and control tasks. The data denoising system obtains the denoising activation features and denoising pause features corresponding to the respective data optimization execution strategies. Based on the denoising activation features and denoising pause features corresponding to each of the data optimization execution strategies, potential noisy image descriptors are determined. Based on the denoising activation features and denoising pause features corresponding to each of the data optimization execution strategies, the correlation degree between the first strategy identifier of each data optimization execution strategy and the second strategy identifier of the remaining data optimization execution strategies is determined. The latent noise image descriptors are mapped into correlation maps based on the determined correlation degree. The data denoising system generates target denoising instruction text based on the correlation map, using the target liveness monitoring information as the denoising benchmark, and covering various spatiotemporal monitoring and control tasks. The target denoising instruction text is a customized data optimization scheme, which includes how to select and execute denoising strategies under different spatiotemporal monitoring and control tasks to achieve the best denoising effect; The control data denoising system sends the target denoising instruction text to the edge-side data optimization device, so that the edge-side data optimization device can optimize the control data of the several spatiotemporal domain monitoring and control tasks through the target denoising instruction text.

2. The method as described in claim 1, characterized in that, The generation of several data optimization execution strategies with denoising activation and denoising pause features based on the target liveness monitoring information and the several spatiotemporal monitoring and control tasks includes: Based on the monitoring expectation mode of the spatiotemporal domain monitoring and control tasks, the several spatiotemporal domain monitoring and control tasks are respectively mapped into several first task knowledge graphs containing denoising activation features and denoising pausing features. The target liveness monitoring information is mapped into a second task knowledge graph containing the same denoising activation features and denoising pause features; The data optimization execution strategies are generated based on the aforementioned first task knowledge graphs and second task knowledge graphs.

3. The method as described in claim 1, characterized in that, The acquisition of denoising activation features and denoising pause features corresponding to each data optimization execution strategy includes: From the aforementioned data optimization execution strategies, a target data optimization execution strategy that is related to the historical execution strategies in the embedded monitoring and control information pool in the monitoring and control dimension is obtained; based on the denoising activation features and denoising pause features of the historical execution strategies, the denoising activation features and denoising pause features corresponding to the target data optimization execution strategy are determined; or, Using the topological labels corresponding to the target liveness monitoring information in the aforementioned data optimization execution strategies as a reference, a target feature space is determined; the strategy node variables of each data optimization execution strategy other than the topological labels corresponding to the target liveness monitoring information are obtained; and the denoising activation features and denoising pause features of each data optimization execution strategy in the target feature space are determined based on the strategy node variables of each data optimization execution strategy.

4. The method as described in claim 1, characterized in that, The step of mapping the latent noise image descriptor into the correlation map based on the determined correlation degree includes: Mining target denoising activation features and target denoising pause features from the latent noise image descriptor; If the target denoising activation feature and the target denoising pause feature correspond to the same data optimization execution strategy, then the graph semantic unit corresponding to the target denoising activation feature and the target denoising pause feature is determined to be a preset semantic code. If the target denoising activation feature and the target denoising pause feature correspond to different data optimization execution strategies, then the cosine similarity between the target denoising pause feature and the target denoising activation feature is taken from the determined correlation and used as the graph semantic unit corresponding to the target denoising activation feature and the target denoising pause feature, so as to obtain the correlation graph.

5. The method as described in claim 1, characterized in that, The acquisition of target liveness monitoring information from the implantable monitoring and control information pool and several spatiotemporal monitoring and control tasks belonging to the implantable monitoring and control information pool includes: Obtain the historical execution strategies created by the noise reduction strategy planning system from the implanted monitoring and control information pool; The execution strategies that exclude the historical execution strategies from the implanted monitoring and control information pool are taken as the execution strategies to be processed. Based on the information involvement weight of the execution strategy to be processed, the implanted monitoring and control information pool is decomposed into several monitoring and control information sets. The several spatiotemporal monitoring and control tasks are determined based on the pending execution strategies within the target monitoring and control information set, and the target liveness monitoring information is determined using the target monitoring and control information set.

6. The method as described in claim 1, characterized in that, The acquisition of target liveness monitoring information from the implantable monitoring and control information pool and several spatiotemporal monitoring and control tasks belonging to the implantable monitoring and control information pool includes: Knowledge mining is performed on the in vivo physiological attribute data corresponding to the implantable monitoring and regulation information pool to obtain a knowledge relationship network of in vivo physiological attributes; Based on the in vivo physiological attribute knowledge network, the priority task events among the various task events of the monitoring and regulation tasks in the implantable monitoring and regulation information pool are determined. If the priority task event does not have an associated priority task event, then the monitoring and control task corresponding to the priority task event shall be taken as the spatiotemporal domain monitoring and control task. If the priority task event has associated priority task events that do not belong to the same data optimization execution strategy, and the correlation degree between the priority task event and the associated priority task event is less than or equal to a set correlation degree, then the monitoring and control task corresponding to the priority task event is taken as the spatiotemporal domain monitoring and control task.

