Skeletal postoperative rehabilitation evaluation method, device and equipment and storage medium

Abstract

The invention relates to the technical field of medical rehabilitation evaluation, in particular to a skeleton postoperative rehabilitation evaluation method, device and equipment and a storage medium. Skeleton forms, structures and healing conditions are shown through skeleton scanning image data, historical natural walking and medical record information data are combined, single-dimension limitation is avoided through multi-source fusion, and a foundation is laid for accurate evaluation; an ant colony algorithm is utilized to construct a search model, iterative search is carried out on a support vector machine model, an optimal path node parameter set is obtained, the global search advantage of the ant colony algorithm is brought into play, and the accuracy of the evaluation model is improved; a Gaussian mixture algorithm is combined, complex distribution characteristics of multi-source data are fused, and the relation between the rehabilitation state and the data is accurately described; according to individual medical records and historical walking data, personalized evaluation is realized, an accurate report is provided for a patient, a doctor is assisted to formulate a treatment scheme meeting individual requirements, and clinical decision is supported.

Description

Technical Field

[0001] The present invention relates to the technical field of medical rehabilitation evaluation, and particularly relates to a method, device, equipment and storage medium for postoperative rehabilitation evaluation of bones. Background Art

[0002] For patients after bone surgery, whether due to accidental injury, after fracture fixation surgery, after joint replacement surgery, or limb dysfunction caused by other bone-related diseases, the postoperative rehabilitation evaluation of bones is crucial for the treatment effect and recovery process of patients. Its accuracy and comprehensiveness directly affect whether patients can recover smoothly. However, the existing postoperative rehabilitation evaluation methods for bones still have significant deficiencies in terms of accuracy and personalization. They often lack in-depth exploration of the personalized characteristics of patients. Neither the physical function information of patients nor the key information such as disease history and treatment history contained in the medical record information of patients has been fully utilized. This makes the evaluation results inaccurate and unable to provide the most suitable guidance for rehabilitation treatment. Therefore, it is of great clinical significance to develop a method for postoperative rehabilitation evaluation of bones that can comprehensively integrate various aspects of information and achieve objective, accurate and comprehensive evaluation. Summary of the Invention

[0003] In order to solve the above-mentioned drawbacks in the prior art, the present invention proposes a method, system, equipment and storage medium for postoperative rehabilitation evaluation of bones.

[0004] To solve the above technical problems, the technical solutions adopted by the present invention are as follows: A method for postoperative rehabilitation evaluation of bones includes: obtaining bone scan image data and a rehabilitation evaluation task; constructing an ant colony search model based on a preset ant colony algorithm, bone scan image data and a rehabilitation evaluation task; obtaining the historical natural walking data and medical record information data of a patient; constructing a support vector machine model based on a preset support vector machine algorithm, historical natural walking data and medical record information data; performing iterative search on the support vector machine model based on the ant colony search model to obtain a set of path node parameters; constructing a postoperative rehabilitation evaluation model based on the set of path node parameters, historical natural walking data, bone scan image data, medical record information data and a preset Gaussian mixture algorithm; obtaining real-time natural walking data, and performing evaluation on the real-time natural walking data through the postoperative rehabilitation evaluation model to obtain a rehabilitation evaluation result.

[0005] Further, the constructing a postoperative rehabilitation evaluation model based on the set of path node parameters, historical natural walking data, bone scan image data, medical record information data and a preset Gaussian mixture algorithm includes: calculating the logarithmic likelihood probability value for the historical natural walking data based on the Gaussian mixture algorithm; Generate an action score set based on the logarithmic likelihood probability value and the preset single-action sequence; calculate the correlation between historical natural walking data, bone scan image data, and medical record information data, and construct a correlation graph adjacency matrix based on the correlation; perform feature extraction on the bone scan image data based on the preset gradient Sobel algorithm to obtain bone area features, bone perimeter features, and shape factor features; construct a Gaussian mixture model based on the action score set, the correlation graph adjacency matrix, the bone area features, the bone perimeter features, and the shape factor features; optimize the Gaussian mixture model according to the path node parameter set to obtain a postoperative rehabilitation evaluation model.

[0006] Further, the performing feature extraction on the bone scan image data based on the preset gradient Sobel algorithm to obtain bone area features, bone perimeter features, and shape factor features includes: performing gradient calculation on the bone scan image data based on the gradient Sobel algorithm to obtain gradient image data, gradient magnitude, and gradient direction; performing feature extraction on the gradient image data based on the double-threshold detection method, the gradient magnitude, and the gradient direction to obtain edge contour features; performing morphological analysis on the edge contour features to obtain bone area features, bone perimeter features, and shape factor features.

[0007] Further, the performing morphological analysis on the edge contour features to obtain bone area features, bone perimeter features, and shape factor features includes: counting the number of pixel points of the bone scan image according to the edge contour features and the gradient magnitude to obtain a pixel point set; converting the pixel point set into bone area features based on a conversion method; traversing the edge contour features to obtain an Euclidean distance set, and accumulating the Euclidean distance set to obtain bone perimeter features; performing feature analysis on the bone area features and the bone perimeter features to obtain shape factor features.

