Wearable device for brain surgery postoperative patient rehabilitation monitoring and data analysis system

Abstract

The invention discloses a wearable device and a data analysis system for brain surgery postoperative patient rehabilitation monitoring, and relates to the technical field of medical health monitoring, the system comprises a federated learning module, a digital twinning module, a rehabilitation evaluation module, a universe interaction module and a feedback control module, and the federated learning module guarantees data privacy and realizes multi-center model training; the digital twinning module constructs a virtual mirror image of a patient and simulates a rehabilitation scheme effect; the rehabilitation evaluation module quantitatively evaluates the rehabilitation condition; the meta-universe interaction module creates a virtual community and supports brain-controlled interaction training; the feedback control module dynamically adjusts a rehabilitation scheme, the wearable device comprises an ultra-flexible sensing layer, a quantum signal acquisition layer, a nerve regulation and control layer, a data processing layer and a wireless communication module, high-sensitivity data acquisition, transmission and analysis are achieved, the accuracy, personalization and interactivity of brain surgery postoperative rehabilitation monitoring are improved, and the brain surgery postoperative rehabilitation monitoring system is suitable for being popularized and applied. And intelligent development of rehabilitation treatment is promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of medical health monitoring, and more specifically, to a wearable device and a data analysis system for the rehabilitation monitoring of patients after brain surgery. Background Art

[0002] With the development of medical technology, multi-disciplinary integration technologies are gradually being adopted for the rehabilitation monitoring and treatment after brain surgery. Traditional methods have deficiencies in data privacy, personalized programs, interactivity, and the accuracy of neuromodulation. To overcome these problems, a new type of rehabilitation monitoring system combines technologies such as federated learning, digital twin, VR / AR, etc. Federated learning realizes cross-hospital model training while protecting privacy, improving prediction accuracy. The digital twin technology constructs a virtual patient mirror, simulates the effects of rehabilitation programs, and supports personalized treatment. The rehabilitation evaluation module comprehensively quantifies the patient's rehabilitation situation. The metaverse interaction module uses VR / AR technology to create an immersive training environment and enhance patient participation. The feedback control module adjusts the training program and neuromodulation parameters according to the evaluation results to achieve treatment optimization. In addition, wearable device technology provides a convenient means of data collection. Its ultra-flexible sensing layer and quantum signal acquisition layer ensure high-precision data collection, while the neuromodulation layer and data processing layer achieve real-time stimulation adjustment and signal processing. The wireless communication module ensures the secure transmission of data, thus improving the rehabilitation monitoring system.

[0003] Existing rehabilitation monitoring and treatment after brain surgery have problems such as insufficient data privacy protection, lack of personalized rehabilitation programs, poor interactivity in rehabilitation training, and low accuracy of neuromodulation. Summary of the Invention

[0004] In order to overcome the problems of insufficient data privacy protection, lack of personalized rehabilitation programs, poor interactivity in rehabilitation training, and low accuracy of neuromodulation existing in the existing rehabilitation monitoring and treatment after brain surgery, the present invention designs a wearable device and a data analysis system for the rehabilitation monitoring of patients after brain surgery, which can effectively solve the above technical problems.

[0005] To solve the above technical problems, the technical solution of the present invention is as follows: A data analysis system for the rehabilitation monitoring of patients after brain surgery, comprising: a federated learning module, a digital twin module, a rehabilitation evaluation module, a metaverse interaction module, and a feedback control module; The federated learning module adopts a distributed machine learning framework to achieve multi-center model training across hospitals on the premise of protecting the privacy of patient data; The digital twin module constructs a virtual mirror of the patient, integrates intraoperative imaging data, real-time physiological parameters, and rehabilitation progress, and simulates the effects of different training programs through a physics engine; The rehabilitation assessment module evaluates the patient's rehabilitation status based on the collected data and the results of model training; The metaverse interaction module creates a virtual rehabilitation community by combining VR / AR technologies. The patient completes social interaction training through a brain-controlled virtual avatar, and the EEG data and virtual environment feedback are interacted in real time; The feedback control module adjusts the rehabilitation training plan and neuromodulation parameters according to the rehabilitation assessment results.

