Interaction control method and system for skeletal muscle training of ICU bedridden patient
Through neural network model and virtual reality technology, the patient's motor intentions and adjustment of training intensity was solved, and the problem of poor rehabilitation training in the ICU environment was solved, personalized and efficient skeletal muscle training was achieved, and the risk of bed-related complications was reduced.
Patent Information
- Application Number
- CN202410940650.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-13
- Publication Date
- 2024-11-08
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art is difficult to accurately identify the exercise intentions of bedridden patients and dynamically adjust the training intensity in an ICU environment, resulting in poor rehabilitation training results and lack of personalization and efficiency.
The neural network model is used to combine virtual reality technology to collect patient limb movement data, identify movement intentions, and adjust the induction intensity in the virtual reality situation based on the gap calculation to achieve dynamic adaptive training.
It provides a personalized and efficient skeletal muscle training program, which significantly improves the patient's rehabilitation effect, reduces the risk of long-term bed-related complications, and optimizes the allocation of medical resources.
Smart Images

Figure CN118919015A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of smart medical care, and more specifically, to an interactive control method and system for skeletal muscle training of bedridden patients in an ICU. Background Art
[0002] With the development of medical technology, the rehabilitation of long-term bedridden patients in the intensive care unit (ICU) has received increasing attention. Long-term bed rest can cause patients to suffer from a series of complications such as muscle atrophy, joint stiffness, and osteoporosis, which seriously affects their rehabilitation process and quality of life, and leads to acquired neuromuscular dysfunction in critically ill patients, namely ICU acquired weakness (ICU-AW). Traditional physical therapy often relies on manual operation, which makes it difficult to achieve personalized, high-intensity continuous training. In addition, in the ICU environment, the complexity of the patient's condition limits the applicability and efficiency of traditional rehabilitation methods.
[0003] In recent years, virtual reality (VR) technology has been gradually introduced into the field of rehabilitation medicine because it can provide an immersive and customizable rehabilitation environment. By simulating various scenarios, VR technology can stimulate patients' interest in sports and increase the fun and participation of training. However, most existing technologies focus on standardized rehabilitation training procedures and lack dynamic adaptability to individual differences of patients, especially in identifying and responding to patients' specific movement intentions, which limits the effectiveness of training programs and the maximization of patients' motor potential.
[0004] Therefore, how to design an interactive rehabilitation training method for bedridden patients in the ICU environment that can not only accurately identify exercise intentions but also dynamically adjust training intensity according to the patient's real-time status has become an urgent problem to be solved. Summary of the invention
[0005] In order to solve the above-mentioned technical problems, the present application is proposed to provide an interactive control method and system for skeletal muscle training for bedridden patients in ICU, which can not only accurately identify exercise intentions but also dynamically adjust the training intensity according to the real-time status of the patients.
[0006] In a first aspect, the present invention provides an interactive control method for skeletal muscle training of ICU bedridden patients, comprising: when a limb of a patient in the ICU generates a first movement, collecting first movement data of bones and / or muscles corresponding to the limb of the patient; inputting the first movement data into a preset first neural network model, and identifying the patient's movement intention through the first neural network model; constructing a virtual reality scenario according to the patient's movement intention, the virtual reality scenario having a plurality of elements, the plurality of elements having preset attribute values, and being used to induce the patient to perform a second movement corresponding to the movement intention according to a preset induction intensity; collecting second movement data of bones and / or muscles corresponding to the limb of the patient when the patient performs the second movement; querying standard movement data corresponding to the second movement, and calculating the second movement. The method comprises the following steps: measuring the difference between the movement data of the bones and / or muscles corresponding to the patient's limbs and the standard movement data corresponding to the second movement, so as to reflect the level of movement disorder of the patient when performing the second movement; inputting the difference between the movement data of the bones and / or muscles corresponding to the patient's limbs and the standard movement data corresponding to the second movement into a preset second neural network model corresponding to the virtual reality scenario, and calculating the attribute value that at least one element of the virtual reality scenario should have through the second neural network model; adjusting the attribute value of the at least one element in the virtual reality scenario according to the attribute value that the at least one element of the virtual reality scenario should have to change the induction intensity; providing the virtual reality scenario to the user, and inducing the patient to perform the second movement again according to the changed induction intensity through the multiple elements in the virtual reality scenario.