[0008] Further, the optimizing the Gaussian mixture model according to the path node parameter set to obtain a postoperative rehabilitation evaluation model includes: generating an optimization function according to the path node parameter set; solving the Gaussian mixture model based on the optimization function to obtain optimized weights, mean vectors, and covariance matrices; performing derivative calculation according to the preset zero derivative, the optimized weights, the mean vectors, and the covariance matrices to obtain optimized model parameters; optimizing the Gaussian mixture model based on the optimized weights, the mean vectors, the covariance matrices, and the optimized model parameters to obtain a postoperative rehabilitation evaluation model.

[0009] Further, the ant colony search model constructed based on the preset ant colony algorithm, bone scan image data, and rehabilitation evaluation task includes: calculating weighted image data based on the bone scan image data and the rehabilitation evaluation task; performing multi-level coarsening operations on the weighted image data based on a preset order to obtain N-level image parameter data; and constructing an ant colony search model based on the N-level image parameter data and the ant colony algorithm.

[0010] Further, the iterative search of the support vector machine model based on the ant colony search model to obtain a set of path node parameters includes: calculating the first derivative and the second derivative of the medical record information data by using the numerical differentiation method; calculating the weight and the bias term according to the first derivative, the second derivative, and a preset partial derivative; generating a pheromone concentration calculation formula according to the weight and the bias term, and the expression of the pheromone concentration calculation formula is as follows: , where is the pheromone concentration, is the weight, is a variable related to the first derivative and the second derivative, is the bias term; performing dimensionality reduction on the medical record information data by using the linear discriminant analysis method to obtain dimensionality-reduced case data; calculating the pheromone concentration based on the pheromone concentration calculation formula for the dimensionality-reduced case data; obtaining the penalty parameter and the kernel function parameter of the support vector machine model, and setting the penalty parameter and the kernel function parameter as path nodes; and performing iterative search on the support vector machine model based on the ant colony search model, the path nodes, and the pheromone concentration to obtain a set of path node parameters.

[0011] Further, a bone postoperative rehabilitation evaluation device includes: a first data acquisition module for acquiring bone scan image data and a rehabilitation evaluation task; a first model construction module for constructing an ant colony search model based on the preset ant colony algorithm, bone scan image data, and rehabilitation evaluation task; a second data acquisition module for acquiring the patient's historical natural walking data and medical record information data; a second model construction module for constructing a support vector machine model based on the preset support vector machine algorithm, historical natural walking data, and medical record information data; a parameter set acquisition module for performing iterative search on the support vector machine model based on the ant colony search model to obtain a set of path node parameters; a third model construction module for constructing a postoperative rehabilitation evaluation model based on the set of path node parameters, historical natural walking data, bone scan image data, medical record information data, and a preset Gaussian mixture algorithm; and a rehabilitation evaluation module for acquiring real-time natural walking data and obtaining a rehabilitation evaluation result by using the postoperative rehabilitation evaluation model for the real-time natural walking data.

[0012] Further, a device for postoperative rehabilitation assessment of bones, the device for postoperative rehabilitation assessment of bones includes: a memory and at least one processor, and instructions are stored in the memory; at least one of the processors calls the instructions in the memory so that the device for postoperative rehabilitation assessment of bones executes each step of the method for postoperative rehabilitation assessment of bones as described in any one of the above.

[0013] Further, a computer-readable storage medium, instructions are stored on the computer-readable storage medium, and when the instructions are executed by a processor, each step of the method for postoperative rehabilitation assessment of bones as described in any one of the above is implemented.

[0014] The beneficial effects of the method for postoperative rehabilitation assessment of bones of the present invention are as follows: The bone morphology, structure and healing status are comprehensively presented through bone scan image data, providing a key basis for bone rehabilitation assessment. At the same time, combining historical natural walking data and medical record information data provides disease background and treatment reference. The multi-source data fusion avoids the limitation of a single dimension, comprehensively covers the key aspects of rehabilitation, and lays a solid foundation for accurate assessment; an ant colony algorithm is used to construct a search model, and the model constructed based on the support vector machine algorithm is iteratively searched to obtain an optimal path node parameter set. This process gives play to the global search advantage of the ant colony algorithm, provides a reference for the subsequent modeling of the assessment model, and improves the accuracy and reliability of the assessment model; by combining the Gaussian mixture algorithm, fully considering the complex distribution characteristics of multi-source data, fusing data at the probability level, accurately depicting the relationship between the rehabilitation state and each data, greatly improving the accuracy and reliability of the assessment model; based on the medical records and historical walking data of the patient individual, personalized assessment is realized, providing a precise rehabilitation assessment report for the patient, assisting the doctor to formulate a treatment plan that meets the individual needs, and providing support for reliable clinical decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The above and / or additional aspects and advantages of the present invention will become obvious and easy to understand from the following description of the embodiments in conjunction with the accompanying drawings, where: Figure 1 is the first flow chart of the method for postoperative rehabilitation assessment of bones provided by the embodiment of the present invention; Figure 2 is the second flow chart of the method for postoperative rehabilitation assessment of bones provided by the embodiment of the present invention; Figure 3 is the third flow chart of the method for postoperative rehabilitation assessment of bones provided by the embodiment of the present invention; Figure 4 is the fourth flow chart of the method for postoperative rehabilitation assessment of bones provided by the embodiment of the present invention; Figure 5The fifth flowchart of a method for evaluating postoperative rehabilitation of bones provided by an embodiment of the present invention; Figure 6 The sixth flowchart of a method for evaluating postoperative rehabilitation of bones provided by an embodiment of the present invention; Figure 7 The seventh flowchart of a method for evaluating postoperative rehabilitation of bones provided by an embodiment of the present invention; Figure 8 A structural schematic diagram of a device for evaluating postoperative rehabilitation of bones provided by an embodiment of the present invention; Figure 9 A structural schematic diagram of a device for evaluating postoperative rehabilitation of bones provided by an embodiment of the present invention. Detailed implementation manners

