Intelligent planning method and system for good limb position based on large model

Abstract

The application relates to an intelligent good-limb-position intelligent planning method and system based on a large model, relates to the AI auxiliary rehabilitation technical field, and comprises the following steps: a perception layer collects multidimensional physiological data and external scene data of a target user through a multimodal sensor, and obtains basic human characteristics, disease information, a rehabilitation stage and a preset lying position; a decision layer searches an adaptive good-limb-position adjustment parameter range according to the disease information, the rehabilitation stage and the preset lying position based on medical rehabilitation big data, takes the range as an optimization space, searches and optimizes according to the basic human characteristics, the multidimensional physiological data and the external scene data, and obtains optimal good-limb-position adjustment parameters; and the optimal parameters are sent to an execution layer to control an intelligent mechanical equipment group to place the good limb position of the target user. The application solves the problem that a traditional good-limb-position placing method relies on fixed standard parameters and general schemes, lacks in-depth understanding and response to individual differences of patients, and causes the placing scheme to be difficult to accurately adapt to actual needs of the patients.

Description

Technical Field

[0001] This application relates to the field of AI-assisted rehabilitation, and in particular to intelligent limb positioning planning methods and systems based on large models. Background Technology

[0002] With the advancement of intelligent rehabilitation medicine, proper limb positioning, as a key aspect of rehabilitation care, presents a significant challenge in improving rehabilitation outcomes, particularly in terms of its precision and personalized adaptation.

[0003] Currently, traditional methods of proper limb positioning rely on fixed standard parameters and general protocols, making it difficult to adapt to dynamic factors such as individual patient differences, real-time physiological states, and changing scenarios. This often results in insufficient targeting and low adaptability of the methods. This not only reduces the auxiliary effect of proper limb positioning on rehabilitation but also increases the complexity and implementation cost of nursing work. Summary of the Invention

[0004] To address the aforementioned technical issues, this application provides an intelligent limb positioning planning method and system based on a large model. This improves upon the traditional limb positioning methods, which rely on fixed standard parameters and general schemes, failing to adequately consider individual patient differences that lead to unsuitable positioning schemes and poor recovery outcomes.

[0005] The embodiments of this application disclose the following technical solutions:

[0006] In a first aspect, embodiments of this application provide an intelligent limb positioning planning method based on a large model, the method comprising:

[0007] The multi-modal sensors in the perception layer collect multi-dimensional physiological data and external scene data of the target user, and obtain the target user's basic human characteristics, disease information, rehabilitation stage and preset lying position;

[0008] At the decision-making level, based on medical rehabilitation big data, the appropriate range of limb position adjustment parameters is obtained according to the disease information, rehabilitation stage and preset lying position. Using the appropriate range of limb position adjustment parameters as the optimization space, the optimal limb position adjustment parameters are searched and optimized according to the basic human characteristics, multi-dimensional physiological data and external scene data to obtain the best limb position adjustment parameters.

[0009] The optimal limb positioning adjustment parameters are sent to the execution layer, and the intelligent mechanical equipment group is controlled to perform the optimal limb positioning for the target user in accordance with the optimal limb positioning adjustment parameters.

[0010] Secondly, embodiments of this application provide an intelligent limb positioning planning system based on a large model, the system comprising:

[0011] The perception data acquisition module is used to collect multi-dimensional physiological data and external scene data of the target user through the multimodal sensors of the perception layer, and to obtain the target user's basic human characteristics, disease information, rehabilitation stage and preset lying position;

[0012] The decision parameter optimization module is used at the decision level to retrieve and obtain the appropriate limb position adjustment parameter range based on medical rehabilitation big data, according to the disease information, rehabilitation stage and preset lying position, and use the appropriate limb position adjustment parameter range as the optimization space to search and optimize the limb position adjustment parameters according to the basic human characteristics, multi-dimensional physiological data and external scene data to obtain the best limb position adjustment parameters.

[0013] The placement control module is used to send the optimal limb positioning adjustment parameters to the execution layer and control the intelligent mechanical equipment group to perform the optimal limb positioning for the target user according to the optimal limb positioning adjustment parameters.

[0014] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0015] This application proposes an intelligent limb positioning planning method and system based on a large model. It achieves intelligent and precise limb positioning by using multi-modal sensors at the perception layer to collect multi-dimensional physiological data, a decision layer to retrieve and iteratively find optimal limb positioning adjustment parameters, and an execution layer to control intelligent mechanical equipment to perform the placement. First, the multi-modal sensors at the perception layer collect multi-dimensional physiological data of the target user, external scene data, basic human characteristics, disease information, rehabilitation stage, and preset lying position. Next, at the decision layer, based on medical rehabilitation big data, it retrieves a suitable range of optimal limb positioning adjustment parameters and optimizes these parameters using a deep learning-based evaluation channel for limb positioning results. Finally, the optimal limb positioning adjustment parameters are sent to the execution layer, where the intelligent mechanical equipment group executes the optimal limb positioning for the target user.

[0016] This technical solution solves the problems of poor rehabilitation effect and low efficiency caused by reliance on human experience and single parameter settings in traditional limb positioning by integrating multimodal data acquisition, parameter optimization by combining medical big data and deep learning, and precise execution by intelligent mechanical equipment. It achieves personalized and scientific limb positioning and provides technical support for patient rehabilitation treatment. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating the intelligent limb positioning planning method based on a large model provided in this application embodiment;

[0019] Figure 2 This is a schematic diagram of the structure of the intelligent limb positioning planning system based on a large model provided in the embodiments of this application.

[0020] The components represented by each number in the attached diagram are explained below:

[0021] Sensing data acquisition module 01, decision parameter optimization module 02, and execution placement control module 03. Detailed Implementation

[0022] This application provides a method and system for intelligent planning of good limb positioning based on a large model, which is used to solve the technical problem that the traditional good limb positioning methods in the prior art rely on fixed standard parameters and general schemes, lack a deep understanding and response to individual differences of patients, and make it difficult to accurately adapt the positioning scheme to the actual needs of each patient.

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0025] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0026] Example 1, as shown in the appendix Figure 1 As shown, this application provides an intelligent planning method for good limb positioning based on a large model, the method comprising the following steps:

[0027] S110: Collect multi-dimensional physiological data and external scene data of the target user through the multimodal sensor of the perception layer, and obtain the target user's basic human characteristics, disease information, rehabilitation stage and preset lying position;

[0028] In this embodiment of the application, in the scenario of intelligent limb positioning planning based on a large model, in order to provide a comprehensive and accurate data foundation for the subsequent decision-making layer to optimize the limb positioning adjustment parameters, it is necessary to obtain the physiological state, scene characteristics and basic information of the target user through the deployment of multimodal sensors in the perception layer and the integration of multi-dimensional information.