[0007] Optionally, in the aforementioned interactive control method for skeletal muscle training of ICU bedridden patients, the collecting of first motion data of bones and / or muscles corresponding to the patient's limbs includes: continuously taking multiple consecutive images of the patient's limbs within a preset time period; identifying the positions of multiple joints and muscle groups on the patient's limbs in the multiple consecutive images; calculating the angular velocity vectors and angular acceleration vectors of the multiple joints on the patient's limbs according to the position changes of the multiple joints and muscle groups on the patient's limbs; calculating the joint torques and power outputs of the multiple joints on the patient's limbs, the joint torque τ of the mth joint among the multiple joints being equal to I m -1 (ω m I m ω m +α m )-g, the power output of the mth joint P = τω m , where ω m represents the angular velocity vector of the mth joint, Represents ω m The derivative with respect to time, αm represents the angular acceleration vector of the mth joint, I m represents a preset moment of inertia matrix, g represents a preset gravity vector; the joint torque and power output of multiple joints on the patient's limbs are used as the first action data for inputting the first neural network model.
[0008] Optionally, the aforementioned interactive control method for skeletal muscle training of ICU bedridden patients, before calculating the joint torque and power output of multiple joints on the patient's limbs, further includes: extracting the material density, center of mass and coverage area of the multiple joints and muscle groups on the patient's limbs from a preset skeletal and muscle system model constructed for the patient's limbs; calculating the moment of inertia matrix for the multiple joints on the patient's limbs based on the material density, center of mass and coverage area of the multiple joints and muscle groups on the patient's limbs, as well as the coverage area of the multiple joints and muscle groups, wherein the moment of inertia matrix I corresponding to the mth joint m =∫ρ(r)(||r|| 2 -rc)dV, V represents the coverage area of the m-th joint and its muscle group, r represents a point in the coverage area of the m-th joint and its muscle group, c represents the center of mass of the m-th joint and its muscle group, and ρ(r) represents the material density of the m-th joint and its muscle group.
[0009] Optionally, in the aforementioned interactive control method for skeletal muscle training for ICU bedridden patients, the collecting of first motion data of bones and / or muscles corresponding to the patient's limbs includes: collecting, through a preset pressure sensor array, the pressure generated when the patient's limbs perform the first motion; based on the known pressure interference caused by the patient's non-purposeful motions, correcting the pressure generated when the patient's limbs perform the first motion; and using the corrected pressure generated when the patient's limbs perform the first motion as the first motion data.
[0010] Optionally, the aforementioned interactive control method for skeletal muscle training of ICU bedridden patients, before inputting the first action data into a preset first neural network model, further comprises: when the first neural network model is trained, calculating the loss function of the first neural network model Where E represents the total number of preset movement intention types, represents the predicted probability corresponding to the nth motion intention calculated during the training of the first neural network model, y n represents the true probability corresponding to the nth motion intention, w nThe weight coefficient corresponding to the nth motion intention is set based on the known scarcity of the nth motion intention; and the parameters of the first neural network model are adjusted according to the size of the loss function L of the first neural network model.
[0011] Optionally, the aforementioned interactive control method for skeletal muscle training of ICU bedridden patients, before adjusting the parameters of the first neural network model according to the size of the loss function L of the first neural network model, further includes: setting a loss correction function for the first neural network model Where T represents the number of trainings currently completed, and Respectively represent the output results of the first neural network model during the tth and t+1th training. and At the same time, The value of is 1, when and At the same time, The value of is 0; based on the loss correction function L cor , the loss function L is modified.
[0012] Optionally, the aforementioned interactive control method for skeletal muscle training of ICU bedridden patients, before inputting the difference between the motion data of the bones and / or muscles corresponding to the patient's limbs and the standard motion data corresponding to the second motion into a preset second neural network model corresponding to the virtual reality scenario, further comprises: setting an original change step length s for any parameter in the second neural network model o When the second neural network model completes the ath training, calculate the change step length of any parameter in the second neural network model Wherein, d is a preset reference value; according to the change step d a , change any parameter in the second neural network model.
[0013] Optionally, the aforementioned interactive control method for skeletal muscle training for ICU bedridden patients, wherein the virtual reality scenario is constructed according to the patient's movement intention, further comprises: extracting the patient's preference characteristics from the patient's historical behavior records; and changing the type of any one of the multiple elements according to the patient's preference characteristics, so that the element of the changed type has the preference characteristics.
[0014] Optionally, in the aforementioned interactive control method for skeletal muscle training of ICU bedridden patients, the adjusting of the attribute value of at least one element in the virtual situation according to the attribute value that at least one element of the virtual reality situation should have to change the induction intensity comprises: setting a sensitivity coefficient μ for the patient according to the physiological reflex data of the patient to reflect the sensitivity of the patient's limbs when receiving induction; setting the attribute adjustment amplitude according to the sensitivity coefficient of the patient Wherein, x represents the attribute value that at least one element of the virtual reality scenario should have, x0 represents the actual attribute value of at least one element of the virtual reality scenario, β and γ are preset values; the attribute value of at least one element of the virtual reality scenario is continuously adjusted according to the attribute adjustment range until the expected attribute value is reached.