[0016] Next, the technical solutions of the present invention will be described clearly and completely with reference to the accompanying drawings. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0017] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order different from that shown or described herein. In addition, the terms "comprising" or "having" and any variation thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0018] For ease of understanding, the specific processes of the embodiments of the present invention are described below. Please refer to Figure 1 , an embodiment of a method for evaluating postoperative rehabilitation of bones in an embodiment of the present invention, includes: 101. Obtain bone scan image data and a rehabilitation evaluation task; In this embodiment, bone scan images can be detected by means such as X-ray imaging, CT scanning, MRI magnetic resonance imaging, radionuclide imaging, and nuclear ultrasound imaging; bone scan image data can visually present the morphology, structure, and healing status of bones, providing key information about the bones themselves for subsequent rehabilitation assessment; for example, through images such as X-rays and CTs, the healing degree of the fracture site, the position and morphology after joint replacement, etc. can be clearly seen; at the same time, clarify the rehabilitation assessment tasks, determine the direction and focus of the assessment, whether it is to focus on limb function recovery, pain relief degree, or other specific rehabilitation indicators, laying a foundation for the entire assessment process; 102. Construct an ant colony search model based on a preset ant colony algorithm, bone scan image data, and rehabilitation assessment tasks; 103. Obtain the patient's historical natural walking data and medical record information data; In this embodiment, the historical natural walking data reflects the physical functions of patients with similar symptoms during the rehabilitation period, such as muscle strength, joint flexibility, and movement coordination, etc.; the medical record information data covers the patient's disease history and treatment history, providing a disease background and past treatment reference for the assessment. The comprehensive utilization of these multi-source data makes the assessment no longer limited to a single dimension, comprehensively covering all key aspects of the patient's rehabilitation, laying a solid foundation for accurate assessment; 104. Construct a support vector machine model based on a preset support vector machine algorithm, historical natural walking data, and medical record information data; 105. Perform iterative search on the support vector machine model based on the ant colony search model to obtain a set of path node parameters; In this embodiment, by using the optimized search strategy of the ant colony search model to continuously search in the parameter space of the support vector machine model, a set of key path point parameters with optimal performance can be found. These sets have an important impact on the final rehabilitation assessment model, helping to provide a reference for the modeling of the assessment model and improving the accuracy and reliability of the assessment model; 106. Construct a postoperative rehabilitation assessment model based on the set of path node parameters, historical natural walking data, bone scan image data, medical record information data, and a preset Gaussian mixture algorithm; In this embodiment, use the Gaussian mixture algorithm to construct a postoperative rehabilitation assessment model. This algorithm can fully consider the complex distribution characteristics of multi-source data, fuse different types of data at the probability level, so as to more accurately describe the relationship between the rehabilitation state and each data, construct a comprehensive and accurate postoperative rehabilitation assessment model, and thus more accurately explore the potential relationship between data, greatly improving the accuracy and reliability of the assessment model, and providing a more accurate tool for rehabilitation assessment; 107. Obtain real-time natural walking data, and use the postoperative rehabilitation evaluation model to evaluate the real-time natural walking data to obtain a rehabilitation evaluation result.

[0019] In this embodiment, evaluating the real-time natural walking data through the postoperative rehabilitation evaluation model can track the patient's rehabilitation progress in real time.