[0006] Preferably, the federated learning module constructs a distributed model training architecture. Each medical institution locally trains a complication prediction model based on LSTM, and the central server aggregates global parameters through dynamic weighting, where the dynamic weighting is determined according to the data volume and data quality of each medical institution; The differential privacy technology is used to add noise to protect data privacy.

[0007] Preferably, the digital twin module fuses intraoperative DTI images, real-time EEG data, and fNIRS blood oxygen parameters, and constructs a dynamic three-dimensional model of the patient's brain function through Unreal Engine; Simulate the changes in cerebrospinal fluid dynamics and calculate the cerebral edema risk index:

[0008] Among them, is the change rate of intracranial pressure, is the second derivative of the change in ventricular volume, and the coefficient is negatively correlated with the change rate of ventricular volume. When the change rate of ventricular volume increases, decreases, thereby reducing the value.

[0009] Preferably, the rehabilitation assessment module calculates the functional connection strength: Dynamically adjust the threshold: Among them, represents the channel pair, is the phase locking value, is the basic threshold.

[0010] Preferably, the metaverse interaction module generates a virtual rehabilitation community scene and realizes the actions of the brain-controlled virtual avatar through a VR / AR headset device; When the attention concentration lasts for a preset time exceeding the threshold, a new scene is unlocked and the neuromodulation intensity is reduced by a preset percentage; Among them, the attention concentration is: , is the power spectral density.

[0011] Preferably, the feedback control module generates a multi-level intervention strategy, including adjusting stimulation parameters, updating the VR training difficulty, and pushing medical advice reminders.

[0012] A wearable device for postoperative rehabilitation monitoring of brain surgery patients, comprising: a super-flexible sensing layer, a quantum signal acquisition layer, a neuromodulation layer, a data processing layer, and a wireless communication module; The super-flexible sensing layer includes: a nano-level flexible electrode array, arranged in a serpentine topology structure, made of a composite of graphene and liquid metal, directly attached to the patient's scalp for collecting electroencephalogram signals, and the nano-level flexible electrode array has a preset thickness and elongation rate; A self-healing conductive material layer that wraps the electrode array, includes a dynamically covalently cross-linked polyborosiloxane matrix and a silver nanowire network, and triggers molecular chain recombination within a preset time when a resistance value mutation is detected, and the self-healing conductive material layer has a preset repair efficiency; The quantum signal acquisition layer includes: a quantum dot sensor array, adopting a core-shell structure, enhancing the sensitivity of electroencephalogram signals through the quantum tunneling effect, detecting weak electrical signals in deep brain regions, and the quantum dot sensor array detects high-frequency oscillation signals in the hippocampus; The neuromodulation layer includes: a closed-loop transcranial electrical stimulation module that dynamically adjusts stimulation parameters according to real-time electroencephalogram characteristics, and the stimulation frequency satisfies a preset formula:

[0013] Wherein, is a dynamic adjustment coefficient, is the power change amount in the θ band, is the amplitude in the α band, is a preset threshold; The data processing layer is used for adaptive noise reduction of quantum signals, and uses an improved wavelet threshold function combined with independent component analysis to remove power frequency interference; Calculate the motor cortex activation index: Wherein, - = 13−30Hz, is a time window function; The wireless communication module supports dual-mode transmission and uploads data to the cloud system through an encryption protocol.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: The federated learning module of the present invention adopts a distributed machine learning framework and differential privacy technology. Each medical institution trains a complication prediction model based on LSTM locally. The central server aggregates global parameters through dynamic weighting and adds noise to protect data privacy, ensuring that patient data is not leaked during the cross-hospital training process, thus solving the problem of insufficient data privacy protection; the digital twin module integrates intraoperative DTI images, real-time EEG data, and fNIRS blood oxygen parameters, constructs a dynamic three-dimensional model of the patient's brain function through Unreal Engine, and simulates the change of cerebrospinal fluid dynamics to calculate the cerebral edema risk index. The rehabilitation evaluation module calculates the functional connection strength and dynamically adjusts the threshold according to the collected data and the model training results, and customizes a precise rehabilitation plan for the patient, solving the problem of lack of personalization in the rehabilitation plan; the metaverse interaction module combines VR / AR technology to create a virtual rehabilitation community. Patients complete social interaction training through brain-controlled virtual avatars. The EEG data and the virtual environment feedback interact in real time. When the attention concentration exceeds the threshold for a preset time, a new scene is unlocked and the neuromodulation intensity is reduced by a preset percentage. This interactive training method improves the patient's participation and rehabilitation compliance, solving the problem of poor interactivity in rehabilitation training; the feedback control module generates multi-level intervention strategies according to the rehabilitation evaluation results, including adjusting stimulation parameters, updating the VR training difficulty, and pushing medical advice reminders. The closed-loop transcranial electrical stimulation module in the neuromodulation layer of the wearable device dynamically adjusts the stimulation parameters according to the real-time EEG characteristics. The data processing layer adaptively denoises the quantum signal and calculates the motor cortex activation degree index. With the combined action of the above technical means, the accuracy of neuromodulation is improved, solving the problem of low accuracy of neuromodulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only exemplary, and for those of ordinary skill in the art, without creative efforts, other implementation drawings can be obtained according to the provided drawings.