[0015] In a second aspect, the present invention provides an interactive control system for skeletal muscle training of ICU bedridden patients, comprising: a first motion data acquisition module, which acquires first motion data of bones and / or muscles corresponding to the limbs of a patient in the ICU when the limbs of the patient produce a first motion; a motion intention recognition module, which inputs the first motion data into a preset first neural network model, and recognizes the patient's motion intention through the first neural network model; a virtual reality scenario construction module, which constructs a virtual reality scenario according to the patient's motion intention, wherein the virtual reality scenario has a plurality of elements, and the plurality of elements have preset attribute values, and are used to induce the patient to perform a second motion corresponding to the motion intention according to a preset induction intensity; a second motion data acquisition module, which acquires second motion data of bones and / or muscles corresponding to the limbs of the patient when the patient performs the second motion; a gap calculation module, which queries the standard corresponding to the second motion A quasi-action data module is used to calculate the difference between the second action data and the standard action data corresponding to the second action, so as to reflect the movement disorder level of the patient when performing the second action; an attribute value analysis module is used to input the difference between the action data of the bones and / or muscles corresponding to the patient's limbs and the standard action data corresponding to the second action into a preset second neural network model corresponding to the virtual reality scenario, and calculate the attribute value that at least one element of the virtual reality scenario should have through the second neural network model; an attribute value adjustment module is used to adjust the attribute value of at least one element in the virtual scenario according to the attribute value that at least one element of the virtual reality scenario should have, so as to change the induction intensity; an induction module is used to provide the virtual reality scenario to the user, and induce the patient to perform the second action again according to the changed induction intensity through the multiple elements in the virtual reality scenario.
[0016] The above one or more technical solutions of the present invention have at least one or more of the following beneficial effects:
[0017] The present invention realizes the integration of biomechanical analysis, deep learning, sensor technology and virtual reality technology, demonstrates the innovative application of interdisciplinary technology in the field of medical rehabilitation, and provides a more scientific, efficient and personalized skeletal muscle training program for ICU bedridden patients, significantly improving the patients' rehabilitation effects and reducing the risk of complications associated with long-term bed rest. At the same time, it opens up new ways for the effective allocation and utilization of medical resources. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.
[0019] Figure 1 is a flow chart of an interactive control method for skeletal muscle training for ICU bedridden patients according to an embodiment of the present application;
[0020] Figure 2 is a partial flow chart of an interactive control method for skeletal muscle training for ICU bedridden patients according to an embodiment of the present application;
[0021] Figure 3 is another partial flow chart of an interactive control method for skeletal muscle training for ICU bedridden patients according to an embodiment of the present application;
[0022] Figure 4 is another partial flow chart of the interactive control method for skeletal muscle training for ICU bedridden patients according to an embodiment of the present application;
[0023] Figure 5 is another partial flow chart of the interactive control method for skeletal muscle training of ICU bedridden patients according to an embodiment of the present application;
[0024] Figure 6 It is a block diagram of an interactive control system for skeletal muscle training of bedridden patients in the ICU according to an embodiment of the present application. DETAILED DESCRIPTION
[0025] Some embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.
[0026] like Figure 1As shown, in one embodiment of the present invention, an interactive control method for skeletal muscle training of ICU bedridden patients is provided, comprising:
[0027] Step S110, when a limb of a patient in the ICU generates a first movement, first movement data of bones and / or muscles corresponding to the limb of the patient are collected.
[0028] In this embodiment, there is no restriction on the method of collecting the first motion data of the patient.
[0029] Step S120, inputting the first motion data into a preset first neural network model, and identifying the patient's movement intention through the first neural network model.
[0030] In this embodiment, by accurately collecting the patient's first motion data and using a neural network model to identify movement intentions, it is helpful to generate a highly personalized virtual reality training scenario for each patient. This personalized design ensures that the training program closely fits the patient's actual rehabilitation needs, improves the pertinence and effectiveness of the training, and accelerates the rehabilitation process.
[0031] Step S130, constructing a virtual reality scenario according to the patient's movement intention, wherein the virtual reality scenario has a plurality of elements, and the plurality of elements have preset attribute values, which are used to induce the patient to perform a second action corresponding to the movement intention according to a preset induction intensity.
[0032] In this embodiment, the virtual reality scenario can be a sports field, a gym, etc., and the multiple elements can be graphic elements such as people, vehicles, animals, and sound elements such as instructions and music, which are used to induce and stimulate the patient to perform a second action. When the patient's exercise intention is to run, the second action induced by the multiple elements to the patient may be leg lifting, etc.
[0033] Step S140, collecting second motion data of bones and / or muscles corresponding to the limbs when the patient performs a second motion.
[0034] Step S150, querying the standard motion data corresponding to the second motion, calculating the difference between the second motion data and the standard motion data corresponding to the second motion, so as to reflect the movement disorder level of the patient when performing the second motion.