[0029] Specifically, the core scope of data collection should first be defined, covering the target user's physical condition and external environment, which will serve as the basic boundary for parameter collection.

[0030] Furthermore, by deploying multimodal sensors at the perception layer, multidimensional physiological data of the target user can be monitored in real time.

[0031] The multi-dimensional physiological data includes current physiological indicators such as heart rate, blood pressure, blood oxygen saturation, body temperature, respiratory rate, electromyography signals, and pain scores, to dynamically reflect the user's real-time physiological state.

[0032] At the same time, external scene data is collected synchronously, such as mattress characteristics, bed surface type, bed frame angle, bedding thickness, conduit wire distribution and external fixing device information, to comprehensively present the environmental characteristics of the user.

[0033] Furthermore, it is also necessary to simultaneously obtain basic information about the target users, including basic human characteristics, disease information, rehabilitation stage, and preset lying position, in order to clarify the individual differences and basic needs of users.

[0034] This step integrates multi-dimensional physiological data, external scene data, and basic information through multi-modal sensors, providing comprehensive data support for decision-makers to retrieve appropriate parameter ranges and conduct precise optimization based on medical rehabilitation big data, thus ensuring the personalization and accuracy of subsequent limb positioning planning.

[0035] Step S110 in the method provided in this application embodiment includes:

[0036] Multi-dimensional physiological data and external scene data of the target user are collected by multi-modal sensors. The multi-dimensional physiological data includes at least heart rate, blood pressure, blood oxygen saturation, body temperature, respiratory rate, electromyography signal and pain score. The external scene data includes at least mattress characteristics, bed surface type, bed frame angle, bedding thickness, distribution of conduit wires and external fixation device information.

[0037] The system acquires the target user's basic human characteristics, disease information, rehabilitation stage, and preset lying position. The basic human characteristics include at least height, weight, body type, and body circumference. The disease information is one of the following: neurological disease, orthopedic disease, internal medicine disease, or surgical disease. The preset lying position is one of the following: supine, prone, lateral, semi-recumbent, prone-lateral, inclined, sitting, lithotomy, or prone-elevated position.

[0038] In this embodiment of the application, in order to provide comprehensive and accurate basic data for intelligent planning of good limb positioning, it is necessary to deploy multimodal sensors and integrate multi-dimensional information to achieve all-round collection of the target user's physiological state, scene characteristics and basic information, so as to ensure the scientificity and adaptability of subsequent decision-making parameter range optimization.

[0039] Specifically, the first step is to define the core dimensions of data collection, including the target user's real-time physiological state, the characteristics of the external scene, and basic individual information, and then construct a complete data collection system.

[0040] Furthermore, multi-dimensional physiological data collection will be conducted. By deploying multimodal sensors at the perception layer, various physiological indicators of the target user can be monitored in real time, and multi-dimensional physiological data can be integrated to obtain the data.

[0041] The multidimensional physiological data includes heart rate, blood pressure, blood oxygen saturation, body temperature, respiratory rate, electromyography signal, and pain score.

[0042] Specifically, heart rate (unit: beats / minute) is collected by photoelectric sensors on the chest to reflect the heart's pumping and circulatory baseline status, and needs to be continuously monitored and the fluctuation range recorded; blood pressure (unit: mmHg) is obtained by non-invasive or invasive monitoring devices to obtain systolic and diastolic pressure, reflecting vascular pressure and circulatory resistance, and needs to be measured and recorded regularly with body position information.

[0043] In addition, blood oxygen saturation (unit: %) is collected by a sensor at the fingertip or earlobe to reflect the patient's oxygen supply status. It needs to be continuously monitored and the accuracy should be within ±2%. Body temperature (unit: ℃) is measured by a sensor on the body surface to reflect the patient's individual metabolism and fever status. It is necessary to ensure that the measurement method is appropriate.

[0044] In addition, respiratory rate (unit: breaths / minute) is acquired through chest and abdominal sensors to reflect the individual patient's respiratory function. Interference needs to be eliminated to ensure accurate counting. Electromyographic signals (unit: μV) are acquired through surface electrode pads on muscles to reflect the individual patient's muscle activity and tension. The electrode positions need to be dynamically adjusted according to the monitoring site.

[0045] In addition, pain scores are provided by patients using a digital scale (0-10 points) to reflect subjective pain perception, and need to be collected before and after adjustment of the surface electrode pads on the muscle. Through multi-dimensional physiological data collected by the multimodal sensor in the sensory layer, the real-time physiological changes of the target user can be comprehensively captured, providing real-time physiological evidence for subsequent optimization of limb positioning parameters.

[0046] Furthermore, external scene data is collected to fully understand the physical environment characteristics of the target user, providing scene-level support for the safety and adaptability of proper limb positioning.

[0047] The external scene data includes mattress characteristics, bed surface type, bed frame angle, bedding thickness, conduit wire distribution, and external fixing device information.

[0048] Specifically, mattress type includes mattress material, firmness, and elasticity coefficient, which are collected by material identification sensors and pressure sensors. These factors affect the support and pressure distribution of the patient's body, thus relating to comfort and the risk of pressure sores. Bed surface type refers to the surface structure of the bed, such as flat beds and multi-functional adjustable beds, which determines the stability and adjustability of the patient's lying position.

[0049] In addition, the bed frame angle (unit: °) refers to the tilt angle of the head and foot of the bed, which affects the patient's positional changes and physiological functions such as breathing and circulation; the bedding thickness refers to the thickness and material of the bedding, which affects the patient's contact sensation with the mattress and the local pressure.

[0050] In addition, catheter and lead distribution refers to the placement of medical auxiliary devices such as urinary catheters, infusion tubes, and monitoring lines on the patient's body to restrict the range of limb movement, and traction or compression must be avoided; external fixation device information refers to the location and nature of fixation devices such as plaster casts, splints, and braces, which are used to impose restrictions on the range of limb movement and safety measures.

[0051] By collecting data from external environments, we can accurately identify key factors in the patient's environment that may affect proper limb positioning, ensuring the safety, accuracy, and suitability of proper limb positioning.

[0052] Furthermore, simply acquiring real-time monitoring data on the target user's multi-dimensional physiological data and external scene data is insufficient to support personalized limb positioning planning. It is also necessary to simultaneously acquire the target user's basic information to clarify the user's individual differences and basic needs.

[0053] The basic information of the target user includes basic human characteristics, disease information, rehabilitation stage, and preset lying position. Basic human characteristics include at least height, weight, body type, and body circumference.