[0016] Figure 1 It is a structural diagram of a data analysis system for the rehabilitation monitoring of patients after brain surgery; DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The drawings are only for exemplary illustration and cannot be construed as a limitation of this patent; To better illustrate this embodiment, some components in the drawings are omitted, enlarged, or reduced, and do not represent the size of the actual product; For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0018] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0019] Embodiment For the data analysis system for the rehabilitation monitoring of patients after brain surgery, please refer to Figure 1 , including: a federated learning module, a digital twin module, a rehabilitation evaluation module, a metaverse interaction module, and a feedback control module; The federated learning module adopts a distributed machine learning framework to achieve multi-center model training across hospitals while protecting the privacy of patient data; The digital twin module constructs a virtual mirror of the patient, integrates intraoperative imaging data, real-time physiological parameters, and rehabilitation progress, and simulates the effects of different training programs through a physics engine; The rehabilitation evaluation module evaluates the patient's rehabilitation status based on the collected data and the results of model training; The metaverse interaction module creates a virtual rehabilitation community by combining VR / AR technologies. Patients complete social interaction training through brain-controlled virtual avatars, and real-time interaction between EEG data and virtual environment feedback is achieved; The feedback control module adjusts the rehabilitation training program and neuromodulation parameters according to the rehabilitation evaluation results.

[0020] In a specific implementation, a distributed machine learning framework is adopted to connect the servers of multiple hospitals to form a multi-center model training network across hospitals. The servers of each hospital serve as local nodes and are responsible for training models based on the patient data of their own hospitals; the central server coordinates the overall situation and is responsible for parameter aggregation and model update; each medical institution locally trains a complication prediction model based on LSTM (Long Short-Term Memory Network). The LSTM network structure is set such that the input layer receives the patient's time series data, such as daily physiological indicators and rehabilitation training data after surgery. The hidden layer contains multiple LSTM units for capturing the long-term dependencies of the data, and the output layer predicts the probability of complications. During the training process, the Adam optimizer is used, the learning rate is set, and the cross-entropy loss is used as the loss function.

[0021] The central server aggregates global parameters through dynamic weighting. The dynamic weighting is evaluated according to the data volume of each medical institution (the weight is proportional to the data volume) and data quality through indicators such as data integrity and accuracy. The weight range is determined from 0 to 1. For example, the model parameters of hospitals with a large data volume account for a higher proportion during aggregation. At the same time, differential privacy technology is adopted to add Laplace noise during local model updates, and the noise scale is adjusted according to data sensitivity to protect the privacy of patient data.

[0022] Integrate DTI images during the operation (clearly showing the orientation and integrity of cerebral white matter fiber tracts), real-time EEG data (reflecting the instantaneous changes in brain electrical activities), and blood oxygen parameters of fNIRS (functional near-infrared spectroscopy imaging) (indicating the dynamic changes in blood oxygen saturation in brain regions). Construct a dynamic three-dimensional model of the patient's brain function through the Unreal Engine. The model uses the patient's cranial anatomical structure as the framework, and accurately maps the DTI image fiber tracts, EEG electrode positions, and fNIRS detection areas into it.

