Home rehabilitation monitoring method for patient after cerebral apoplexy interventional operation and related device

Abstract

The invention discloses a method and a related device for monitoring home rehabilitation of a patient after a cerebral apoplexy interventional operation, and the method comprises the steps: obtaining a flexible pressure sensor, an optical sensor and a three-axis acceleration sensor which are integrated on a wearable device of the patient; the blood pressure feature, the blood glucose feature and the motion feature of the patient are collected in real time and input into the stroke postoperative rehabilitation risk model, the rehabilitation risk level of the patient is obtained, and the rehabilitation risk level comprises a low risk, a medium risk and a high risk; inputting the blood pressure characteristics, the blood glucose characteristics, the motion characteristics and the rehabilitation risk levels into a stroke postoperative rehabilitation suggestion model to generate a personalized rehabilitation suggestion scheme; the blood pressure features, the blood glucose features, the motion features, the rehabilitation risk levels and the personalized rehabilitation suggestion schemes are synchronized to a cloud server, so that a doctor can remotely monitor the rehabilitation condition of the patient. Personalized rehabilitation suggestions are provided by monitoring the physiological and motion data of the patient in real time, and remote monitoring and management of the rehabilitation process of the patient are achieved.

Description

Technical Field

[0001] The present invention relates to the field of medical health rehabilitation monitoring technology, and in particular to a method and device for home rehabilitation monitoring of patients after stroke interventional surgery, and a computing device. Background Art

[0002] Post-stroke rehabilitation is crucial for patients' functional recovery and improvement of their quality of life. The traditional rehabilitation model mainly relies on inpatient rehabilitation and outpatient follow-up, which has limitations such as time-consuming, high cost, and inability to monitor in real time. Home rehabilitation, as an emerging rehabilitation model, has received increasing attention. However, existing home rehabilitation monitoring methods still have many shortcomings. For example, some wearable devices can only monitor single indicators such as step count and heart rate, which cannot fully reflect the patient's physiological state. In addition, traditional blood pressure and blood sugar monitoring requires patients to go to the hospital regularly or use invasive equipment, which is cumbersome to operate and has poor patient compliance. Although telemedicine platforms can achieve remote guidance, they lack objective physiological data support, making it difficult to conduct accurate risk assessment and personalized rehabilitation guidance.

[0003] To solve the above problems, the present invention proposes a method for home rehabilitation monitoring of patients after stroke intervention surgery, which provides personalized rehabilitation advice by real-time monitoring of the patient's physiological and movement data, thereby realizing remote monitoring and management of the patient's rehabilitation process. Summary of the Invention

[0004] In view of the above problems, the present invention provides a method and apparatus, and a computing device for home rehabilitation monitoring of patients after stroke interventional surgery.

[0005] According to one aspect of the present invention, a method for monitoring home rehabilitation of patients after stroke interventional surgery is provided, comprising:

[0006] The flexible pressure sensor, optical sensor, and triaxial accelerometer integrated into the patient's wearable device collects the patient's blood pressure, blood sugar, and motion characteristics in real time. The blood pressure characteristics include systolic peak pressure, diastolic trough pressure, mean arterial pressure, pulse pressure, and heart rate variability. The blood sugar characteristics include average blood sugar value, blood sugar fluctuation amplitude, percentage of time meeting blood sugar standards, duration of high blood sugar, and duration of low blood sugar. The motion characteristics include total steps, average cadence, activity intensity, sitting time, and activity posture.

[0007] Inputting the blood pressure characteristics, blood sugar characteristics, and movement characteristics into a post-stroke rehabilitation risk model to obtain a patient's rehabilitation risk level, wherein the rehabilitation risk level includes low risk, medium risk, and high risk; wherein the post-stroke rehabilitation risk model includes a client-side update model and a server-side aggregation model;

[0008] Inputting the blood pressure characteristics, blood sugar characteristics, movement characteristics, and rehabilitation risk level into a post-stroke rehabilitation advice model to generate a personalized rehabilitation advice plan; wherein the post-stroke rehabilitation advice model includes a state encoding layer, an action decision layer, a personalized adaptation layer, and an advice generation layer;

[0009] The blood pressure characteristics, blood sugar characteristics, exercise characteristics, rehabilitation risk level and personalized rehabilitation recommendation plan are synchronized to a cloud server for doctors to remotely monitor the patient's recovery.