[0020] In this embodiment, the bone morphology, structure, and healing status are comprehensively presented through bone scan image data, providing a key basis for bone rehabilitation evaluation. At the same time, combining historical natural walking data and medical record information data provides disease background and treatment references. The multi-source data fusion avoids the limitations of a single dimension, comprehensively covers the key aspects of rehabilitation, and lays a solid foundation for accurate evaluation. The ant colony algorithm is used to construct a search model to iteratively search for the model constructed based on the support vector machine algorithm to obtain the optimal path node parameter set. This process gives play to the global search advantage of the ant colony algorithm, provides a reference for the subsequent modeling of the evaluation model, and improves the accuracy and reliability of the evaluation model. By combining the Gaussian mixture algorithm, fully considering the complex distribution characteristics of multi-source data, fusing data at the probability level, accurately depicting the relationship between the rehabilitation state and each data, greatly improving the accuracy and reliability of the evaluation model. Based on the patient's individual medical records and historical walking data, personalized evaluation is realized, providing a precise rehabilitation evaluation report for the patient, assisting doctors in formulating treatment plans that meet individual needs, and providing support for reliable clinical decisions.

[0021] Please refer to Figure 2 , the second embodiment of a bone postoperative rehabilitation evaluation method in the embodiment of the present invention includes: 201. Calculate the historical natural walking data based on the Gaussian mixture algorithm to obtain the log-likelihood probability value; In this embodiment, for historical natural walking data, these data may contain multiple different motion patterns or states. Through the Gaussian mixture algorithm, it can be decomposed into different Gaussian components, each component representing a potential motion pattern, and the log-likelihood probability value is calculated based on the Gaussian components. The log-likelihood probability value reflects the possibility that the patient is in a certain specific motion model. Calculating the log-likelihood probability value helps to evaluate the fitting degree of the model to the historical natural walking data. The higher the value, the stronger the ability of the model to explain the data, providing a basis for the subsequent generation of the action score set; 202. Generate an action score set based on the log-likelihood probability value and the preset single-action sequence; In this embodiment, the single-action sequence is scored through the log-likelihood probability value. This process is essentially to screen and quantify the key actions in the natural walking data, highlighting those actions that are more representative and important for describing the patient's natural walking state, providing more valuable feature information for subsequent further analysis and modeling; 203. Calculate the correlation between historical natural walking data, bone scan image data, and medical record information data, and construct an adjacency matrix of the correlation graph based on the correlation. In this embodiment, the adjacency matrix of the correlation graph describes the degree of association between these data in an intuitive mathematical form, and the elements in the matrix represent the correlation strength between the corresponding data. 204. Extract features from the bone scan image data based on the preset gradient Sobel algorithm to obtain bone area features, bone perimeter features, and shape factor features. 205. Construct a Gaussian mixture model based on the action score set, the adjacency matrix of the correlation graph, bone area features, bone perimeter features, and shape factor features. In this embodiment, the action score set highlights the key action information in the historical natural walking data, the adjacency matrix of the correlation graph reveals the relationship between different types of data, and the bone area, perimeter, and shape factor features provide important information from the perspective of bone structure. The Gaussian mixture model constructed in this way can comprehensively consider various data information and more comprehensively describe the patient's rehabilitation status. The Gaussian mixture model fuses this information at the probability level, enabling the model to capture complex non-linear relationships between data and providing a powerful modeling tool for rehabilitation assessment. 206. Optimize the Gaussian mixture model according to the path node parameter set to obtain a postoperative rehabilitation assessment model.

[0022] In this embodiment, the path node parameter set is obtained in previous algorithm steps (such as the iterative search of the ant colony search model for the support vector machine model), and it contains parameter information that is important for optimizing the model performance. Applying these parameters to the Gaussian mixture model can adjust the model structure, parameter settings, etc., so that the model better meets the actual rehabilitation assessment needs. Through this optimization, the postoperative rehabilitation assessment model can better utilize the characteristics of multi-source data, improve the accuracy and reliability of the assessment, and provide more accurate prediction and analysis for the final rehabilitation assessment result. In this embodiment, the Gaussian mixture algorithm is used to analyze historical natural walking data, calculate the log-likelihood probability value, and then generate an action score set to accurately screen and quantify key actions, deeply excavate the body function information contained in the walking data, and provide detailed and key feature data for rehabilitation evaluation; calculate the correlation between historical natural walking, bone scan images, and medical record information data, construct an adjacency matrix, and intuitively present the data association; at the same time, extract bone image features, fuse the action score set, adjacency matrix, and bone features to construct a Gaussian mixture model, comprehensively consider multi-source data, describe the rehabilitation state from multiple dimensions, and capture the complex non-linear relationship between data; based on the set of path node parameters obtained from the previous algorithm steps, optimize the Gaussian mixture model, adjust the model structure and parameters to make it more suitable for the actual needs of rehabilitation evaluation, and improve the accuracy and reliability of the evaluation.