[0054] Among them, height (unit: cm) and weight (unit: kg) are used to assess limb proportions and load-bearing capacity; body type includes thin, normal, obese, etc., which usually affect the selection of support points; body circumference (unit: cm) such as shoulder circumference, waist circumference, thigh circumference, etc. are related to the size adaptation of the support device. These data can be collected by laser rangefinders, weighing sensors and circumference measurement tools or entered by medical staff.

[0055] In addition, the disease information includes one of the following: neurological diseases, orthopedic diseases, internal medicine diseases, and surgical diseases, in order to clarify the special requirements of proper limb positioning for preventing joint deformities and protecting special parts under different diseases.

[0056] Among them, neurological diseases (such as stroke and Parkinson's disease) require special attention to the prevention of limb spasticity and joint contractures; orthopedic diseases (such as fractures and arthritis) require appropriate fixation and healing of the affected area; internal medicine diseases (such as heart failure) may have special requirements for the angle of lying position; and surgical diseases (such as the postoperative recovery period) require avoiding pressure on the surgical site. This information can be obtained through electronic medical records or confirmation by medical staff.

[0057] Meanwhile, the rehabilitation stage needs to be determined in conjunction with the progression of the disease, including the acute phase, subacute phase, recovery phase, and sequelae phase.

[0058] In the acute phase, the focus is on stabilizing vital signs and preventing complications. Precise positioning of limbs is necessary to avoid worsening of the condition due to improper positioning. In the subacute phase, while maintaining stable vital signs, the focus is on maintaining early joint mobility and mild muscle strength training.

[0059] In addition, the recovery period should also include functional recovery training to promote the reconstruction of limb function and create favorable postural conditions for active or passive rehabilitation training. Patients in the sequelae stage often have long-term functional impairments or limb deformities; therefore, proper limb positioning focuses on slowing the functional decline process through long-term postural management to reduce the lasting impact of complications on daily activities. The goals of proper limb positioning differ at different rehabilitation stages and must be determined based on rehabilitation assessment records or the judgment of medical staff.

[0060] Furthermore, the preset lying position is one of the following: supine, prone, lateral, semi-recumbent, prone-lateral, inclined, sitting, lithotomy, or prone-elevated.

[0061] Among them, the supine position requires attention to prevent pressure sores and joint deformities, the lateral position requires adjustment of the support points to avoid local pressure, and the semi-recumbent position is often used to improve respiratory function, etc. The setting is determined by medical staff according to the treatment needs or the patient's tolerance.

[0062] This step, through the comprehensive collection of multi-dimensional physiological data and external scene data of the target user by multimodal sensors, and the systematic integration of basic information such as human body characteristics, disease information, rehabilitation stage and preset lying position, provides a reliable and comprehensive basis for decision-makers to search for the appropriate limb position adjustment parameter range based on medical rehabilitation big data, and to carry out precise optimization within this range in combination with individual characteristics, ensuring that subsequent limb position planning can fully adapt to the user's personalized needs.

[0063] S120: At the decision-making level, based on medical rehabilitation big data, the appropriate limb position adjustment parameter range is obtained according to the disease information, rehabilitation stage and preset lying position. Using the appropriate limb position adjustment parameter range as the optimization space, the limb position adjustment parameters are searched and optimized according to the basic human characteristics, multi-dimensional physiological data and external scene data to obtain the best limb position adjustment parameters.

[0064] In this embodiment of the application, in the scenario of intelligent limb positioning planning based on a large model, in order to select limb positioning adjustment parameters that conform to the individual characteristics of the target user from massive data, it is necessary to achieve accurate definition of parameter range and efficient acquisition of optimal parameters through retrieval of medical rehabilitation big data and deep learning-driven search optimization.

[0065] Specifically, the basic parameter framework is built by first retrieving a suitable set of sample parameters from medical rehabilitation big data, using disease information, rehabilitation stage, and preset bed position as the core constraints.

[0066] Furthermore, based on medical rehabilitation big data, guided by proper limb positioning, the parameters for adjusting proper limb positioning that meet the constraints are retrieved to obtain a sample set of proper limb positioning adjustment parameters.

[0067] Among them, the good limb positioning adjustment parameters include good limb positioning placement parameters and placement execution parameters. By screening high-frequency data of the sample parameter set, the range of suitable good limb positioning adjustment parameters is determined, and an effective boundary is defined for subsequent optimization.

[0068] Meanwhile, a channel for evaluating the results of proper limb positioning was constructed based on deep learning. The model was trained using sample schemes and historical assessment data from medical rehabilitation big data, enabling it to evaluate the adaptability of different adjustment schemes.

[0069] Based on this, multiple good limb positioning adjustment parameters are randomly selected within the range of suitable good limb positioning adjustment parameters. Multiple good limb positioning adjustment schemes are constructed by combining the target user's basic human characteristics, multi-dimensional physiological data and external scene data. The placement adaptability of each scheme is output through the good limb positioning result evaluation channel.

[0070] Finally, using the range of limb positioning adjustment parameters that best suits the desired position as the optimization space, iterative optimization is carried out based on the placement adaptability, continuously filtering and adjusting the parameters to finally obtain the optimal limb positioning adjustment parameters.

[0071] This step, through the combination of medical rehabilitation big data retrieval and intelligent optimization, ensures the scientific and personalized nature of the limb positioning adjustment parameters, providing a core basis for the intelligent mechanical equipment group at the execution level to perform precise limb positioning operations for the target user according to the optimal parameters.

[0072] Step S120 in the method provided in this application embodiment includes:

[0073] Based on medical rehabilitation big data, guided by proper limb positioning and constrained by the disease information, rehabilitation stage, and preset supine position, a proper limb positioning adjustment parameter retrieval is performed to obtain a sample set of proper limb positioning adjustment parameters. The proper limb positioning adjustment parameters include proper limb positioning parameters and positioning execution parameters. The proper limb positioning parameters include several positioning parameters for several joint parts. The positioning parameters include joint angle parameters, spatial position parameters, support and fixation parameters, and safety constraint parameters. The positioning execution parameters include positioning force and positioning speed.

[0074] The sample set of good limb positioning adjustment parameters is screened using high-frequency data, and the range of suitable good limb positioning adjustment parameters is determined based on the high-frequency data screening results.

[0075] A deep learning-based evaluation channel for proper limb positioning results was constructed.

[0076] Within the range of adaptive limb positioning adjustment parameters, multiple limb positioning adjustment parameters are randomly selected, and multiple limb positioning adjustment schemes are constructed by combining the basic human characteristics, multi-dimensional physiological data and external scene data.

[0077] Using the evaluation channel for the good limb positioning results, the placement adaptability of the multiple good limb positioning adjustment schemes is evaluated, and multiple placement adaptability scores are output.