[0010] In an optional manner, the flexible pressure sensor continuously measures PPG signals non-invasively at the patient's wrist or fingertip by photoplethysmography;

[0011] The optical sensor continuously monitors the glucose concentration in the patient's subcutaneous tissue fluid in a microneedle patch manner using near-infrared spectroscopy technology.

[0012] In an optional manner, the state encoding layer includes a physiological state encoder, a risk level encoder, and a context encoder; wherein the physiological state encoder uses a dual-stream CNN-LSTM network to process blood pressure, blood sugar, and motion data, and the physiological state encoder includes a spatial stream and a temporal stream;

[0013] The action decision layer includes an Actor network, a Critic network, and an action space; wherein the action space includes exercise intensity, medication time, diet, sleep quality, and follow-up visit time;

[0014] The personalized adaptation layer includes a capsule network and a dual-tower Siamese network; the capsule network includes a basic capsule and a dynamic capsule;

[0015] The suggestion generation layer includes a template engine and a T5-small model.

[0016] In an optional embodiment, the client update model runs a lightweight LSTM network on the home device to learn the user's individualized physiological rhythm, wherein the lightweight LSTM network includes 64 hidden layer units and GRU variant units;

[0017] The server-side model aggregates multiple patient data through the FedAvg algorithm to perform global model updates.

[0018] In an optional manner, the capsule network is used to extract features of different granularities, the basic capsule layer is used to extract original features, and the dynamic capsule layer is used to perform dynamic routing based on the original features to learn high-level feature representations;

[0019] The dual-tower Siamese network includes two network branches with identical structures but shared weights, which are used to process the patient's physiological state and rehabilitation advice respectively. The similarity between the patient and the rehabilitation advice is learned by minimizing the distance between the vector representations output by the two branches.

[0020] In an optional manner, the calculation formula for the rehabilitation risk level is:

[0021]

[0022] in, is the abnormality function of physiological indicators; is the time-varying weight coefficient; These are the standardized scores for blood pressure, blood sugar, and motor characteristics; are the start time and end time of monitoring respectively.

[0023] In an optional manner, the differential equation satisfied by updating the time-varying weight coefficient is:

[0024]

[0025] in, is the weight of the i-th factor; For time; is the gradient of the loss function of the i-th factor; is the change in health status.

[0026] In an optional manner, the policy gradient optimization formula of the Actor network is:

[0027]

[0028] in, is the strategy parameter gradient; For strategy expectations under is the action at time step t; is the state at time step t; is the discount factor; is the time step rewards; for The baseline function.

[0029] According to another aspect of the present invention, a device for monitoring home rehabilitation of patients after stroke interventional surgery is provided, comprising:

[0030] The data acquisition module is used to collect the patient's blood pressure characteristics, blood glucose characteristics, and motion characteristics in real time through a flexible pressure sensor, optical sensor, and triaxial accelerometer integrated into the patient's wearable device. The blood pressure characteristics include systolic peak pressure, diastolic trough pressure, mean arterial pressure, pulse pressure, and heart rate variability indicators. The blood glucose characteristics include mean blood glucose value, blood glucose fluctuation amplitude, percentage of time in which blood glucose reaches the target, duration of high blood glucose, and duration of low blood glucose. The motion characteristics include total number of steps, average cadence, activity intensity, sitting time, and activity posture.

[0031] a rehabilitation risk assessment module, configured to input the blood pressure characteristics, blood sugar characteristics, and movement characteristics into a post-stroke rehabilitation risk model to obtain a patient's rehabilitation risk level, wherein the rehabilitation risk level includes low risk, medium risk, and high risk; wherein the post-stroke rehabilitation risk model includes a client-side update model and a server-side aggregation model;

[0032] A rehabilitation advice generation module, configured to input the blood pressure characteristics, blood sugar characteristics, movement characteristics, and rehabilitation risk level into a post-stroke rehabilitation advice model to generate a personalized rehabilitation advice plan; wherein the post-stroke rehabilitation advice model includes a state encoding layer, an action decision layer, a personalized adaptation layer, and an advice generation layer;

[0033] The data synchronization module is used to synchronize the blood pressure characteristics, blood sugar characteristics, exercise characteristics, rehabilitation risk level and personalized rehabilitation recommendation plan to the cloud server for doctors to remotely monitor the patient's rehabilitation status.