[0035] In this embodiment, by calculating the gap between the second motion data and the standard motion data, a new method for quantitatively evaluating the patient's movement disorder is actually implemented. This quantitative evaluation helps medical staff to accurately grasp the patient's rehabilitation progress, adjust the rehabilitation plan in time, and ensure the scientificity and accuracy of the treatment measures.
[0036] Step S160, input the difference between the motion data of the bones and / or muscles corresponding to the patient's limbs and the standard motion data corresponding to the second motion into a preset second neural network model corresponding to the virtual reality scenario, and calculate the attribute value that at least one element of the virtual reality scenario should have through the second neural network model.
[0037] Step S170: adjusting the attribute value of at least one element in the virtual scene to change the induction intensity according to the attribute value that at least one element of the virtual reality scene should have.
[0038] The innovation of this embodiment is that the induction intensity and element attributes of the virtual reality scenario can be adjusted in real time according to the actual performance of the patient in performing the second action, thus realizing dynamic adaptive training. This method can automatically optimize the training program according to the patient's immediate feedback and progress, avoid overtraining or undertraining, and improve the safety and efficiency of training.
[0039] Step S180: providing the virtual reality scenario to the user, and inducing the patient to perform the second action again according to the changed induction intensity through various elements in the virtual reality scenario.
[0040] According to the technical solution of this embodiment, the integration of biomechanical analysis, deep learning, sensor technology and virtual reality technology is realized, demonstrating the innovative application of interdisciplinary technology in the field of medical rehabilitation, providing ICU bedridden patients with a more scientific, efficient and personalized skeletal muscle training program, significantly improving the patients' rehabilitation effects, reducing the risk of complications related to long-term bed rest, and opening up new ways for the effective allocation and utilization of medical resources.
[0041] like Figure 2 As shown, in one embodiment of the present invention, an interactive control method for skeletal muscle training for ICU bedridden patients is provided. Compared with the aforementioned technical solution, the interactive control method for skeletal muscle training for ICU bedridden patients in this embodiment, step S110 includes:
[0042] Step S210, continuously taking a plurality of images of the patient's limbs within a preset time period.
[0043] Step S220, identifying the positions of multiple joints and muscle groups on the patient's limbs in a plurality of consecutive images.
[0044] Step S230, calculating the angular velocity vectors and angular acceleration vectors of the multiple joints on the patient's limbs according to the position changes of the multiple joints and muscle groups on the patient's limbs.
[0045] In this embodiment, by continuously taking images of the patient's limbs and identifying the positions of their joints and muscle groups, combined with advanced image processing and recognition technology, high-precision capture of the patient's limb movements is achieved.
[0046] Step S240, calculating the joint torque and power output of multiple joints on the patient's limbs, the joint torque of the mth joint among the multiple joints The power output of the mth joint is P = τω m , where ω m represents the angular velocity vector of the mth joint, Represents ω m The derivative with respect to time, α m represents the angular acceleration vector of the mth joint, I m represents the preset moment of inertia matrix, and g represents the preset gravity vector.
[0047] In this embodiment, by calculating the angular velocity vector, angular acceleration vector, joint torque and power output of the joint, an in-depth biomechanical analysis is provided. This method enables the training program to be formulated based on accurate physiological and mechanical parameters, improves the scientificity and accuracy of the training, and helps to more effectively restore the patient's muscle strength and joint function.
[0048] Furthermore, before this step, the method may further include: (1) extracting the material density, center of mass and coverage area of multiple joints and muscle groups on the patient's limbs from a preset skeletal and muscle system model constructed for the patient's limbs. (2) calculating the moment of inertia matrix for multiple joints on the patient's limbs based on the material density, center of mass and coverage area of multiple joints and muscle groups on the patient's limbs, and the coverage area of multiple joints and muscle groups, wherein the moment of inertia matrix I corresponding to the mth joint is m =∫ρ(r)(||r|| 2 -rc)dV, V represents the coverage area of the m-th joint and its muscle group, r represents a point in the coverage area of the m-th joint and its muscle group, c represents the center of mass of the m-th joint and its muscle group, and ρ(r) represents the material density of the m-th joint and its muscle group.
[0049] In this embodiment, by extracting key parameters from the patient's limb bone and muscle system model, including material density, center of mass position and muscle coverage area, a more accurate calculation of the joint moment of inertia matrix can be provided to the patient. The application of this personalized parameter significantly improves the accuracy of joint torque and power output calculation, ensures the pertinence and effectiveness of rehabilitation training, and makes the training plan more in line with the patient's actual situation.
[0050] Step S250, using the joint torque and power output of multiple joints on the patient's limbs as the first motion data for inputting into the first neural network model.
[0051] According to the technical solution of this embodiment, accurate assessment and control of joint torque and power output can help prevent common complications of long-term bedridden patients, such as muscle atrophy, joint stiffness, etc. By maintaining a reasonable mechanical load, promoting blood circulation, and enhancing muscle vitality, the technical solution of this embodiment significantly improves the patient's quality of rehabilitation and quality of life, and shortens the rehabilitation period.