[0078] Using the range of suitable limb positioning adjustment parameters as the search space, the optimal limb positioning adjustment parameters are searched and optimized based on the multiple placement adaptability parameters to obtain the best limb positioning adjustment parameters.

[0079] In this embodiment of the application, in order to accurately select the optimal limb positioning adjustment parameters that match the individual characteristics of the target user from a large amount of medical rehabilitation data, it is necessary to rely on the retrieval capabilities of medical rehabilitation big data and the intelligent optimization driven by deep learning to achieve the accurate definition of the parameter range and the reliable acquisition of the optimal parameters, thereby providing a scientific and personalized basis for the optimal limb positioning operation of the execution layer.

[0080] First, based on medical rehabilitation big data, guided by proper limb positioning, and with disease information, rehabilitation stage, and preset lying position as core constraints, a set of suitable sample proper limb positioning adjustment parameters is retrieved from the medical rehabilitation big data to build a basic parameter framework.

[0081] The parameters for proper limb positioning include placement parameters and execution parameters. Placement parameters include joint angles, spatial position, support and fixation, and safety constraints. Execution parameters include placement force and speed.

[0082] Based on this, by screening high-frequency data of the sample set of good limb positioning adjustment parameters, the range of suitable good limb positioning adjustment parameters is determined, thus defining an effective boundary for subsequent optimization operations.

[0083] For example, for a target user with an orthopedic disease (post-femoral neck fracture surgery), in the recovery period, and whose preset lying position is semi-recumbent, a large number of similar patients' good limb position adjustment parameters are retrieved from medical rehabilitation big data.

[0084] Among the parameters for proper limb positioning, the hip joint angle is mostly concentrated between 15° and 30°, the knee joint angle is mostly between 20° and 40°, and the support and fixation parameters indicate that soft padding should be placed on the waist and the affected lower limb for support; the positioning execution parameters show that the positioning force is mostly 20-30N and the positioning speed is mostly 5-10cm / s.

[0085] Based on this, after high-frequency data screening of these parameters, the range of suitable limb positioning adjustment parameters for this type of user was determined to be: hip joint angle 18°-28°, knee joint angle 25°-35°, lumbar support thickness 5-8cm, placement force 22-28N, placement speed 6-9cm / s, etc., which clarified the effective range for subsequent optimization of the suitable limb positioning adjustment parameters for this user.

[0086] Furthermore, an evaluation channel for optimal limb positioning results is constructed based on deep learning to scientifically and quantitatively assess the adaptability of different optimal limb positioning adjustment schemes, providing a reliable evaluation standard for subsequent parameter optimization.

[0087] The method provided in this application includes the following steps in "constructing an evaluation channel for good limb positioning results based on deep learning":

[0088] Based on medical rehabilitation big data, guided by proper limb positioning, and constrained by the disease information, rehabilitation stage, and preset supine position, a set of sample proper limb positioning adjustment schemes is collected, and historical positioning evaluation data of different sample proper limb positioning adjustment schemes are obtained. The sample positioning adaptability is comprehensively evaluated to obtain a sample positioning adaptability set. The historical positioning evaluation data includes at least subjective pain score, physiological indicator stability score, and pressure distribution uniformity score.

[0089] Using the sample set of good limb positioning adjustment schemes as input data and the sample placement fitness set as supervised data, a deep learning model is trained until convergence to obtain an evaluation channel for good limb positioning results.

[0090] In this embodiment of the application, in order to achieve accurate quantitative assessment of the adaptability of the limb positioning adjustment scheme, it is necessary to train a deep learning model based on medical rehabilitation big data, construct an evaluation channel for the limb positioning results, and learn the correlation characteristics between the scheme parameters and the actual positioning effect through the model in order to adapt to the evaluation needs of different diseases, rehabilitation stages and supine positions.

[0091] Specifically, based on medical rehabilitation big data, guided by proper limb positioning, and combined with disease information, rehabilitation stage, and pre-set supine position constraints, a set of sample proper limb positioning adjustment schemes was collected.

[0092] These limb positioning adjustment programs cover placement and execution parameters for different joint parts, such as shoulder joint angle of 30°-60°, hip joint support height of 5-10cm, placement force of 20-30N, and placement speed of 5-8cm / s.

[0093] Simultaneously, historical placement assessment data corresponding to each good limb positioning adjustment plan is obtained, including patients' subjective pain scores, physiological indicator stability scores, and pressure distribution uniformity scores. The sample placement fitness is obtained through weighted calculation, forming a sample placement fitness set, which provides basic data for model training.

[0094] The subjective pain score is obtained by having patients provide feedback on a numerical rating scale of 0-10 based on their own feelings, with 0 indicating no pain and 10 indicating the most severe pain.

[0095] In addition, the physiological stability score is based on the fluctuation range of physiological indicators such as heart rate, blood pressure, and blood oxygen saturation before and after placement. The smaller the fluctuation, the higher the score, with a maximum score of 100 points.

[0096] In addition, the pressure distribution uniformity score is obtained by measuring the contact pressure distribution between different parts of the patient's body and the mattress using pressure sensors. The more uniform the distribution, the higher the score, with a maximum score of 100.

[0097] Based on this, the above three types of scores are integrated and calculated by assigning different weights. The sample placement fitness is obtained by comprehensively evaluating the formula "sample placement fitness = subjective pain score × pain weight + physiological index stability score × physiological weight + pressure distribution uniformity score × pressure weight" to fully reflect the fit of the good limb position adjustment plan.

[0098] The weighting of each score is determined based on the degree of influence of each score on the effect of good limb positioning under different diseases, rehabilitation stages, and preset lying positions. For example, in the postoperative rehabilitation stage of orthopedic diseases, the weight of pressure distribution uniformity may be relatively high.

[0099] For example, in a lateral decubitus position adjustment scheme for a stroke patient in the recovery period, the patient's subjective pain score is 2 points, the physiological index stability score is 90 points, and the pressure distribution uniformity score is 85 points. If the pain weight is 0.4, the physiological weight is 0.3, and the pressure weight is 0.3, then the sample placement fitness of this scheme = 2 × (0.4) + 90 × 0.3 + 85 × 0.3 = 1.2 + 27 + 25.5 = 62.3 points.

[0100] Finally, by weighting the subjective pain score, physiological stability score, and pressure distribution uniformity score corresponding to each good limb positioning adjustment scheme according to preset weights, a sample placement fitness set that can comprehensively reflect the suitability of each scheme is obtained.

[0101] Furthermore, using a set of good limb positioning adjustment schemes as input and a set of sample placement fitness as supervision, an evaluation model based on a deep neural network is trained to construct an evaluation channel for good limb positioning results.