[0034] According to another aspect of the present invention, there is provided a computing device, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus;

[0035] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the above-mentioned method for home rehabilitation monitoring of patients after stroke intervention surgery.

[0036] According to the solution provided by the present invention, the patient's blood pressure characteristics, blood sugar characteristics and movement characteristics are collected in real time through the flexible pressure sensor, optical sensor and three-axis acceleration sensor integrated on the patient's wearable device; wherein, the blood pressure characteristics include systolic pressure peak value, diastolic pressure trough value, mean arterial pressure, pulse pressure and heart rate variability index; the blood sugar characteristics include blood sugar average value, blood sugar fluctuation amplitude, percentage of blood sugar reaching standard time, high blood sugar duration and low blood sugar duration; the movement characteristics include total number of steps, average step frequency, activity intensity, sitting time and activity posture; the blood pressure characteristics, blood sugar characteristics and movement characteristics are input into the stroke postoperative rehabilitation risk model to obtain The patient's rehabilitation risk level is determined, and the rehabilitation risk level includes low risk, medium risk, and high risk; wherein the post-stroke rehabilitation risk model includes a client update model and a server aggregation model; the blood pressure characteristics, blood sugar characteristics, movement characteristics, and rehabilitation risk level are input into a post-stroke rehabilitation suggestion model to generate a personalized rehabilitation suggestion plan; wherein the post-stroke rehabilitation suggestion model includes a state encoding layer, an action decision layer, a personalized adaptation layer, and a suggestion generation layer; the blood pressure characteristics, blood sugar characteristics, movement characteristics, rehabilitation risk level, and personalized rehabilitation suggestion plan are synchronized to a cloud server for doctors to remotely monitor the patient's rehabilitation status. The present invention provides personalized rehabilitation suggestions by monitoring the patient's physiological and movement data in real time, thereby enabling doctors to remotely monitor and manage the patient's rehabilitation process.

[0037] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0039] Figure 1 A schematic flow chart of a method for monitoring home rehabilitation of patients after stroke interventional surgery according to an embodiment of the present invention is shown;

[0040] Figure 2 A schematic diagram of a framework of a device for monitoring home rehabilitation of patients after stroke interventional surgery according to an embodiment of the present invention is shown;

[0041] Figure 3 A schematic structural diagram of a computing device according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0042] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0043] Figure 1 FIG2 shows a flow chart of a method for monitoring home rehabilitation of patients after stroke intervention surgery according to an embodiment of the present invention. Figure 1 As shown, the following steps are included:

[0044] Step S101, through the flexible pressure sensor, optical sensor and three-axis acceleration sensor integrated on the patient's wearable device, the patient's blood pressure characteristics, blood sugar characteristics and movement characteristics are collected in real time; wherein, the blood pressure characteristics include systolic blood pressure peak value, diastolic blood pressure trough value, mean arterial pressure, pulse pressure and heart rate variability index; the blood sugar characteristics include blood sugar average value, blood sugar fluctuation amplitude, percentage of blood sugar reaching standard time, high blood sugar duration and low blood sugar duration; the movement characteristics include total number of steps, average cadence, activity intensity, sitting time and activity posture.

[0045] In this embodiment, the wearable device can realize real-time and continuous monitoring of the patient's physiological data, avoid the errors that may exist in traditional intermittent measurements, and more accurately reflect the patient's physiological change trends. Non-invasive measurements are performed using flexible pressure sensors, optical sensors, and three-axis accelerometers to avoid the discomfort and risks brought by traditional invasive measurements, thereby improving patient compliance. Microneedle patches can be considered minimally invasive, which improves convenience and patient compliance compared to finger pricking to measure blood sugar. Blood pressure characteristics include not only traditional systolic and diastolic blood pressure, but also mean arterial pressure, pulse pressure, and heart rate variability indicators, which can more deeply reflect the patient's cardiovascular function. In addition to the average blood sugar value, blood sugar characteristics also include blood sugar fluctuation amplitude, percentage of time when blood sugar reaches the standard, duration of high blood sugar, and duration of low blood sugar, which more comprehensively evaluate the patient's blood sugar control. Exercise characteristics include total number of steps, average cadence, activity intensity, sitting time, and activity posture.