[0052] The technical solution provided in another embodiment of the present invention can replace Figure 2 In the technical solution corresponding to the embodiment, step S110 includes:
[0053] (1) The pressure generated when the patient's limbs perform the first movement is collected through a preset pressure sensor array.
[0054] In this embodiment, the pressure changes caused by the patient's limb movements are accurately captured by the preset pressure sensor array, which significantly improves the sensitivity and accuracy of motion data collection compared to traditional methods. Especially in a complex environment such as the ICU, this technology can effectively eliminate the interference introduced by the patient's non-purposeful movements (such as micro-movements caused by breathing and heartbeat), ensuring the authenticity and reliability of the data.
[0055] (2) Based on the known pressure disturbance caused by the patient's non-purposeful movements, the pressure generated when the patient's limbs perform the first movement is corrected.
[0056] (3) The pressure generated when the corrected patient's limb performs the first action is used as the first action data
[0057] According to the technical solution of this embodiment, by intelligently correcting the pressure interference caused by non-purposeful actions, the first action data input into the neural network model is ensured to be purer, reducing the possibility of misjudgment. This directly improves the accuracy of movement intention recognition, makes the subsequent virtual reality scenario construction and rehabilitation training closer to the actual needs of patients, and improves the effectiveness and pertinence of training.
[0058] like Figure 3 As shown, an embodiment of the present invention provides an interactive control method for skeletal muscle training for ICU bedridden patients. Compared with the aforementioned technical solution, the interactive control method for skeletal muscle training for ICU bedridden patients in this embodiment, before step S120, further includes:
[0059] Step S310, when the first neural network model is trained, the loss function of the first neural network model is calculated Where E represents the total number of preset movement intention types, represents the predicted probability corresponding to the nth motion intention calculated during the training of the first neural network model, y n represents the true probability corresponding to the nth motion intention, w n is a weight coefficient corresponding to the n-th motion intention, which is set based on the known scarcity of the n-th motion intention.
[0060] In this embodiment, by introducing a weighted loss function in the first neural network model training stage, the data imbalance problem between types of motion intentions can be effectively solved, especially for those rare or difficult to capture motion intentions. By giving a higher weight coefficient to scarce motion intentions, the model will pay more attention to the learning of these intentions during the training process, significantly improving the balance and accuracy of all motion intention recognition.
[0061] Step S320, setting a loss correction function for the first neural network model Where T represents the number of trainings currently completed, and Respectively represent the output results of the first neural network model during the tth and t+1th training. and At the same time, The value of is 1, when and At the same time, The value of is 0.
[0062] In this embodiment, by introducing the loss correction function, the continuity of the prediction results during the model training process is effectively monitored and taken into account, especially the inconsistent prediction results between consecutive training steps are penalized. This mechanism helps to reduce fluctuations during the training process, enhance the stability of the model output and the smooth transition of the prediction sequence, and ensure the consistency and reliability of the patient's movement intention recognition.
[0063] Step S330, based on the loss correction function L cor , and correct the loss function L.
[0064] Step S340: adjusting the parameters of the first neural network model according to the size of the loss function L of the first neural network model.
[0065] According to the technical solution of this embodiment, by dynamically adjusting the weight coefficient to adapt to the scarcity of different motion intentions, it is conducive to effectively utilizing limited training resources, especially for specific motion intentions with small amounts of data, to achieve optimal allocation of resources. The loss correction function prompts the model to pay more attention to the consistency of continuous prediction results during training, avoiding unnecessary parameter oscillations, thereby improving training efficiency.
[0066] like Figure 4 As shown, an embodiment of the present invention provides an interactive control method for skeletal muscle training for ICU bedridden patients. Compared with the aforementioned technical solution, the interactive control method for skeletal muscle training for ICU bedridden patients in this embodiment, before step S160, further includes:
[0067] Step S410, setting an original change step size s for any parameter in the second neural network model o .
[0068] Step S420, when the second neural network model completes the ath training, calculate the change step length of any parameter in the second neural network model Where d is the preset reference value; according to the change step s a , change any parameter in the second neural network model.
[0069] According to the technical solution of this embodiment, by introducing a dynamic adjustment mechanism for the parameter change step size, the learning rate of the second neural network model is automatically optimized according to the training process, thereby achieving an effective balance between training efficiency and model convergence. As the training progresses, the change step size is dynamically adjusted through a specific formula that takes into account the number of training iterations and the preset benchmark value, and uses an exponential decay strategy to gradually reduce the step size, ensuring that the model can quickly explore the optimal solution in the early stage, while being able to make fine adjustments in the later stage to avoid overfitting and improve the generalization ability of the model.