[0102] Specifically, a deep neural network framework consisting of an input layer, hidden layers, and an output layer is constructed. The input layer receives the parameter features from the limb positioning adjustment scheme, the hidden layer adopts a multi-layer fully connected structure with three hidden layers (each with 128, 64, and 32 neurons respectively), and extracts non-linear features through the ReLU activation function. The output layer outputs the placement fitness prediction value through a linear activation function.

[0103] Furthermore, the sample data is divided in an 8:1:1 ratio. For example, 4,000 sets are selected from 5,000 sets as the training set, 500 sets as the validation set, and 500 sets as the test set, covering a variety of diseases such as neurological diseases and orthopedic diseases, as well as the entire rehabilitation process from the acute phase to the sequelae phase, to ensure the comprehensiveness of the data distribution.

[0104] Furthermore, during the model training phase, the parameters of the sample placement schemes from the training set are input into the network, and the network parameters are adjusted using the Adam optimization algorithm to gradually reduce the mean squared error between the output predicted placement fitness and the sample placement fitness. The initial learning rate is set to 0.001, and the learning rate decays to half of the previous rate every 100 iterations.

[0105] Meanwhile, the model performance is monitored in real time using the validation set. If the mean squared error of the validation set decreases by less than 0.5 points over 20 consecutive rounds, the model is considered to have converged.

[0106] Furthermore, during the testing phase, the test set is input into the model. If the average deviation between the predicted placement fitness and the actual sample placement fitness is less than 3 points, the evaluation channel is deemed qualified and can be put into practical use.

[0107] Finally, the evaluation channel for the completed good limb positioning results can adjust the scheme parameters according to the input good limb positioning and output accurate positioning adaptability, providing a quantitative basis for subsequent parameter optimization.

[0108] For example, for a patient in the subacute phase after orthopedic surgery with a pre-set semi-recumbent position, the parameters of the optimal limb positioning adjustment scheme are: hip joint angle 25°, support thickness 7cm, placement force 26N, and placement speed 7cm / s. After inputting these parameters into the evaluation channel of the optimal limb positioning results after training, the channel outputs a placement fitness score of 83.

[0109] Meanwhile, according to the evaluation of medical staff, this good limb positioning adjustment plan can effectively reduce the patient's pain, maintain the stability of physiological indicators and even pressure distribution. The actual placement adaptability score is 83 points, which is consistent with the output value of the evaluation channel. This fully reflects the accuracy of the evaluation channel of the good limb positioning result and provides a reliable quantitative reference for selecting the optimal parameters from multiple plans.

[0110] Furthermore, after the evaluation channel for proper limb positioning results is constructed, the proper limb positioning adjustment parameters are iteratively optimized to obtain the proper limb positioning adjustment parameters that best match the individual characteristics and rehabilitation needs of the target user.

[0111] Specifically, firstly, multiple limb positioning adjustment parameters are randomly selected within the range of suitable limb positioning adjustment parameters. Then, combined with the target user's basic human characteristics, multi-dimensional physiological data, and external scene data, multiple targeted limb positioning adjustment schemes are constructed.

[0112] For example, for a target user suffering from a neurological disease (stroke), in the recovery period, with the preset lying position being the lateral position, the basic human characteristics are: height 165cm, weight 60kg, normal body shape, shoulder circumference 90cm, multi-dimensional physiological data are: heart rate 70 beats / minute, blood pressure 120 / 80mmHg, blood oxygen saturation 98%, pain score 2 points, and external scene data are: mattress is medium firm, bed frame angle is 0°, bedding thickness is 2cm, and there are no conduits, wires, or external fixation devices.

[0113] Based on this, within the range of suitable limb positioning adjustment parameters (such as shoulder joint angle 30°-60°, hip joint angle 15°-30°, placement force 20-30N, placement speed 5-8cm / s), three sets of parameters were randomly selected to construct three good limb positioning adjustment schemes.

[0114] Among them, the first good limb position adjustment scheme is a shoulder joint angle of 40°, a hip joint angle of 20°, a placement force of 22N, and a placement speed of 6cm / s; the second good limb position adjustment scheme is a shoulder joint angle of 50°, a hip joint angle of 25°, a placement force of 25N, and a placement speed of 7cm / s; the third good limb position adjustment scheme is a shoulder joint angle of 35°, a hip joint angle of 18°, a placement force of 24N, and a placement speed of 6.5cm / s.

[0115] Furthermore, using the established evaluation channel for optimal limb positioning results, the placement fitness of several randomly selected optimal limb positioning adjustment schemes was evaluated. Specifically, the parameters of each scheme were input into the evaluation channel, which then used its internal neural network to calculate and output the corresponding placement fitness score. For example, after evaluation, the placement fitness scores for the three schemes mentioned above were 78, 85, and 76 points, respectively.

[0116] Furthermore, using the range of suitable limb positioning adjustment parameters as the search space, the optimal limb positioning adjustment parameters are searched and optimized based on the multiple output placement adaptability values ​​to obtain the best limb positioning adjustment parameters.

[0117] The method provided in this application embodiment includes the step of "using the range of suitable limb positioning adjustment parameters as the optimization space, searching and optimizing the suitable limb positioning adjustment parameters according to the multiple placement adaptability to obtain the optimal limb positioning adjustment parameters" as follows:

[0118] Based on the multiple placement fitness values, the multiple good limb position adjustment parameters are sorted in descending order of placement fitness values, and the good limb position adjustment parameters are set as the initial solution to construct an initial solution sequence;

[0119] The first solution in the initial solution sequence is set as the optimal solution, the next preset proportion of the initial solutions are set as the poor solutions, and the remaining initial solutions are set as the inferior solutions. The same number of parameters are randomly selected within the range of the adaptive limb position adjustment parameters to replace the poor solutions, thereby obtaining an updated solution sequence. The preset proportion is greater than or equal to 2% and less than or equal to 20%, and the preset proportion gradually decreases as the number of optimization attempts increases.

[0120] Based on the updated solution sequence, with the optimal solution as the adjustment direction, the other solutions in the updated solution sequence are adjusted according to the preset parameter adjustment step size to obtain a secondary updated solution sequence. If there are other solutions in the secondary updated solution sequence with a placement fitness greater than that of the optimal solution, then the optimal solution is replaced by the solution with a placement fitness greater than that of the optimal solution.

[0121] Continue iterative optimization by replacing superior solutions, filtering inferior solutions, eliminating poor solutions, supplementing parameters, and adjusting solutions until the preset number of convergences is reached. Then, output the superior solutions in the current updated solution sequence as the optimal limb position adjustment parameters.