[0046] For example, flexible pressure sensors can be integrated into wristbands or fingertip wearable devices to ensure close contact with the patient's skin for accurate measurement of pulse wave signals (PPG). Optical sensors can be attached to the patient's skin using microneedle patches, continuously monitoring glucose concentration in subcutaneous tissue fluid using near-infrared spectroscopy. Three-axis accelerometers can be integrated into wearable devices to monitor the patient's movement status (such as step count, cadence, and activity intensity).

[0047] Step S102: Input the blood pressure characteristics, blood sugar characteristics, and movement characteristics into a post-stroke rehabilitation risk model to obtain the patient's rehabilitation risk level, which includes low risk, medium risk, and high risk. The post-stroke rehabilitation risk model includes a client update model and a server aggregation model.

[0048] In this embodiment, the client-side update model learns the user's individualized circadian rhythm, making risk assessment more realistic. The server-side aggregated model leverages common data from multiple users to enhance the model's generalization capabilities and improve its adaptability to new users. Real-time collected physiological characteristic data is quickly input into the risk model to promptly identify problems that arise during the patient's recovery and make appropriate rehabilitation adjustments. Furthermore, the client-side update model and the server-side aggregated model enable model training without leaking the user's original data.

[0049] In an optional embodiment, the client update model runs a lightweight LSTM network on the home device to learn the user's individualized physiological rhythm, wherein the lightweight LSTM network includes 64 hidden layer units and GRU variant units;

[0050] The server-side model aggregates multiple patient data through the FedAvg algorithm to perform global model updates.

[0051] In this embodiment, the client's lightweight LSTM network learns and models the individual physiological rhythm of each user to capture individual differences. The server aggregates the data of multiple patients through the FedAvg algorithm to update the global model, avoiding the problem of insufficient model generalization ability caused by relying solely on individual data. The data does not need to be uploaded to the server, but is retained on the local device for training, thereby protecting the user's privacy. The server only aggregates the parameters updated by the local model and does not directly access the original data. In addition, the client is a lightweight LSTM network with low computing resource consumption, which is suitable for running on home devices. The server only needs to aggregate the model parameters, and the communication volume is much smaller than directly transmitting the original data.

[0052] In an optional manner, the calculation formula for the rehabilitation risk level is:

[0053]

[0054] in, is the abnormality function of physiological indicators; is the time-varying weight coefficient; These are the standardized scores for blood pressure, blood sugar, and motor characteristics; are the start time and end time of monitoring respectively.

[0055] In this embodiment, the introduction of a physiological indicator abnormality function quantifies the degree of abnormality of different physiological indicators, thereby more accurately assessing a patient's recovery risk. The time integral considers changes in a patient's physiological indicators over a certain period of time, rather than relying solely on data at a single point in time. This better captures long-term trends in a patient's recovery process, thereby providing more targeted rehabilitation recommendations.

[0056] In an optional manner, the differential equation satisfied by updating the time-varying weight coefficient is:

[0057]

[0058] in, is the weight of the i-th factor; For time; is the gradient of the loss function of the i-th factor; is the change in health status.

[0059] In this embodiment, real-time updates of the time-varying weight coefficients through differential equations dynamically adapt to changes in the patient's health status, thereby providing a more accurate rehabilitation risk assessment. The change in health status considers the patient's overall health trend, preventing fluctuations in a single indicator from interfering with weight adjustments.

[0060] Step S103: Input the blood pressure characteristics, blood sugar characteristics, movement characteristics and rehabilitation risk level into a post-stroke rehabilitation advice model to generate a personalized rehabilitation advice plan; wherein, the post-stroke rehabilitation advice model includes a state encoding layer, an action decision layer, a personalized adaptation layer and an advice generation layer.

[0061] In this embodiment, the state encoding layer and personalized adaptation layer dynamically adapt to changes in the patient's physiological state, ensuring the timeliness and adaptability of rehabilitation recommendations. The action decision layer utilizes an actor network and a critic network to intelligently select the optimal rehabilitation action (such as exercise intensity, medication timing, diet, sleep quality, and follow-up appointment schedule) based on the patient's physiological state and rehabilitation goals. The recommendation generation layer utilizes template engines and T5-small models to quickly generate clear and understandable rehabilitation recommendations, improving patient compliance.