[0023] Please refer to Figure 3 , the third embodiment of a bone postoperative rehabilitation evaluation method in the embodiment of the present invention includes: 301. Perform gradient calculation on the bone scan image data based on the gradient Sobel algorithm to obtain gradient image data, gradient magnitude, and gradient direction; In this embodiment, the gradient Sobel algorithm calculates the gradient of each pixel point in the bone scan image data by applying a specific template for convolution operation in the horizontal and vertical directions; the gradient image data records the gradient information of each pixel point, the gradient magnitude reflects the severity of the gray change of the pixel point, and the gradient direction indicates the direction of the largest gray change; in the bone scan image, the gray difference between the bone and the surrounding tissue will be significantly reflected in the gradient magnitude and gradient direction, which helps to accurately identify the edge of the bone in the subsequent process; for example, the gray change at the bone edge is relatively large, and its gradient magnitude value is relatively high, and the edge direction can be determined through the gradient direction. In this embodiment, in bone edge recognition, the gradient Sobel algorithm is used to calculate the gradient of the bone scan image data, clearly presenting the gradient information, gradient magnitude, and direction of each pixel point; the obvious manifestation of the gray difference between the bone and the surrounding tissue in these data lays a foundation for accurately identifying the bone edge, ensuring that the subsequent analysis is based on an accurate bone contour; in the feature extraction link, the double-threshold detection method screens pixel points based on the gradient magnitude and direction, effectively removing noise interference, retaining the true bone edge information, and extracting accurate edge contour features, further improving the reliability of bone feature analysis; in terms of bone feature quantification and application, morphological analysis is performed on the edge contour features to obtain bone area, perimeter, and shape factor features. These features quantify the bone state from multiple dimensions. In scenarios such as fracture healing monitoring, they can intuitively reflect the growth of callus and bone remodeling, etc., provide accurate data support for clinical diagnosis, assist in the scientific formulation and adjustment of rehabilitation programs, and improve the effect and efficiency of rehabilitation treatment.

[0024] 302. Feature extraction is performed on the gradient image data based on the double-threshold detection method, gradient magnitude, and gradient direction to obtain edge contour features. In this embodiment, the double-threshold detection method uses two preset thresholds (a high threshold and a low threshold) to screen the gradient magnitude; pixel points above the high threshold are determined as strong edge points, and pixel points below the low threshold are considered not to be edge points; for pixel points between the high and low thresholds, if they are connected to strong edge points, they are also regarded as edge points. Combining the gradient direction information, this method can effectively remove noise interference and retain the true bone edge information, thereby extracting relatively accurate edge contour features; for example, in bone scan images, some tiny noises may generate small gradient magnitudes, but they can be filtered out by the double-threshold detection, leaving only the effective information related to the bone edge. 303. Morphological analysis is performed on the edge contour features to obtain bone area features, bone perimeter features, and shape factor features.

[0025] In this embodiment, the bone area feature is calculated, which is the number of pixels enclosed by the contour, reflecting the occupied range of the bone in the image; the bone perimeter feature is the length of the edge contour, which can be used to evaluate the morphological changes of the bone; the shape factor feature comprehensively considers information such as area and perimeter, and can more comprehensively depict the shape characteristics of the bone; for example, for the monitoring of the fracture healing process, the change in bone area can reflect the growth of callus, the change in perimeter may imply bone remodeling, and the analysis of the shape factor helps to determine whether the bone has returned to its normal shape. Please refer to Figure 4 , the fourth embodiment of a method for postoperative rehabilitation assessment of bones in the embodiments of the present invention includes: 401. Pixel point counting is performed on the bone scan image according to the edge contour features and gradient magnitude to obtain a pixel point set. In this embodiment, the edge contour features define the boundary range of the bone in the image, and the gradient magnitude can help distinguish the pixel differences between the bone and surrounding tissues. By combining these two to perform pixel point counting on the bone scan image, the pixel points belonging to the bone part can be accurately screened out to form a pixel point set. 402. The pixel point set is converted into a bone area feature based on a conversion method. In this embodiment, the conversion method can be a method based on pixel counting, a geometric calculation method, a method based on region calculation after image segmentation, etc. 403. The edge contour features are traversed to obtain a set of Euclidean distances, and the set of Euclidean distances is accumulated to obtain the bone perimeter feature. In this embodiment, the Euclidean distance can accurately describe the Euclidean distance between two adjacent pixel points. By accumulating these distances, the perimeter feature of the bone is obtained. The bone perimeter obtained in this way can reflect the contour length of the bone, which is of great significance for observing changes in bone morphology. For example, during the fracture healing process, changes in the bone perimeter may imply the remodeling and repair conditions of the bone; 404. Perform feature analysis on the bone area feature and the bone perimeter feature to obtain the shape factor feature.

[0026] In this embodiment, the bone boundary is defined using the edge contour feature, and the pixel difference between the bone and the surrounding tissue is distinguished by combining the gradient magnitude. The bone pixel points are accurately counted to form a set, which lays a solid foundation for subsequent feature quantification. The pixel point set is converted into the bone area feature through a conversion method to achieve a quantitative assessment of the bone size; at the same time, the Euclidean distance is obtained by traversing the edge contour and accumulated to obtain the perimeter feature, which accurately reflects the contour length of the bone and is of great significance for observing changes in bone morphology. For example, it can imply the bone remodeling and repair conditions during fracture healing; The area, perimeter, and shape factor features comprehensively evaluate the bone state from multiple dimensions and provide comprehensive information support in disease diagnosis, treatment plan formulation, and rehabilitation monitoring.