[0023] Use computational fluid dynamics algorithms to simulate the flow state of cerebrospinal fluid in the ventricular system, including flow velocity, flow direction, pressure distribution, etc. Calculate the cerebral edema risk index according to the simulation results. The formula is: Cerebral edema risk index , where k is negatively correlated with the ventricular volume change rate. When the ventricular volume change rate increases, k decreases, thereby reducing the value of the cerebral edema risk index. When the ventricular volume increases rapidly, it indicates that cerebrospinal fluid accumulation may trigger cerebral edema. At this time, the risk index decreases, and the system issues a warning.

[0024] Calculate the functional connectivity strength between different regions of the brain. The formula is: Functional connectivity strength , where is the phase locking value of the channel pair , reflecting the synchronization of electrical activities in two brain regions; the basic threshold is obtained based on the statistical data of a large number of normal people and is used for normalization processing; the threshold is dynamically adjusted according to factors such as the patient's rehabilitation stage and training intensity. For example, in the initial stage of rehabilitation, in order to sensitively detect weak functional connectivity changes, the basic threshold is set to 0.3; as rehabilitation progresses and the patient's brain function gradually recovers, in order to focus on strong connection relationships, the basic threshold is increased to 0.5.

[0025] Use the Unity 3D engine to generate virtual rehabilitation community scenes, including diverse environments such as parks, gyms, and social squares. The patient wears a VR / AR headset device and controls the actions of the virtual avatar through EEG signals. After the EEG signals are preprocessed (filtered and denoised) and feature extracted, such as the power spectral density of alpha waves and beta waves, they are mapped into movement instructions for the virtual avatar. For example, when the patient concentrates, the virtual avatar walks forward; when the patient relaxes, the avatar stops. When the attention concentration lasts for a preset time, such as 30 seconds exceeding the threshold, such as 1.5, a new scene is unlocked and the neuromodulation intensity is reduced by a preset percentage, such as 10%, to reward the patient and adapt to their rehabilitation progress.

[0026] Generate multi-level intervention strategies based on the rehabilitation assessment results. If the assessment finds that the patient's motor function recovery is slow, adjust the stimulation parameters, such as increasing the stimulation frequency of the closed-loop transcranial electrical stimulation module by 10%; if the patient's cognitive function has improved, then update the VR training difficulty and increase the task complexity; at the same time, regularly push medical advice reminders, such as reminding the patient to take medicine on time and have regular check-ups, etc.

[0027] The closed-loop transcranial electrical stimulation module of the neuromodulation layer performs electrical stimulation on a specific area of the patient's brain through a preset electrode array according to the adjusted stimulation parameters. During the stimulation process, the patient's electroencephalogram response is monitored in real time. If abnormal discharges or discomfort symptoms occur, the stimulation is immediately paused and an alarm is issued. At the same time, the data processing layer continuously analyzes the patient's performance data under the new training difficulty, evaluates the intervention effect, and provides a basis for the next round of feedback control.

[0028] The federated learning module constructs a distributed model training architecture. Each medical institution locally trains a complication prediction model based on LSTM. The central server aggregates global parameters through dynamic weighting, where the dynamic weighting is determined according to the data volume and data quality of each medical institution; the differential privacy technology is used to add noise to protect data privacy.

[0029] The digital twin module fuses intraoperative DTI images, real-time electroencephalogram data, and fNIRS blood oxygen parameters, and constructs a dynamic three-dimensional model of the patient's brain function through Unreal Engine; simulates the changes in cerebrospinal fluid dynamics and calculates the cerebral edema risk index:

[0030] Among them, is the change rate of intracranial pressure, is the second derivative of the change in ventricular volume, and the coefficient is negatively correlated with the change rate of ventricular volume. When the change rate of ventricular volume increases, decreases, thereby reducing the value of .

[0031] Unreal Engine is a very powerful game engine developed by Epic Games. It is not only widely used in game development, but also applied to multiple fields such as movies, television, architecture, automotive, simulation, and others.

[0032] The rehabilitation evaluation module calculates the functional connectivity strength: Dynamically adjust the threshold: Among them, represents the channel pair, is the phase locking value, is the basic threshold.