[0070] In one embodiment of the present invention, an interactive control method for skeletal muscle training for bedridden ICU patients is provided. Compared with the aforementioned technical solution, in the interactive control method for skeletal muscle training for bedridden ICU patients of this embodiment, step S130 further includes:
[0071] (1) Extract the patient’s preference characteristics from the patient’s historical behavior records.
[0072] (2) According to the patient's preference characteristics, the type of any one of the multiple elements is changed, and the changed type of the element has the preference characteristics.
[0073] According to the technical solution of this embodiment, by integrating the patient's historical behavior records and extracting preference features, the elements in the virtual reality scenario can be adjusted in a personalized way to make it closer to the patient's interests and preferences. This personalized customization not only improves the attractiveness of the training and increases the patient's participation, but also stimulates the patient's training motivation, allowing the patient to participate in rehabilitation activities more actively while enjoying the training process, significantly improving the effect and efficiency of rehabilitation training.
[0074] like Figure 5As shown, in one embodiment of the present invention, an interactive control method for skeletal muscle training for ICU bedridden patients is provided. Compared with the aforementioned technical solution, the interactive control method for skeletal muscle training for ICU bedridden patients in this embodiment, step S170 includes:
[0075] Step S510, according to the patient's physiological reflex data, a sensitivity coefficient μ is set for the patient to reflect the sensitivity of the patient's limbs when receiving induction.
[0076] In this embodiment, by introducing the patient's physiological reflex data to set the sensitivity coefficient, accurate consideration of the individual differences of each patient is achieved, ensuring a high degree of personalization of the induced intensity adjustment in the virtual reality scenario.
[0077] Step S520: Setting the attribute adjustment range according to the patient's sensitivity coefficient Among them, x represents the attribute value that at least one element of the virtual reality scenario should have, x0 represents the actual attribute value of at least one element of the virtual reality scenario, and β and γ are preset values.
[0078] Step S530: continuously adjusting the attribute value of at least one element of the virtual reality scenario according to the attribute adjustment range until the desired attribute value is reached.
[0079] According to the technical solution of this embodiment, the attribute adjustment range calculated by using the sensitivity coefficient and other preset parameters can delicately reflect the patient's immediate response to rehabilitation training, making the element attribute value adjustment in the virtual reality scenario more accurate and adaptable. This dynamic adaptability optimizes the patient's rehabilitation experience, reduces discomfort during training, and increases the patient's cooperation and satisfaction.
[0080] like Figure 6 As shown, in one embodiment of the present invention, an interactive control system for skeletal muscle training for ICU bedridden patients is provided, comprising:
[0081] The first motion data collection module 610 collects first motion data of bones and / or muscles corresponding to the limbs of the patient in the ICU when the limbs of the patient produce a first motion.
[0082] In this embodiment, there is no restriction on the method of collecting the first motion data of the patient.
[0083] The movement intention recognition module 620 inputs the first motion data into a preset first neural network model, and recognizes the patient's movement intention through the first neural network model.
[0084] In this embodiment, by accurately collecting the patient's first motion data and using a neural network model to identify movement intentions, it is helpful to generate a highly personalized virtual reality training scenario for each patient. This personalized design ensures that the training program closely fits the patient's actual rehabilitation needs, improves the pertinence and effectiveness of the training, and accelerates the rehabilitation process.
[0085] The virtual reality scenario construction module 630 constructs a virtual reality scenario according to the patient's movement intention. The virtual reality scenario has multiple elements, and the multiple elements have preset attribute values, which are used to induce the patient to perform a second action corresponding to the movement intention according to a preset induction intensity.
[0086] In this embodiment, the virtual reality scenario can be a sports field, a gym, etc., and the multiple elements can be graphic elements such as people, vehicles, animals, and sound elements such as instructions and music, which are used to induce and stimulate the patient to perform a second action. When the patient's exercise intention is to run, the second action induced by the multiple elements to the patient may be leg lifting, etc.
[0087] The second motion data collection module 640 collects second motion data of the bones and / or muscles corresponding to the limbs when the patient performs the second motion.
[0088] The gap calculation module 650 queries the standard action data corresponding to the second action, and calculates the gap between the second action data and the standard action data corresponding to the second action, so as to reflect the movement disorder level of the patient when performing the second action.
[0089] In this embodiment, by calculating the gap between the second motion data and the standard motion data, a new method for quantitatively evaluating the patient's movement disorder is actually implemented. This quantitative evaluation helps medical staff to accurately grasp the patient's rehabilitation progress, adjust the rehabilitation plan in time, and ensure the scientificity and accuracy of the treatment measures.
[0090] The attribute value analysis module 660 inputs the difference between the motion data of the bones and / or muscles corresponding to the patient's limbs and the standard motion data corresponding to the second motion into a preset second neural network model corresponding to the virtual reality scenario, and calculates the attribute value that at least one element of the virtual reality scenario should have through the second neural network model.