[0062] In an optional manner, the state encoding layer includes a physiological state encoder, a risk level encoder, and a context encoder; wherein the physiological state encoder uses a dual-stream CNN-LSTM network to process blood pressure, blood sugar, and motion data, and the physiological state encoder includes a spatial stream and a temporal stream;

[0063] The action decision layer includes an Actor network, a Critic network, and an action space; wherein the action space includes exercise intensity, medication time, diet, sleep quality, and follow-up visit time;

[0064] The personalized adaptation layer includes a capsule network and a dual-tower Siamese network; the capsule network includes a basic capsule and a dynamic capsule;

[0065] The suggestion generation layer includes a template engine and a T5-small model.

[0066] In this embodiment, the dual-stream CNN-LSTM network in the state encoding layer effectively integrates blood pressure, blood glucose, and exercise data (both time series and non-time series). The action decision layer uses an actor-critic network to model the process of generating rehabilitation recommendations as a reinforcement learning problem. The actor network selects the optimal rehabilitation action, while the critic network evaluates the quality of these actions, thereby continuously optimizing the actor network's strategy. The personalized adaptation layer further enhances the model's ability to model individual differences through the use of capsule networks and dual-tower Siamese networks. The capsule network learns features of varying granularity to capture key information, while the Siamese network learns similarities between patient status and rehabilitation recommendations to recommend more suitable solutions. The recommendation generation layer combines a template engine with the T5-small model to ensure structured and standardized recommendations, making them easier to understand and more practical.

[0067] In an optional manner, the capsule network is used to extract features of different granularities, the basic capsule layer is used to extract original features, and the dynamic capsule layer is used to perform dynamic routing based on the original features to learn high-level feature representations;

[0068] The dual-tower Siamese network includes two network branches with identical structures but shared weights, which are used to process the patient's physiological state and rehabilitation advice respectively. The similarity between the patient and the rehabilitation advice is learned by minimizing the distance between the vector representations output by the two branches.

[0069] In this embodiment, a dual-tower Siamese network calculates the similarity between the patient's physiological state and each rehabilitation recommendation, thereby selecting the most appropriate one. For example, if a patient's blood pressure and blood sugar data indicate a high-risk state, while their exercise profile indicates low activity, personalized rehabilitation recommendations emphasizing increased exercise and dietary adjustments may be selected to help the patient recover.

[0070] In an optional manner, the policy gradient optimization formula of the Actor network is:

[0071]

[0072] in, is the strategy parameter gradient; For strategy expectations under is the action at time step t; is the state at time step t; is the discount factor; is the time step rewards; for The baseline function.

[0073] In this example, the policy gradient method directly optimizes policy parameters to maximize expected reward. Unlike value function methods, this approach does not require estimating a value function but instead directly adjusts the policy to improve performance. This allows the actor network to more directly control behavior and is applicable to continuous action spaces.

[0074] In step S104, the blood pressure characteristics, blood sugar characteristics, exercise characteristics, rehabilitation risk level and personalized rehabilitation recommendation plan are synchronized to a cloud server for doctors to remotely monitor the patient's rehabilitation status.

[0075] In this embodiment, the patient data is uploaded to the cloud in real time, and the doctor can remotely monitor the patient's various physiological indicators, movement data, risk level and rehabilitation suggestions in real time, detect abnormal conditions in time and intervene to avoid emergencies, thereby improving the rehabilitation effect, rehabilitation enthusiasm and compliance.