[0033] The metaverse interaction module generates a virtual rehabilitation community scene and realizes the actions of the brain-controlled virtual avatar through VR / AR headset devices; when the attention concentration exceeds the threshold for a preset time, a new scene is unlocked and the neuromodulation intensity is reduced by a preset percentage; among them, the attention concentration is: , is the power spectral density.

[0034] The feedback control module generates a multi-level intervention strategy, including adjusting stimulation parameters, updating VR training difficulty, and pushing medical advice reminders.

[0035] A wearable device for postoperative rehabilitation monitoring of brain surgery patients, comprising: a super-flexible sensing layer, a quantum signal acquisition layer, a neuromodulation layer, a data processing layer, and a wireless communication module; The super-flexible sensing layer includes: a nano-scale flexible electrode array, arranged in a serpentine topology structure, made of a composite of graphene and liquid metal, directly attached to the patient's scalp for collecting electroencephalogram signals, and the nano-scale flexible electrode array has a preset thickness and elongation rate; A self-healing conductive material layer, wrapping the electrode array, containing a dynamically covalently cross-linked polyborosiloxane matrix and a silver nanowire network, triggering molecular chain recombination within a preset time when a resistance value mutation is detected, and the self-healing conductive material layer has a preset repair efficiency; The quantum signal acquisition layer includes: a quantum dot sensor array, adopting a core-shell structure, enhancing the sensitivity of electroencephalogram signals through the quantum tunneling effect, detecting weak electrical signals in deep brain regions, and the quantum dot sensor array detects high-frequency oscillation signals in the hippocampus; The neuromodulation layer includes: a closed-loop transcranial electrical stimulation module, dynamically adjusting stimulation parameters according to real-time electroencephalogram characteristics, and the stimulation frequency satisfies a preset formula:

[0036] wherein, is the dynamic adjustment coefficient, is the power change amount in the θ band, is the amplitude in the α band, is the preset threshold; The data processing layer is used for adaptive noise reduction of quantum signals, and uses an improved wavelet threshold function combined with independent component analysis to remove power frequency interference; Calculate the motor cortex activation index: wherein, - = 13−30Hz, is the time window function; The wireless communication module supports dual-mode transmission and uploads data to the cloud system through an encryption protocol.

[0037] Design a wearable device that fits the patient's head, which is in the shape of a helmet or headband as a whole, and uses a lightweight and highly elastic material as the base, such as memory plastic, to ensure that the device can adapt to different patients' head shapes and is comfortable and stable to wear. The interior of the device is partitioned by function and each layer of components is integrated in an orderly manner to ensure that all parts cooperate closely without affecting the patient's daily activities.

[0038] A nanoscale flexible electrode array with a serpentine topology, made of a composite of graphene and liquid metal, is arranged on the inner side of the device according to the international 10 - 20 lead standard to directly attach to the patient's scalp to collect electroencephalogram (EEG) signals. The thickness of the electrode array is set to 100 microns and the elongation rate is 300%, ensuring a tight fit to the scalp without hindering the patient's head movement.

[0039] Wrap a self - healing conductive material layer outside the electrode array, which includes a polyborosiloxane matrix with dynamically covalently cross - linked bonds and a silver nanowire network. When a sudden change in resistance value is detected, the material triggers the recombination of molecular chains for self - healing within 30 seconds, and the repair efficiency reaches 90%, ensuring the continuity and stability of signal acquisition.

[0040] Install a quantum dot sensor array on the inner side of the device, above the ultra - flexible sensing layer. The sensor has a core - shell structure, with a semiconductor quantum dot as the core and an insulating material as the shell. The sensitivity of the EEG signal is enhanced through the quantum tunneling effect, and it focuses on detecting high - frequency oscillation signals in the hippocampus. The sensor array is connected to the ultra - flexible sensing layer through micro - wires to ensure efficient and stable signal transmission.

[0041] Install a closed - loop transcranial electrical stimulation module inside the device, near key areas such as the motor cortex of the brain. This module includes electrical stimulation electrodes and a control circuit. The electrical stimulation electrodes are made of flexible materials and can fit tightly to the scalp. The module dynamically adjusts the stimulation frequency according to the real - time EEG characteristics, according to the formula where is the dynamic adjustment coefficient, with an initial value of 0.5, is the change in power in the θ band, is the amplitude in the α band, is the preset threshold; the preset threshold is 0.6. When specific changes occur in the patient's EEG activity, the module automatically adjusts the stimulation parameters to promote the recovery of nerve function.