[0122] In this embodiment of the application, in order to accurately locate the parameters that best match the individual characteristics and rehabilitation needs of the target user within the range of parameters for proper limb positioning, it is necessary to perform multiple rounds of iterative optimization and dynamically optimize the parameter combination in combination with the positioning adaptability, so as to achieve the goal of efficiently screening the optimal solution from the set of candidate parameters and providing accurate basis for proper limb positioning operation in the execution layer.

[0123] Specifically, firstly, based on the placement adaptability of multiple good limb positioning adjustment schemes, the corresponding good limb positioning adjustment parameters are sorted in descending order, and these parameters are used as initial solutions to construct an initial solution sequence.

[0124] For example, in a certain scenario, 100 good limb positioning adjustment schemes are generated. After the good limb positioning result evaluation channel outputs the placement fitness, the initial solution sequence is obtained by sorting the placement fitness values, where the fitness values ​​of the solutions from the 1st to the 100th position decrease sequentially.

[0125] Furthermore, the first solution in the initial solution sequence is selected as the optimal solution, and the solutions that follow the sequence at a predetermined proportion (e.g., 10%) are designated as suboptimal solutions, while the remaining solutions are designated as inferior solutions. Within the range of parameters for adjusting the limb position, new parameters equal to the number of suboptimal solutions are randomly selected to replace the suboptimal solutions, thus forming an updated solution sequence.

[0126] The preset ratio is usually greater than or equal to 2% and less than or equal to 20% to achieve a dynamic balance between search range and accuracy during the optimization process.

[0127] Meanwhile, the preset ratio will gradually decrease as the number of optimization attempts increases (e.g., from 10% to 2%). In the early stage, a larger ratio is used to expand the search range, and in the later stage, the ratio is reduced to finely adjust the solution near the optimal solution. This balances the breadth and accuracy of the search and ensures that the final limb positioning adjustment parameters can better adapt to the individual characteristics and rehabilitation needs of the target user.

[0128] For example, if the initial solution sequence contains 100 solutions and the preset ratio for the first optimization is 10%, then the last 10 solutions (100 × 10% = 10) are set as poor solutions. Ten new parameters are randomly selected within the range of the adjustment parameters for the best limb position to replace them, resulting in an updated solution sequence containing 100 solutions. In the fifth optimization, the preset ratio is reduced from 10% to 5%, and only the last 5 poor solutions (100 × 5% = 5) are replaced to conduct a refined search near the high-quality solutions.

[0129] Furthermore, based on the updated solution sequence, other solutions are adjusted according to a preset step size with the optimal solution as the adjustment direction, so as to generate a secondary updated solution sequence.

[0130] If there is a solution in the second update solution sequence with a placement fitness higher than the current optimal solution, then the optimal solution is replaced with the solution with higher placement comfort to achieve iterative optimization of the optimal solution.

[0131] For example, if the preset step size is 2° for each adjustment of the joint angle and 0.5N for each adjustment of the placement force, assuming the optimal solution has a shoulder joint angle of 50° and a placement force of 25N, and a suboptimal solution has a shoulder joint angle of 45° and a placement force of 23N, after adjustment, the suboptimal solution's shoulder joint angle becomes 48° and the placement force becomes 23.5N. After evaluation by the good limb positioning result evaluation channel, its placement comfort score increases from 76 to 87, which is higher than the current optimal solution's 85. Therefore, the solution with a placement comfort score of 87 replaces the current optimal solution with a score of 85 and becomes the new optimal solution.

[0132] Based on this, the iterative optimization process of "replacing the best solution, screening the worst solution, eliminating the poor solution, supplementing the parameters, and adjusting the solution" is repeated to continuously narrow the search range and gradually approach the best combination of limb position adjustment parameters, so as to ensure that each iteration can move towards a better solution.

[0133] At the same time, a preset number of convergences is set. When the number of iterations reaches the target, the best solution in the current updated solution sequence is output as the optimal limb position adjustment parameter.

[0134] For example, if the preset number of convergences is 30, after the 30th iteration, if the fitness of the current optimal solution fluctuates by no more than 1 point for 5 consecutive iterations and is significantly higher than other solutions, then the optimization is stopped and the optimal solution in the current updated solution sequence is set as the best limb position adjustment parameter.

[0135] Ultimately, through the multi-round iterative optimization mechanism described above, the optimal limb positioning parameters that best match the individual characteristics and rehabilitation needs of the target user can be accurately identified within the range of suitable limb positioning adjustment parameters, providing a scientific and precise basis for subsequent limb positioning operations at the execution level.

[0136] S130: Send the optimal limb positioning adjustment parameters to the execution layer, and control the intelligent mechanical equipment group to perform the optimal limb positioning of the target user according to the optimal limb positioning adjustment parameters.

[0137] In this embodiment of the application, in order to transform the determined optimal limb positioning adjustment parameters into actual good limb positioning operations, it is necessary to use an intelligent mechanical device group at the execution layer to accurately execute the positioning action according to the parameters, so as to ensure that the positioning process is safe, stable and meets the individual needs of the patient, thereby realizing the intelligent and automated placement of good limbs.

[0138] Specifically, the optimal limb positioning parameters obtained from the output are first sent to the control unit of the execution layer in the form of standardized instructions. These instructions must include details such as the target parameters of each joint, the execution sequence, the force threshold, and safety constraints, providing a clear basis for the coordinated operation of the intelligent mechanical equipment group.

[0139] Furthermore, the intelligent mechanical equipment group at the execution layer carries out collaborative work based on the received parameter instructions.

[0140] In the method provided in this application embodiment, the intelligent mechanical equipment group includes a lightweight robotic arm, a support and fixing device, a drive device, and a force control device.

[0141] Among them, the lightweight robotic arm is the core execution component. Its end effector can plan a precise motion trajectory based on the joint angle parameters and spatial position parameters in the optimal limb positioning adjustment parameters. With the power support provided by the drive device, it drives the patient's limb to move slowly at the placement speed specified by the parameters until it reaches the target position.

[0142] In addition, the support and fixation device automatically adjusts the position, height, and hardness of the pillow according to the support and fixation parameters to provide stable support for the patient's head, torso, limbs, and other parts, preventing the limbs from shifting after placement.

[0143] Meanwhile, the force control device monitors the force when the robotic arm contacts the patient's body in real time to ensure that the actual placement force does not exceed the safety threshold set by the parameters. When an abnormal force is detected, it will immediately report to the control unit so that the robotic arm's movement can be adjusted in time to ensure the patient's safety.

[0144] Throughout the entire process, all devices worked together and strictly followed the optimal limb positioning adjustment parameters to perform the operation, ultimately achieving the best limb positioning for the target user and realizing the precise implementation from parameter planning to actual operation.