[0076] According to the solution provided by the present invention, the patient's blood pressure characteristics, blood sugar characteristics and movement characteristics are collected in real time through the flexible pressure sensor, optical sensor and three-axis acceleration sensor integrated on the patient's wearable device; wherein, the blood pressure characteristics include systolic pressure peak value, diastolic pressure trough value, mean arterial pressure, pulse pressure and heart rate variability index; the blood sugar characteristics include blood sugar average value, blood sugar fluctuation amplitude, percentage of blood sugar reaching standard time, high blood sugar duration and low blood sugar duration; the movement characteristics include total number of steps, average step frequency, activity intensity, sitting time and activity posture; the blood pressure characteristics, blood sugar characteristics and movement characteristics are input into the stroke postoperative rehabilitation risk model to obtain The patient's rehabilitation risk level is determined, and the rehabilitation risk level includes low risk, medium risk, and high risk; wherein the post-stroke rehabilitation risk model includes a client update model and a server aggregation model; the blood pressure characteristics, blood sugar characteristics, movement characteristics, and rehabilitation risk level are input into a post-stroke rehabilitation suggestion model to generate a personalized rehabilitation suggestion plan; wherein the post-stroke rehabilitation suggestion model includes a state encoding layer, an action decision layer, a personalized adaptation layer, and a suggestion generation layer; the blood pressure characteristics, blood sugar characteristics, movement characteristics, rehabilitation risk level, and personalized rehabilitation suggestion plan are synchronized to a cloud server for doctors to remotely monitor the patient's rehabilitation status. The present invention provides personalized rehabilitation suggestions by monitoring the patient's physiological and movement data in real time, thereby enabling doctors to remotely monitor and manage the patient's rehabilitation process.

[0077] Figure 2 The schematic diagram of the framework of the device for monitoring home rehabilitation of patients after stroke interventional surgery according to an embodiment of the present invention is shown. The device for monitoring home rehabilitation of patients after stroke interventional surgery includes:

[0078] Data acquisition module 210 is configured to collect, in real time, the patient's blood pressure characteristics, blood glucose characteristics, and motion characteristics through flexible pressure sensors, optical sensors, and triaxial accelerometers integrated into the patient's wearable device. The blood pressure characteristics include systolic peak pressure, diastolic trough pressure, mean arterial pressure, pulse pressure, and heart rate variability. The blood glucose characteristics include mean blood glucose value, blood glucose fluctuation amplitude, percentage of time meeting blood glucose standards, duration of high blood glucose, and duration of low blood glucose. The motion characteristics include total steps, average cadence, activity intensity, sitting time, and activity posture.

[0079] The rehabilitation risk assessment module 220 is configured to input the blood pressure characteristics, blood glucose characteristics, and movement characteristics into a post-stroke rehabilitation risk model to obtain a patient's rehabilitation risk level, wherein the rehabilitation risk level includes low risk, medium risk, and high risk. The post-stroke rehabilitation risk model includes a client-side update model and a server-side aggregation model.

[0080] The rehabilitation advice generation module 230 is configured to input the blood pressure characteristics, blood sugar characteristics, movement characteristics, and rehabilitation risk level into a post-stroke rehabilitation advice model to generate a personalized rehabilitation advice plan; wherein the post-stroke rehabilitation advice model includes a state encoding layer, an action decision layer, a personalized adaptation layer, and an advice generation layer;

[0081] The data synchronization module 240 is used to synchronize the blood pressure characteristics, blood sugar characteristics, exercise characteristics, rehabilitation risk level and personalized rehabilitation recommendation plan to the cloud server for doctors to remotely monitor the patient's rehabilitation status.

[0082] Figure 3 The schematic diagram of the structure of the computing device embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the computing device.

[0083] like Figure 3 As shown, the computing device may include: a processor (processor) 302, a communication interface (Communications Interface) 304, a memory (memory) 306, and a communication bus 308.

[0084] Processor 302, communication interface 304, and memory 306 communicate with each other via communication bus 308. Communication interface 304 is used to communicate with other devices, such as client devices or other server network elements. Processor 302 is used to execute program 310, which specifically performs the steps described in the aforementioned embodiment of the method for monitoring home rehabilitation of patients after stroke interventional surgery.

[0085] Specifically, the program 310 may include program codes, which include computer operation instructions.

[0086] Processor 302 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The one or more processors included in a computing device may be of the same type, such as one or more CPUs, or different types, such as one or more CPUs and one or more ASICs.