7. The method according to any one of claims 1 to 6, characterized in that, The generation of target denoising indication text based on the correlation map, using the target liveness monitoring information as the denoising benchmark, and involving various spatiotemporal monitoring and control tasks, includes: Based on the correlation map and long short-term memory model, the task chain characteristics of each spatiotemporal monitoring and control task are determined. Using the target liveness monitoring information as the denoising benchmark, and based on the task chain characteristics of each spatiotemporal monitoring and control task, a target denoising instruction text with the highest timeliness weight is generated; The task chain characteristics for determining each spatiotemporal monitoring and control task based on the correlation map and long short-term memory model include: Based on the correlation map, the target denoising compliance feature with the highest correlation to the denoising benchmark of the target liveness monitoring information is determined from various spatiotemporal monitoring and control tasks. The target denoising compliance feature is used as the denoising benchmark for the derived liveness monitoring information. Based on the correlation map, the target denoising compliance feature with the highest correlation to the denoising benchmark of the derived liveness monitoring information is determined from the remaining spatiotemporal monitoring and control tasks other than the target spatiotemporal monitoring and control task. Jump to the step of using the target denoising compliance feature as the denoising benchmark for the derived liveness monitoring information, and based on the correlation map, determine the target denoising compliance feature with the highest correlation to the denoising benchmark for the derived liveness monitoring information from the remaining spatiotemporal monitoring and control tasks other than the target spatiotemporal monitoring and control task, until all spatiotemporal monitoring and control tasks have been traversed. The traversal time sequence label of each target spatiotemporal domain monitoring and control task is used as the task chain feature of each spatiotemporal domain monitoring and control task. The spatiotemporal domain monitoring and control task is a neural signal monitoring and control task; the step of generating the target denoising indication text with the highest timeliness weight based on the task chain features of each spatiotemporal domain monitoring and control task includes: Based on the task chain characteristics of each spatiotemporal domain monitoring and control task, the neural signal monitoring and control elements between two consecutive spatiotemporal domain monitoring and control tasks are determined. If the neural signal monitoring and control element represents a labeled monitoring and control task that has completed denoising optimization in the implanted monitoring and control information pool between the two consecutive spatiotemporal domain monitoring and control tasks, then the target denoising instruction text is generated based on the two consecutive spatiotemporal domain monitoring and control tasks and the labeled monitoring and control task that has completed denoising optimization. If the neural signal monitoring and control elements indicate that there are no related task items between the two consecutive spatiotemporal domain monitoring and control tasks, then, taking the two consecutive spatiotemporal domain monitoring and control tasks as a reference, several spatiotemporal domain monitoring and control tasks are decomposed to obtain a first target spatiotemporal domain monitoring and control task set and a second target spatiotemporal domain monitoring and control task set; based on the first target spatiotemporal domain monitoring and control task set and the second target spatiotemporal domain monitoring and control task set, a first alternative denoising instruction text and a second alternative denoising instruction text are generated respectively.

8. The method according to any one of claims 1 to 6, characterized in that, After sending the target denoising instruction text to the edge-side data optimization device, the method further includes: Obtain the current device status characteristics of the edge-side data optimization device; If the current device status features do not match the target denoising instruction text, and the number of spatiotemporal monitoring and control tasks to be processed is less than a set number, then the correlation between the spatiotemporal monitoring and control tasks to be processed is identified based on the cross-modal decision tree, and the denoising instruction update text is determined based on the correlation between the current device status features and the spatiotemporal monitoring and control tasks to be processed. If the current device status features do not match the target denoising instruction text, and the number of spatiotemporal monitoring and control tasks to be processed is greater than the set number, then a new denoising instruction update text with the highest timeliness weight among all the spatiotemporal monitoring and control tasks to be processed will be generated based on the current device status features. The denoising instruction update text is sent to the edge-side data optimization device so that the edge-side data optimization device can optimize the spatiotemporal domain monitoring and control task to be processed through the denoising instruction update text.

9. A data denoising system for control, characterized in that, The method includes at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, causing the at least one processor to perform the method according to any one of claims 1-8.

Images