[0145] For example, for a stroke patient in a lateral decubitus position, the optimal limb positioning parameters are: shoulder flexion 70°, hip flexion 30°, placement force 25N, placement speed 6cm / s, and a 5cm thick support pad should be placed at the lower back and knee joints.

[0146] During execution, the lightweight robotic arm, driven by the drive unit, adjusts the patient's upper limb to a 70° forward flexion position of the shoulder joint and a 30° flexion position of the hip joint with a placement force of 25N and a placement speed of 6cm / s.

[0147] At the same time, the support and fixation device extends a 5cm thick support pad to the lower back and knee joints and fixes it in place; the force control device monitors the contact force throughout the process, and when the force approaches the 30N threshold during a certain adjustment, it automatically issues a deceleration command to ensure that the force is stable at around 25N.

[0148] In addition, the robot arm's motion data, support device status information, and force control device feedback data will be collected in real time and dynamically compared with the optimal limb positioning adjustment parameters. If parameter deviations occur, such as joint angle errors exceeding 5° or force fluctuations exceeding 3N, the control unit will immediately issue correction commands to adjust the robot arm's movements and support device status until they meet the requirements of the optimal limb positioning adjustment parameters.

[0149] Ultimately, through the precise and coordinated operation of the intelligent mechanical equipment group, the patient's limbs were stably placed in the target optimal position. The entire process required no manual intervention, which improved placement efficiency and ensured placement accuracy, providing a reliable guarantee for the patient's rehabilitation treatment.

[0150] The embodiments of this application, through the specific implementation methods described above, achieve the following technical effects:

[0151] This application proposes a large-scale model-based intelligent limb positioning planning method. First, multi-dimensional physiological data and external scene data of the target user are collected through multi-modal sensors in the perception layer. The multi-dimensional physiological data includes heart rate, blood pressure, and blood oxygen saturation, while the external scene data includes mattress characteristics and bed frame angles. Basic human characteristics, disease information, rehabilitation stage, and preset lying position are also obtained. Next, at the decision layer, based on medical rehabilitation big data, the appropriate range of limb positioning adjustment parameters is retrieved according to disease information, rehabilitation stage, and preset lying position. This range is determined through high-frequency screening of the sample parameter set. Then, using this range of limb positioning adjustment parameters as the optimization space, combined with data such as basic human characteristics, the optimal limb positioning adjustment parameters are searched and optimized using a deep learning-constructed evaluation channel for limb positioning results. Finally, the optimal limb positioning adjustment parameters are sent to the execution layer, controlling an intelligent mechanical device group composed of lightweight robotic arms to perform limb positioning, thereby achieving personalized and precise rehabilitation assistance.

[0152] The method provided in this application, through the technical solution of "multi-dimensional data acquisition - big data retrieval and intelligent iterative optimization - intelligent mechanical equipment group collaborative execution", solves the problems of traditional limb positioning relying on human experience, fixed parameter settings, and poor adaptability. It realizes personalized, scientific and automated limb positioning planning, improves positioning accuracy and efficiency, and provides reliable technical support for patient rehabilitation.

[0153] Example 2, as shown in the appendix Figure 2 As shown, based on the inventive concept of the intelligent limb positioning planning method based on a large model provided in Embodiment 1, this application also provides an intelligent limb positioning planning system based on a large model, specifically including:

[0154] The perception data acquisition module 01 is used to acquire multi-dimensional physiological data and external scene data of the target user through the multimodal sensor of the perception layer, and to obtain the target user's basic human characteristics, disease information, rehabilitation stage and preset lying position;

[0155] The decision parameter optimization module 02 is used at the decision level to retrieve and obtain the appropriate limb position adjustment parameter range based on medical rehabilitation big data, according to the disease information, rehabilitation stage and preset lying position, and to search and optimize the limb position adjustment parameters based on the basic human characteristics, multi-dimensional physiological data and external scene data, in order to obtain the best limb position adjustment parameters.

[0156] The placement control module 03 is used to send the optimal limb position adjustment parameters to the execution layer and control the intelligent mechanical equipment group to perform the optimal limb position placement for the target user according to the optimal limb position adjustment parameters.

[0157] In one embodiment, the sensing data acquisition module 01 is further configured to:

[0158] Multi-dimensional physiological data and external scene data of the target user are collected by multi-modal sensors. The multi-dimensional physiological data includes at least heart rate, blood pressure, blood oxygen saturation, body temperature, respiratory rate, electromyography signal and pain score. The external scene data includes at least mattress characteristics, bed surface type, bed frame angle, bedding thickness, distribution of conduit wires and external fixation device information.

[0159] The system acquires the target user's basic human characteristics, disease information, rehabilitation stage, and preset lying position. The basic human characteristics include at least height, weight, body type, and body circumference. The disease information is one of the following: neurological disease, orthopedic disease, internal medicine disease, or surgical disease. The preset lying position is one of the following: supine, prone, lateral, semi-recumbent, prone-lateral, inclined, sitting, lithotomy, or prone-elevated position.

[0160] In one embodiment, the decision parameter optimization module 02 is further configured to:

[0161] Based on medical rehabilitation big data, guided by proper limb positioning and constrained by the disease information, rehabilitation stage, and preset supine position, a proper limb positioning adjustment parameter retrieval is performed to obtain a sample set of proper limb positioning adjustment parameters. The proper limb positioning adjustment parameters include proper limb positioning parameters and positioning execution parameters. The proper limb positioning parameters include several positioning parameters for several joint parts. The positioning parameters include joint angle parameters, spatial position parameters, support and fixation parameters, and safety constraint parameters. The positioning execution parameters include positioning force and positioning speed.

[0162] The sample set of good limb positioning adjustment parameters is screened using high-frequency data, and the range of suitable good limb positioning adjustment parameters is determined based on the high-frequency data screening results.

[0163] A deep learning-based evaluation channel for proper limb positioning results was constructed.

[0164] Within the range of adaptive limb positioning adjustment parameters, multiple limb positioning adjustment parameters are randomly selected, and multiple limb positioning adjustment schemes are constructed by combining the basic human characteristics, multi-dimensional physiological data and external scene data.

[0165] Using the evaluation channel for the good limb positioning results, the placement adaptability of the multiple good limb positioning adjustment schemes is evaluated, and multiple placement adaptability scores are output.

[0166] Using the range of suitable limb positioning adjustment parameters as the search space, the optimal limb positioning adjustment parameters are searched and optimized based on the multiple placement adaptability parameters to obtain the best limb positioning adjustment parameters.

[0167] In one embodiment, the placement control module 03 is further configured to:

[0168] The intelligent mechanical equipment group includes a lightweight robotic arm, a support and fixing device, a drive device, and a force control device.