[0027] Please refer to Figure 5 , the fifth embodiment of a postoperative rehabilitation assessment method for bones in the embodiment of the present invention includes: 501. Generate an optimization function according to the path node parameter set; In this embodiment, generating an optimization function according to the path node parameter set is actually integrating these parameters into an objective function. The design of this objective function is usually to make the model reach the optimal in certain evaluation metrics, such as minimizing the prediction error of the model, maximizing the fitting degree, etc. For example, if the path node parameter set contains parameters highly correlated with data features, then the optimization function may be constructed based on these parameters to emphasize the impact of these key features on the model performance; 502. Solve the Gaussian mixture model based on the optimization function to obtain the optimized weights, mean vectors, and covariance matrices; 503. Perform derivative calculations according to the preset zero derivative, optimized weights, mean vectors, and covariance matrices to obtain the optimized model parameters; In this embodiment, solving the Gaussian mixture model based on the optimization function generated in step 501 is to find a set of optimal weights, mean vectors, and covariance matrices through optimization algorithms (such as the expectation-maximization algorithm, etc.). The mean vector determines the central position of each Gaussian distribution, and the covariance matrix describes the shape and distribution range of each Gaussian distribution. By solving these optimized parameters, the Gaussian mixture model can better fit the data; 504. Optimize the Gaussian mixture model based on the optimized weights, mean vectors, covariance matrices, and optimized model parameters to obtain a postoperative rehabilitation evaluation model.

[0028] In this embodiment, the optimized model parameters obtained in step 503 are applied to the Gaussian mixture model for optimization. The finally obtained postoperative rehabilitation evaluation model can more accurately evaluate the patient's rehabilitation status because it is based on multi-source data (such as historical natural walking data, bone scan image data, and medical record information data, etc.) and is obtained through a series of optimization steps. In this embodiment, an optimization function is generated according to the path node parameter set, and key parameters are incorporated into the objective function for optimization for specific evaluation indicators, such as minimizing prediction errors and improving the fitting degree, highlighting the influence of parameters closely related to data characteristics, laying a good foundation for the optimization of the Gaussian mixture model; use the optimization function to solve the Gaussian mixture model to obtain the optimized weights, mean vectors, and covariance matrices, clarify the importance, central position, and distribution range of each Gaussian distribution, greatly improving the fitting ability of the Gaussian mixture model for multi-source data (such as historical natural walking data, bone scan image data, medical record information data), and fully mining the potential relationships between data; integrate the optimized parameters to deeply optimize the Gaussian mixture model to form a postoperative rehabilitation evaluation model. This model can more accurately evaluate the patient's rehabilitation status with multi-source data and multi-step optimization, helping to improve the effect and efficiency of rehabilitation treatment.

[0029] Please refer to Figure 6 , the sixth embodiment of a method for evaluating postoperative rehabilitation of bones in the embodiments of the present invention includes: 601. Calculate weighted image data according to the bone scan image data and the rehabilitation evaluation task. In this embodiment, the bone scan image data contains rich information such as the shape and structure of the bone, and the rehabilitation evaluation task determines which key features need to be extracted from these data for evaluation. Combining the two to obtain the weighted image data means that weight distribution is carried out on different information in the image. For example, if the rehabilitation evaluation focuses on a specific part of the bone (such as the fracture site), then the pixel information of this part will be given a higher weight when calculating the weighted image data, highlighting its importance in the evaluation. This is done to more specifically focus on the image features closely related to the rehabilitation evaluation when processing the data later, laying a foundation for accurate evaluation. 602. Perform multi-level coarsening operations on the weighted image data based on a preset order to obtain N-level image parameter data. In this embodiment, the preset order determines the degree and level of coarsening of the weighted image data. This operation is performed on the weighted image data because the original weighted image data may contain too many details, which may interfere with the extraction and analysis of the overall features in some cases. Through multi-level coarsening, starting from the original image data, each level gradually reduces the amount and complexity of data while retaining key information. For example, in the higher-level coarsening, some local subtle changes may be ignored, and more attention is paid to the overall bone morphology and structural features. The N-level image parameter data finally obtained is the result after different levels of abstraction. Each level contains image feature information at different scales, which provides a multi-scale perspective for the subsequent construction of the ant colony search model; 603. An ant colony search model is constructed based on N-level image parameter data and an ant colony algorithm.