[0042] The data processing layer uses an improved wavelet threshold function combined with independent component analysis method to perform adaptive noise reduction on quantum signals. The wavelet threshold function dynamically adjusts the threshold according to the signal energy, and independent component analysis separates and removes noise components such as power frequency interference to improve the signal purity.

[0043] Based on the noise - reduced EEG signals, calculate the motor cortex activation index, and the formula is motor cortex activation degree = where, - = 13 - 30 Hz, is a time window function, where, is the amplitude of the electroencephalogram signal in the β band (13 - 30 Hz), w(t) is the time window function, such as the Hanning window, with a window length of 1 second. This index reflects the activity level of the patient's motor cortex and provides an important basis for rehabilitation assessment.

[0044] On the side or top of the device, a wireless communication module supporting dual - mode transmission (Bluetooth 4.2 and Wi - Fi 6) is installed. The module uploads the collected data to the cloud system through the AES - 256 encryption protocol to ensure the security and stability of data transmission. At the same time, the module has low - power characteristics, extending the battery life of the device.

[0045] The device is built - in with a high - efficiency lithium battery and adopts a low - power design to ensure that the device can work continuously for more than 10 hours. At the same time, it is equipped with a charging interface and a power display device, which is convenient for users to understand the power status of the device in time and charge. The outer layer of the device is made of skin - friendly materials to reduce the irritation to the patient's skin. The internal structure is reasonably designed to ensure that each layer of components fits closely but does not overly compress the scalp, improving the comfort of the patient's wearing.

[0046] System operation process: The patient wears the wearable device, and components such as the ultra - flexible sensing layer and quantum signal acquisition layer of the device collect electroencephalogram, physiological and other data in real - time. After the data is preliminarily processed (such as noise reduction, feature extraction, etc.) by the data processing layer, it is encrypted and uploaded to the cloud system through the wireless communication module. The cloud system distributes the received data to each local hospital node for training the complication prediction model based on LSTM. After each node completes the training, it sends the model parameters to the central server. The central server dynamically weights and aggregates the parameters according to the data volume and quality, updates the global model, and distributes the new model to each node for iterative cycling. At the same time, the digital twin module fuses multi - source data such as intraoperative images to construct a dynamic three - dimensional model of the patient's brain function, and uses a physics engine to simulate the changes in brain function under different rehabilitation training schemes, such as simulating the impact of changes in electrical stimulation intensity on the cerebral edema risk index. The rehabilitation assessment module calculates evaluation indicators such as functional connection strength based on the collected data and model training results. The feedback control module generates multi - level intervention strategies such as adjusting stimulation parameters based on the assessment results and executes them through components such as the neuromodulation layer. The patient completes social interaction training through brain - controlled virtual avatars in the virtual rehabilitation community, and data such as attention concentration is real - time fed back to the system to trigger interactive actions such as scene switching, forming a closed - loop of rehabilitation training.

[0047] The same or similar reference numerals correspond to the same or similar components; The terms describing the positional relationship in the drawings are for illustrative purposes only and should not be construed as a limitation of this patent; Obviously, the above embodiments of the present invention are merely examples for clearly explaining the present invention, rather than limitations on the implementation manners of the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation manners here. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the claims of the present invention.

Claims

1. Data analysis system for postoperative rehabilitation monitoring of brain surgery patients, characterized in that, It includes: A federated learning module, a digital twin module, a rehabilitation evaluation module, a metaverse interaction module, and a feedback control module; The federated learning module uses a distributed machine learning framework to achieve multi-center model training across hospitals while protecting the privacy of patient data; The digital twin module constructs a virtual mirror of the patient, integrates intraoperative imaging data, real-time physiological parameters, and rehabilitation progress, and simulates the effects of different training plans through a physics engine; The rehabilitation evaluation module evaluates the patient's rehabilitation situation based on the collected data and the results of model training; The metaverse interaction module combines VR / AR technology to create a virtual rehabilitation community. Patients complete social interaction training by controlling virtual avatars with their brains, and the electroencephalogram data and virtual environment feedback are interacted in real time; The feedback control module adjusts the rehabilitation training plan and neuromodulation parameters according to the rehabilitation evaluation results.