[0091] The attribute value adjustment module 670 adjusts the attribute value of at least one element in the virtual scene to change the induction strength according to the attribute value that at least one element of the virtual reality scene should have.
[0092] The innovation of this embodiment is that the induction intensity and element attributes of the virtual reality scenario can be adjusted in real time according to the actual performance of the patient in performing the second action, thus realizing dynamic adaptive training. This method can automatically optimize the training program according to the patient's immediate feedback and progress, avoid overtraining or undertraining, and improve the safety and efficiency of training.
[0093] The induction module 680 provides the virtual reality scenario to the user, and induces the patient to perform the second action again according to the changed induction intensity through various elements in the virtual reality scenario.
[0094] According to the technical solution of this embodiment, the integration of biomechanical analysis, deep learning, sensor technology and virtual reality technology is realized, demonstrating the innovative application of interdisciplinary technology in the field of medical rehabilitation, providing ICU bedridden patients with a more scientific, efficient and personalized skeletal muscle training program, significantly improving the patients' rehabilitation effects, reducing the risk of complications related to long-term bed rest, and opening up new ways for the effective allocation and utilization of medical resources.
[0095] The basic principles of the present application are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present application are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present application. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, not for limitation, and the above details do not limit the present application to being implemented by adopting the above specific details.
[0096] The block diagrams of the devices, apparatuses, equipment, and systems involved in this application are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.
[0097] It should also be noted that in the apparatus, device and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.
[0098] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
[0099] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.
Claims
1. An interactive control method for skeletal muscle training of ICU bedridden patients, characterized in that: include: When a limb of a patient in the ICU produces a first movement, collecting first movement data of bones and / or muscles corresponding to the limb of the patient; Inputting the first motion data into a preset first neural network model, and identifying the patient's movement intention through the first neural network model; Constructing a virtual reality scenario according to the patient's movement intention, wherein the virtual reality scenario has a plurality of elements, and the plurality of elements have preset attribute values, and are used to induce the patient to perform a second action corresponding to the movement intention according to a preset induction intensity; collecting second motion data of bones and / or muscles corresponding to the limbs when the patient performs the second motion; querying standard motion data corresponding to the second motion, and calculating a gap between the second motion data and the standard motion data corresponding to the second motion, so as to reflect the movement disorder level of the patient when performing the second motion; Inputting the difference between the motion data of the bones and / or muscles corresponding to the limbs of the patient and the standard motion data corresponding to the second motion into a preset second neural network model corresponding to the virtual reality scenario, and calculating the attribute value that at least one element of the virtual reality scenario should have through the second neural network model; According to the attribute value that the at least one element of the virtual reality scenario should have, adjusting the attribute value of the at least one element in the virtual scenario to change the induction intensity; The virtual reality scenario is provided to the user, and the patient is induced to perform the second action again according to the changed induction intensity through the multiple elements in the virtual reality scenario.
2. The interactive control method for skeletal muscle training of ICU bedridden patients according to claim 1, characterized in that: Collecting first motion data of bones and / or muscles corresponding to the patient's limbs includes: continuously capturing a plurality of consecutive images of the patient's limbs within a preset time period; Identify the positions of multiple joints and muscle groups on the patient's limbs in the multiple consecutive images; Calculating angular velocity vectors and angular acceleration vectors of multiple joints on the patient's limbs according to position changes of multiple joints and muscle groups on the patient's limbs; Calculate the joint torque and power output of multiple joints on the patient's limbs, the joint torque of the mth joint in the multiple joints The power output of the mth joint is P = τω m , where ω m represents the angular velocity vector of the mth joint, Represents ω m The derivative with respect to time, α m represents the angular acceleration vector of the mth joint, I m represents the preset moment of inertia matrix, and g represents the preset gravity vector; The joint torque and power output of multiple joints on the patient's limbs are used as the first motion data for input into the first neural network model.
3. The interactive control method for skeletal muscle training of ICU bedridden patients according to claim 2, characterized in that: Prior to calculating the joint torques and power outputs of the plurality of joints on the patient's limb, the method further comprises: Extracting the material density, center of mass and coverage area of multiple joints and muscle groups on the patient's limbs from a preset skeletal and muscle system model constructed for the patient's limbs; The moment of inertia matrix is calculated for the multiple joints on the patient's limbs according to the material density, center of mass and coverage area of the multiple joints and muscle groups on the patient's limbs, and the coverage area of the multiple joints and muscle groups on the patient's limbs, wherein the moment of inertia matrix I corresponding to the mth joint is m =∫ρ(r)(||r|| 2 -rc)dV, V represents the coverage area of the m-th joint and its muscle group, r represents a point in the coverage area of the m-th joint and its muscle group, c represents the center of mass of the m-th joint and its muscle group, and ρ(r) represents the material density of the m-th joint and its muscle group.