[0087] The memory 306 is used to store the program 310. The memory 306 may include a high-speed RAM memory, or may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0088] According to the solution provided by the present invention, the patient's blood pressure characteristics, blood sugar characteristics and movement characteristics are collected in real time through the flexible pressure sensor, optical sensor and three-axis acceleration sensor integrated on the patient's wearable device; wherein, the blood pressure characteristics include systolic pressure peak value, diastolic pressure trough value, mean arterial pressure, pulse pressure and heart rate variability index; the blood sugar characteristics include blood sugar average value, blood sugar fluctuation amplitude, percentage of blood sugar reaching standard time, high blood sugar duration and low blood sugar duration; the movement characteristics include total number of steps, average step frequency, activity intensity, sitting time and activity posture; the blood pressure characteristics, blood sugar characteristics and movement characteristics are input into the stroke postoperative rehabilitation risk model to obtain The patient's rehabilitation risk level is determined, and the rehabilitation risk level includes low risk, medium risk, and high risk; wherein the post-stroke rehabilitation risk model includes a client update model and a server aggregation model; the blood pressure characteristics, blood sugar characteristics, movement characteristics, and rehabilitation risk level are input into a post-stroke rehabilitation suggestion model to generate a personalized rehabilitation suggestion plan; wherein the post-stroke rehabilitation suggestion model includes a state encoding layer, an action decision layer, a personalized adaptation layer, and a suggestion generation layer; the blood pressure characteristics, blood sugar characteristics, movement characteristics, rehabilitation risk level, and personalized rehabilitation suggestion plan are synchronized to a cloud server for doctors to remotely monitor the patient's rehabilitation status. The present invention provides personalized rehabilitation suggestions by monitoring the patient's physiological and movement data in real time, thereby enabling doctors to remotely monitor and manage the patient's rehabilitation process.

[0089] Those skilled in the art will appreciate that the modules in the devices of the embodiments can be adaptively modified and deployed in one or more devices different from the embodiments. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and furthermore, they can be divided into multiple sub-modules, sub-units, or sub-components. All features disclosed in this specification (including the accompanying claims, abstract, and drawings), as well as all processes or units of any method or device disclosed therein, can be combined in any combination, unless at least some of such features and / or processes or units are mutually exclusive. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature serving the same, equivalent, or similar purpose. Furthermore, those skilled in the art will appreciate that although some embodiments herein include certain features and not others included in other embodiments, combinations of features from different embodiments are intended to fall within the scope of the present invention and form different embodiments. For example, in the claims below, any of the claimed embodiments may be used in any combination. The present invention may be implemented by means of hardware comprising several distinct elements and by means of a suitably programmed computer. In a unit claim that lists several means, several of these means may be embodied by the same hardware item. Unless otherwise specified, the steps in the above embodiments should not be understood as limiting the order of execution.

Claims

1. A method for monitoring home rehabilitation of patients after stroke intervention surgery, characterized in that: include: The flexible pressure sensor, optical sensor, and triaxial accelerometer integrated into the patient's wearable device collects the patient's blood pressure, blood sugar, and motion characteristics in real time. The blood pressure characteristics include systolic peak pressure, diastolic trough pressure, mean arterial pressure, pulse pressure, and heart rate variability. The blood sugar characteristics include average blood sugar value, blood sugar fluctuation amplitude, percentage of time meeting blood sugar standards, duration of high blood sugar, and duration of low blood sugar. The motion characteristics include total steps, average cadence, activity intensity, sitting time, and activity posture. Inputting the blood pressure characteristics, blood sugar characteristics, and movement characteristics into a post-stroke rehabilitation risk model to obtain a patient's rehabilitation risk level, wherein the rehabilitation risk level includes low risk, medium risk, and high risk; wherein the post-stroke rehabilitation risk model includes a client-side update model and a server-side aggregation model; Inputting the blood pressure characteristics, blood sugar characteristics, movement characteristics, and rehabilitation risk level into a post-stroke rehabilitation advice model to generate a personalized rehabilitation advice plan; wherein the post-stroke rehabilitation advice model includes a state encoding layer, an action decision layer, a personalized adaptation layer, and an advice generation layer; The blood pressure characteristics, blood sugar characteristics, exercise characteristics, rehabilitation risk level and personalized rehabilitation recommendation plan are synchronized to a cloud server for doctors to remotely monitor the patient's recovery.

2. The method for home rehabilitation monitoring of patients after stroke interventional surgery according to claim 1, characterized in that: The flexible pressure sensor continuously measures PPG signals non-invasively at the patient's wrist or fingertip using a photoplethysmography method; The optical sensor continuously monitors the glucose concentration in the patient's subcutaneous tissue fluid in a microneedle patch manner using near-infrared spectroscopy technology.