2. The data analysis system for postoperative rehabilitation monitoring of brain surgery patients according to claim 1, wherein The federated learning module constructs a distributed model training architecture. Each medical institution locally trains a complication prediction model based on LSTM, and the central server aggregates global parameters through dynamic weighted aggregation, where the dynamic weight is determined according to the data volume and data quality of each medical institution; The differential privacy technology is used to add noise to protect data privacy.

3. The data analysis system for postoperative rehabilitation monitoring of brain surgery patients according to claim 2, wherein The digital twin module fuses intraoperative DTI images, real-time electroencephalogram data, and fNIRS blood oxygen parameters, and constructs a dynamic three-dimensional model of the patient's brain function through Unreal Engine; Simulate the changes in cerebrospinal fluid dynamics and calculate the cerebral edema risk index: ； wherein, is the rate of change of intracranial pressure, is the second derivative of the change in ventricular volume, and the coefficient is negatively correlated with the rate of change of ventricular volume. When the rate of change of ventricular volume increases, decreases, thereby reducing the value.

4. The data analysis system for postoperative rehabilitation monitoring of brain surgery patients according to claim 1, characterized in that, The rehabilitation evaluation module calculates the functional connectivity strength: ； Dynamic threshold adjustment: ; Among them, represents a channel pair, is the phase-locked value, is the base threshold.

5. The data analysis system for postoperative rehabilitation monitoring of brain surgery patients according to claim 1, wherein, The metaverse interaction module generates a virtual rehabilitation community scene and realizes the actions of controlling virtual avatars through VR / AR headset devices; When the attention concentration lasts for a preset time and exceeds the threshold, a new scene is unlocked and the neuromodulation intensity is reduced by a preset percentage; Among them, the attention concentration is: , is the power spectral density.

6. The data analysis system for postoperative rehabilitation monitoring of brain surgery patients according to claim 5, characterized in that, The feedback control module generates a multi-level intervention strategy, including adjusting stimulation parameters, updating the VR training difficulty, and pushing medical advice reminders.

7. A wearable device for postoperative rehabilitation monitoring of patients undergoing brain surgery, which performs data analysis through the data analysis system for postoperative rehabilitation monitoring of patients undergoing brain surgery according to any one of claims 1-6, characterized in that, It includes: A super-flexible sensing layer, a quantum signal acquisition layer, a neuromodulation layer, a data processing layer, and a wireless communication module; The super-flexible sensing layer includes: a nanoscale flexible electrode array, arranged in a serpentine topology structure, made of a composite of graphene and liquid metal, directly attached to the patient's scalp for collecting electroencephalogram signals, and the nanoscale flexible electrode array has a preset thickness and elongation rate; A self-healing conductive material layer wraps the electrode array, contains a polyborosiloxane matrix with dynamically covalently cross-linked and a silver nanowire network. When a resistance value mutation is detected, molecular chain recombination is triggered within a preset time, and the self-healing conductive material layer has a preset repair efficiency; The quantum signal acquisition layer includes: a quantum dot sensor array, with a core-shell structure, enhancing the sensitivity of electroencephalogram signals through the quantum tunneling effect, detecting weak electrical signals in deep brain regions, and the quantum dot sensor array detects high-frequency oscillation signals in the hippocampus; The neuromodulation layer includes: a closed-loop transcranial electrical stimulation module, which dynamically adjusts stimulation parameters according to real-time electroencephalogram characteristics, and the stimulation frequency satisfies a preset formula: ； Among them, is the dynamic adjustment coefficient, is the power change amount in the θ band, is the amplitude in the α band, is the preset threshold; The data processing layer is used to perform adaptive noise reduction on quantum signals, and an improved wavelet threshold function combined with independent component analysis is used to remove power frequency interference; Calculate the motor cortex activation index: ; Among them, - = 13 - 30 Hz, is the time window function; The wireless communication module supports dual-mode transmission and uploads data to the cloud system through an encryption protocol.

Images