4. The interactive control method for skeletal muscle training of ICU bedridden patients according to claim 1, characterized in that: Collecting first motion data of bones and / or muscles corresponding to the patient's limbs includes: collecting the pressure generated when the patient's limb performs the first action through a preset pressure sensor array; Based on the known pressure disturbance caused by the non-purposeful movement of the patient, correct the pressure generated when the patient's limb performs the first movement; The corrected pressure generated when the patient's limb performs the first motion is used as the first motion data.
5. The interactive control method for skeletal muscle training of ICU bedridden patients according to claim 1, characterized in that: Before inputting the first action data into a preset first neural network model, the method further includes: When the first neural network model is trained, the loss function of the first neural network model is calculated Where E represents the total number of preset movement intention types, represents the predicted probability corresponding to the nth motion intention calculated during the training of the first neural network model, y n represents the true probability corresponding to the nth motion intention, w n is a weight coefficient corresponding to the n-th motion intention, which is set based on the known scarcity of the n-th motion intention; According to the size of the loss function L of the first neural network model, the parameters of the first neural network model are adjusted.
6. The interactive control method for skeletal muscle training of ICU bedridden patients according to claim 5, characterized in that: Before adjusting the parameters of the first neural network model according to the size of the loss function L of the first neural network model, the method further includes: Set the loss correction function for the first neural network model Where T represents the number of trainings currently completed, and Respectively represent the output results of the first neural network model during the tth and t+1th training. and At the same time, The value of is 1, when and At the same time, The value of is 0; Based on the loss correction function L cor , the loss function L is modified.
7. The interactive control method for skeletal muscle training of ICU bedridden patients according to claim 1, characterized in that: Before inputting the difference between the motion data of the bones and / or muscles corresponding to the patient's limbs and the standard motion data corresponding to the second motion into a preset second neural network model corresponding to the virtual reality scenario, the method further includes: Set the original change step size s for any parameter in the second neural network model o ; When the second neural network model completes the ath training, the change step length of any parameter in the second neural network model is calculated. Wherein, d is the preset reference value; According to the change step s a , change any parameter in the second neural network model.
8. The interactive control method for skeletal muscle training of ICU bedridden patients according to claim 1, characterized in that: Constructing a virtual reality scenario according to the patient's movement intention includes: Extracting the patient's preference characteristics from the patient's historical behavior records; According to the patient's preference characteristics, the type of any one of the plurality of elements is changed, and the element of the changed type has the preference characteristics.
9. The interactive control method for skeletal muscle training of ICU bedridden patients according to claim 1, characterized in that: According to the attribute value that at least one element of the virtual reality scenario should have, adjusting the attribute value of the at least one element in the virtual scenario to change the induction intensity, including: According to the physiological reflex data of the patient, a sensitivity coefficient μ is set for the patient to reflect the sensitivity of the patient's limbs when receiving induction; Set the attribute adjustment amplitude according to the patient's sensitivity coefficient Wherein, x represents the attribute value that at least one element of the virtual reality scenario should have, x0 represents the actual attribute value that at least one element of the virtual reality scenario should have, and β and γ are preset values; The attribute value of at least one element of the virtual reality scenario is continuously adjusted according to the attribute adjustment amplitude until a desired attribute value is reached.
10. An interactive control system for skeletal muscle training of ICU bedridden patients, characterized in that: include: A first motion data acquisition module, which acquires first motion data of bones and / or muscles corresponding to the limbs of the patient in the ICU when the limbs of the patient produce a first motion; a movement intention recognition module, inputting the first motion data into a preset first neural network model, and recognizing the patient's movement intention through the first neural network model; a virtual reality scenario construction module, which constructs a virtual reality scenario according to the patient's movement intention, wherein the virtual reality scenario has a plurality of elements, and the plurality of elements have preset attribute values, and is used to induce the patient to perform a second action corresponding to the movement intention according to a preset induction intensity; A second motion data collection module is used to collect second motion data of bones and / or muscles corresponding to the limbs of the patient when the patient performs the second motion; a gap calculation module, which queries the standard action data corresponding to the second action, and calculates the gap between the second action data and the standard action data corresponding to the second action, so as to reflect the movement disorder level of the patient when performing the second action; an attribute value analysis module, inputting the difference between the motion data of the bones and / or muscles corresponding to the limbs of the patient and the standard motion data corresponding to the second motion into a preset second neural network model corresponding to the virtual reality scenario, and calculating the attribute value that at least one element of the virtual reality scenario should have through the second neural network model; an attribute value adjustment module, which adjusts the attribute value of at least one element in the virtual situation according to the attribute value that at least one element of the virtual reality situation should have, so as to change the induction intensity; The induction module provides the virtual reality scenario to the user, and induces the patient to perform the second action again according to the changed induction intensity through the multiple elements in the virtual reality scenario.