[0169] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0170] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0171] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for intelligent planning of good position of limbs based on a large model, characterized in that, The method comprises: acquiring multi-dimensional physiological data and external scene data of a target user through a multi-modal sensor of a perception layer, and acquiring basic human characteristics, disease information, rehabilitation stages and preset lying positions of the target user; in a decision layer, based on medical rehabilitation big data, searching and acquiring an adaptive good limb position adjustment parameter range according to the disease information, the rehabilitation stages and the preset lying positions, and taking the adaptive good limb position adjustment parameter range as an optimization space, searching and optimizing good limb position adjustment parameters according to the basic human characteristics, the multi-dimensional physiological data and the external scene data, to obtain optimal good limb position adjustment parameters, comprising: constructing a good limb position placement result evaluation channel based on deep learning, comprising: based on medical rehabilitation big data, taking good limb position placement as guidance and taking the disease information, the rehabilitation stages and the preset lying positions as conditional constraints, collecting a sample good limb position adjustment scheme set and acquiring historical placement evaluation data of different sample good limb position adjustment schemes, comprehensively evaluating to obtain sample placement fitness, and acquiring a sample placement fitness set, wherein the historical placement evaluation data at least includes subjective pain score, physiological index stability score and stress distribution uniformity score; taking the sample good limb position adjustment scheme set as input data and taking the sample placement fitness set as supervised data, training a deep learning model to convergence to obtain a good limb position placement result evaluation channel; randomly selecting a plurality of good limb position adjustment parameters within the adaptive good limb position adjustment parameter range, and combining the basic human characteristics, the multi-dimensional physiological data and the external scene data to construct a plurality of good limb position adjustment schemes; using the good limb position placement result evaluation channel to respectively evaluate the placement fitness of the plurality of good limb position adjustment schemes, and outputting a plurality of placement fitnesses; taking the adaptive good limb position adjustment parameter range as an optimization space, searching and optimizing good limb position adjustment parameters according to the plurality of placement fitnesses to obtain optimal good limb position adjustment parameters, comprising: based on the plurality of placement fitnesses, sorting the plurality of good limb position adjustment parameters in descending order of placement fitness, and setting the good limb position adjustment parameters as initial solutions to construct an initial solution sequence; setting the first solution of the initial solution sequence as a good solution, setting the initial solutions of a preset proportion as poor solutions, setting the remaining initial solutions as inferior solutions, randomly selecting the same number of parameters within the adaptive good limb position adjustment parameter range to replace the poor solutions to obtain an updated solution sequence, wherein the preset proportion is greater than or equal to 2% and less than or equal to 20%, and the preset proportion gradually decreases as the number of optimization increases; based on the updated solution sequence, adjusting other solutions of the updated solution sequence in the adjustment direction of the good solution according to a preset parameter adjustment step to obtain a secondary updated solution sequence, and if the secondary updated solution sequence has a solution with a placement fitness greater than that of the good solution, replacing the good solution with the solution with the greater placement fitness; continuing the iteration of good solution replacement-inferior solution screening-poor solution elimination-parameter supplement-solution adjustment until a preset convergence number is reached, and outputting the good solution in the current updated solution sequence as the optimal good limb position adjustment parameter; The optimal good limb position adjustment parameter is sent to an execution layer, and the intelligent mechanical device group is controlled to perform the good limb position placing of the target user according to the optimal good limb position adjustment parameter.

2. The large model-based intelligent good position intelligent planning method of claim 1, wherein, The multi-dimensional physiological data and external scene data of the target user are collected by the multi-modal sensors of the perception layer, and the basic human characteristics, disease information, rehabilitation stage and preset lying position of the target user are obtained, including: The multi-dimensional physiological data and external scene data of the target user are collected by the multi-modal sensors, wherein the multi-dimensional physiological data at least includes heart rate, blood pressure, blood oxygen saturation, body temperature, respiratory rate, electromyographic signal and pain score, and the external scene data at least includes mattress characteristics, bed surface type, bed frame angle, bedding thickness, catheter and lead distribution and external fixation device information; The basic human characteristics, disease information, rehabilitation stage and preset lying position of the target user are obtained, wherein the basic human characteristics at least include height, weight, body type category and body girth, the disease information is one of nervous system diseases, orthopedic diseases, internal medicine diseases and surgical diseases, and the preset lying position is one of supine position, prone position, lateral position, semi-recumbent position, prone lateral position, inclined lying position, sitting position, lithotomy position and prone elevated position.

3. The large model-based intelligent good position intelligent planning method of claim 1, wherein, Based on medical rehabilitation big data, the adaptive good limb position adjustment parameter range is retrieved according to the disease information, rehabilitation stage and preset lying position, including: Based on medical rehabilitation big data, the good limb position adjustment parameter retrieval is performed with the good limb position placing as a guide and the disease information, rehabilitation stage and preset lying position as a conditional constraint, and a sample good limb position adjustment parameter set is obtained, wherein the good limb position adjustment parameter includes good limb position placing parameter and placing execution parameter, the good limb position placing parameter includes a plurality of placing parameters of a plurality of joint parts, the placing parameter includes joint angle parameter, spatial position parameter, support fixing parameter and safety constraint parameter, and the placing execution parameter includes placing force and placing speed; The sample good limb position adjustment parameter set is subjected to high-frequency data screening, and the adaptive good limb position adjustment parameter range is determined according to the high-frequency data screening result.

4. The large model-based intelligent good position intelligent planning method of claim 1, wherein, The intelligent mechanical device group includes a lightweight mechanical arm, a support fixing device, a driving device and a force control device.

5. An intelligent smart positioning system for smart positioning of a limb based on a large model, characterized by, The system is used for executing the intelligent good limb position planning method based on a large model according to any one of claims 1-4, and the system includes: The perception data acquisition module is used for collecting the multi-dimensional physiological data and external scene data of the target user by the multi-modal sensors of the perception layer, and obtaining the basic human characteristics, disease information, rehabilitation stage and preset lying position of the target user; The decision parameter optimization module is used for searching and optimizing the good limb position adjustment parameter in the decision layer based on medical rehabilitation big data, according to the disease information, rehabilitation stage and preset lying position, and taking the adaptive good limb position adjustment parameter range as an optimization space, and searching and optimizing the good limb position adjustment parameter according to the basic human characteristics, multi-dimensional physiological data and external scene data, to obtain the optimal good limb position adjustment parameter. The execution placing control module is configured to send the optimal good limb position adjustment parameter to an execution layer and control the intelligent mechanical device group to execute the good limb position placing of the target user according to the optimal good limb position adjustment parameter.

Images