[0030] In this embodiment, the ant colony algorithm is an optimization algorithm that simulates the foraging behavior of ants and has a strong global search capability. The N-level image parameter data contains the skeletal image features at multiple scales. The ant colony algorithm is combined with the ant colony algorithm to construct an ant colony search model, so that the ant colony search model can find the optimal solution in a complex solution space. In this embodiment, weighted image data is calculated based on bone scan image data and rehabilitation assessment tasks, and image information is weighted according to assessment requirements, highlighting key features. For example, pixels at fracture sites are given high weights, so that subsequent data processing can focus on key points and lay the foundation for accurate assessment. Based on the preset order, the weighted image data is multi-level coarsened to obtain N-level image parameter data. This operation removes excessive detail interference, retains key information, abstracts images from different levels, forms multi-scale feature information, and comprehensively covers the overall morphology and local details of the bones, providing a rich perspective for the construction of ant colony algorithm models. The ant colony search model constructed by combining N-level image parameter data with the ant colony algorithm has global search capabilities and can find the optimal solution in a complex data solution space, so that the model can mine potential key features, improve the accuracy and effectiveness of the model for rehabilitation assessment, and help achieve personalized and precise bone postoperative rehabilitation assessment.

[0031] See also Figure 7 A seventh embodiment of a bone surgery rehabilitation assessment method in the embodiments of the present invention includes: 701. Use numerical differentiation method to calculate the derivative of medical record information data to obtain first-order derivative and second-order derivative; In this embodiment, the numerical differentiation method is used to approximate the change rate of some treatment data and patient physical function indicators in the medical record information data over time or other potential variables. The first-order derivative reflects the speed of feature change, such as how fast the severity of symptoms changes over time; the second-order derivative reflects the change in the speed of change, such as whether the speed of change of the severity of symptoms is accelerating or slowing down. These derivative information helps to mine the dynamic features and trends hidden in the medical record information data; 702. Obtain weight and bias terms according to the first-order derivative, the second-order derivative and the preset bias derivative; 703. Generate a pheromone concentration calculation formula based on the weight and bias term. The pheromone concentration calculation formula is as follows: , where is the pheromone concentration, is the weight, are variables related to the first and second derivatives, is the bias term; In this embodiment, the weight determines the degree of influence of each derivative-related variable on the pheromone concentration. For example, if the weight corresponding to the first-order derivative of a certain indicator feature in the medical record information data is large, it indicates that the change rate of the feature has a more critical impact on the pheromone concentration. The bias term is used as a constant to adjust the function output as a whole; 704. Use linear discriminant analysis method to reduce the dimension of medical record information data to obtain reduced dimension case data; 705. Calculate the reduced dimension case data based on the pheromone concentration calculation formula to obtain the pheromone concentration; In this embodiment, the medical record data that has been processed by LDA (Linear Discriminant Analysis) is substituted into the pheromone concentration calculation formula. The data after dimensionality reduction retains key information. The pheromone concentration is calculated by this formula, which not only improves the calculation efficiency, but also ensures that the calculation of the pheromone concentration is based on the most representative medical record features, so that the calculation results can more accurately reflect the relationship between the medical record data and the evaluation target; 706. Obtain the penalty parameter and kernel function parameter of the support vector machine model, and set the penalty parameter and kernel function parameter as the path node; In this embodiment, these two parameters are set as path nodes, which provides a search basis for the ant colony search model to explore the performance of the SVM model under different parameter combinations; 707. Based on the ant colony search model, path nodes and pheromone concentrations, the support vector machine model is iteratively searched to obtain a path node parameter set.

[0032] 708. Perform an iterative search on the support vector machine model based on the ant colony search model, the path nodes and the pheromone concentration to obtain a path node parameter set; In this embodiment, the ant colony search model simulates the foraging behavior of ants, and the pheromone concentration plays a guiding role in this process. Path nodes with higher pheromone concentrations (i.e., a specific combination of penalty parameters and kernel function parameters) are more likely to be selected by ants. As the iteration proceeds, the ants continue to explore different path nodes and gradually find the parameter combination that optimizes the performance of the support vector machine model to form a path node parameter set. This iterative search process utilizes the relationship between the medical record data characteristics reflected by the pheromone concentration and the model performance; In this embodiment, numerical differentiation is used to derive the medical record information, and the dynamic characteristics of treatment data and physical function indicators, such as the speed and acceleration of symptom changes, are excavated to provide key information for in-depth understanding of the patient's condition and break through the limitations of traditional analysis; the weights and bias terms are determined based on the derivatives and preset biases, and the pheromone concentration calculation formula is generated to accurately characterize the relationship between medical record characteristics and pheromone concentrations and improve evaluation accuracy; linear discriminant analysis is used for dimensionality reduction, key information is retained, computational complexity is reduced, and computational efficiency is improved to ensure that the pheromone concentration calculation is based on core medical record characteristics and reflects the true relationship between data and evaluation targets; finally, the penalty parameters and kernel function parameters of the support vector machine are set as path nodes, and with the help of the ant colony search model and guided by the pheromone concentration, the optimal parameter combination is iteratively searched, which not only optimizes the performance of the support vector machine model, but also enhances the model's adaptability to complex medical record data, providing more reliable support for medical decision-making and facilitating precision medicine.