3. The method for home rehabilitation monitoring of patients after stroke interventional surgery according to claim 1, characterized in that: The state encoding layer includes a physiological state encoder, a risk level encoder, and a context encoder; wherein the physiological state encoder uses a dual-stream CNN-LSTM network to process blood pressure, blood sugar, and exercise data, and the physiological state encoder includes a spatial stream and a temporal stream; The action decision layer includes an Actor network, a Critic network, and an action space; wherein the action space includes exercise intensity, medication time, diet, sleep quality, and follow-up visit time; The personalized adaptation layer includes a capsule network and a dual-tower Siamese network; the capsule network includes a basic capsule and a dynamic capsule; The suggestion generation layer includes a template engine and a T5-small model.

4. The method for home rehabilitation monitoring of patients after stroke interventional surgery according to claim 1, characterized in that: The client update model runs a lightweight LSTM network on the home device to learn the user's individualized physiological rhythm. The lightweight LSTM network includes 64 hidden layer units and GRU variant units; The server-side model aggregates multiple patient data through the FedAvg algorithm to perform global model updates.

5. The method for home rehabilitation monitoring of patients after stroke interventional surgery according to claim 3, characterized in that: The capsule network is used to extract features of different granularities, the basic capsule layer is used to extract original features, and the dynamic capsule layer is used to perform dynamic routing based on the original features to learn high-level feature representations; The dual-tower Siamese network includes two network branches with identical structures but shared weights, which are used to process the patient's physiological state and rehabilitation advice respectively. The similarity between the patient and the rehabilitation advice is learned by minimizing the distance between the vector representations output by the two branches.

6. The method for home rehabilitation monitoring of patients after stroke interventional surgery according to claim 1, characterized in that: The calculation formula for the rehabilitation risk level is: ； in, is the abnormality function of physiological indicators; is the time-varying weight coefficient; These are the standardized scores for blood pressure, blood sugar, and motor characteristics; are the start time and end time of monitoring respectively.

7. The method for home rehabilitation monitoring of patients after stroke interventional surgery according to claim 6, characterized in that: The differential equation satisfied by updating the time-varying weight coefficient is: ； in, is the weight of the i-th factor; For time; is the gradient of the loss function of the i-th factor; is the change in health status.

8. The method for home rehabilitation monitoring of patients after stroke interventional surgery according to claim 3, characterized in that: The policy gradient optimization formula of the Actor network is: ； in, is the strategy parameter gradient; For strategy expectations under is the action at time step t; is the state at time step t; is the discount factor; is the time step rewards; for The baseline function.

9. A home rehabilitation monitoring device for patients after stroke intervention surgery, characterized in that: include: The data acquisition module is used to collect the patient's blood pressure characteristics, blood glucose characteristics, and motion characteristics in real time through a flexible pressure sensor, optical sensor, and triaxial accelerometer integrated into the patient's wearable device. The blood pressure characteristics include systolic peak pressure, diastolic trough pressure, mean arterial pressure, pulse pressure, and heart rate variability indicators. The blood glucose characteristics include mean blood glucose value, blood glucose fluctuation amplitude, percentage of time in which blood glucose reaches the target, duration of high blood glucose, and duration of low blood glucose. The motion characteristics include total number of steps, average cadence, activity intensity, sitting time, and activity posture. a rehabilitation risk assessment module, configured to input the blood pressure characteristics, blood sugar characteristics, and movement characteristics into a post-stroke rehabilitation risk model to obtain a patient's rehabilitation risk level, wherein the rehabilitation risk level includes low risk, medium risk, and high risk; wherein the post-stroke rehabilitation risk model includes a client-side update model and a server-side aggregation model; A rehabilitation advice generation module, configured to input the blood pressure characteristics, blood sugar characteristics, movement characteristics, and rehabilitation risk level into a post-stroke rehabilitation advice model to generate a personalized rehabilitation advice plan; wherein the post-stroke rehabilitation advice model includes a state encoding layer, an action decision layer, a personalized adaptation layer, and an advice generation layer; The data synchronization module is used to synchronize the blood pressure characteristics, blood sugar characteristics, exercise characteristics, rehabilitation risk level and personalized rehabilitation recommendation plan to the cloud server for doctors to remotely monitor the patient's rehabilitation status.

10. A computing device comprising: A processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the above-mentioned method for home rehabilitation monitoring of patients after stroke intervention surgery.

Images