[0033] A bone post-operative rehabilitation assessment method according to an embodiment of the present invention is described above. A bone post-operative rehabilitation assessment device according to an embodiment of the present invention is described below. Figure 8 In one embodiment of the present invention, a bone post-operative rehabilitation assessment device comprises: The first data acquisition module 1 is used to acquire bone scanning image data and rehabilitation assessment tasks; A first model building module 2 is used to build an ant colony search model based on a preset ant colony algorithm, bone scan image data and rehabilitation assessment tasks; The second data acquisition module 3 is used to acquire the patient's historical natural walking data and medical record information data; The second model building module 4 is used to build a support vector machine model based on a preset support vector machine algorithm, historical natural walking data and medical record information data; A parameter set acquisition module 5 is used to iteratively search the support vector machine model based on the ant colony search model to obtain a path node parameter set; The third model building module 6 is used to build a postoperative rehabilitation assessment model based on the path node parameter set, historical natural walking data, bone scan image data and medical record information data and a preset Gaussian mixture algorithm; A rehabilitation evaluation module 7 is configured to obtain real-time natural walking data and use a postoperative rehabilitation evaluation model to evaluate the real-time natural walking data to obtain a rehabilitation evaluation result.

[0034] In this embodiment, the bone morphology, structure, and healing status are comprehensively presented through bone scan image data, providing a key basis for bone rehabilitation evaluation. At the same time, historical natural walking data and medical record information data are combined to provide disease background and treatment references. The multi-source data fusion avoids the limitations of a single dimension, comprehensively covers the key aspects of rehabilitation, and lays a solid foundation for accurate evaluation. The ant colony algorithm is used to construct a search model to iteratively search for a model constructed based on the support vector machine algorithm to obtain an optimal set of path node parameters. This process takes advantage of the global search ability of the ant colony algorithm to provide a reference for the subsequent modeling of the evaluation model, improving the accuracy and reliability of the evaluation model. By combining the Gaussian mixture algorithm, the complex distribution characteristics of multi-source data are fully considered, and the data is fused at the probability level to accurately describe the relationship between the rehabilitation state and each data, greatly improving the accuracy and reliability of the evaluation model. Based on the medical records and historical walking data of individual patients, personalized evaluation is realized, providing a precise rehabilitation evaluation report for patients, assisting doctors in formulating treatment plans that meet individual needs, and providing support for reliable clinical decisions.

[0035] Figure 9 FIG. is a schematic structural diagram of a bone postoperative rehabilitation evaluation device provided by an embodiment of the present invention. The bone postoperative rehabilitation evaluation device 900 may vary greatly due to different configurations or performances, and may include one or more processors (central processing units, CPUs) 913 (for example, one or more processors) and a memory 920, and one or more storage media 930 (for example, one or more mass storage devices) for storing application programs 933 or data 932. Among them, the memory 920 and the storage media 930 may be short-term storage or persistent storage. The program stored in the storage media 930 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on a bone postoperative rehabilitation evaluation device 900. Further, the processor 913 may be configured to communicate with the storage media 930 and execute a series of instruction operations in the storage media 930 on a bone postoperative rehabilitation evaluation device 900 to implement the steps of a bone postoperative rehabilitation evaluation method provided by the above method embodiments.

[0036] A postoperative skeletal rehabilitation assessment device 900 may further include one or more power supplies 940, one or more wired or wireless network interfaces 950, one or more input / output interfaces 960, and / or one or more operating systems 931, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, and so on. Those skilled in the art can understand that Figure 9 The structure of a postoperative skeletal rehabilitation assessment device shown does not constitute a limitation on the postoperative skeletal rehabilitation assessment device 900, and it may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0037] A computer-readable storage medium, on which instructions are stored, and when the instructions are executed by a processor, each step of a postoperative skeletal rehabilitation assessment method as described above is implemented.

[0038] The present invention and its embodiments have been described above. Such a description is not restrictive. What is shown in the drawings is only one of the embodiments of the present invention, and the actual content is not limited thereto. In general, if those of ordinary skill in the art are inspired by it and, without departing from the spirit of the present invention, design similar structural modes and embodiments to this technical solution without creative efforts, they shall fall within the protection scope of the present invention.

Claims

1. A method for postoperative rehabilitation assessment of bones, characterized in that, Including: Obtaining bone scan image data and rehabilitation assessment tasks; Constructing an ant colony search model based on a preset ant colony algorithm, bone scan image data, and rehabilitation assessment tasks; Obtaining the patient's historical natural walking data and medical record information data; Constructing a support vector machine model based on a preset support vector machine algorithm, historical natural walking data, and medical record information data; Performing iterative search on the support vector machine model based on the ant colony search model to obtain a set of path node parameters; Constructing a postoperative rehabilitation assessment model based on the set of path node parameters, historical natural walking data, bone scan image data, medical record information data, and a preset Gaussian mixture algorithm; Obtaining real-time natural walking data and using the postoperative rehabilitation assessment model to process the real-time natural walking data to obtain a rehabilitation assessment result.

2. The method for evaluating postoperative rehabilitation of bones according to claim 1, wherein The constructing of the postoperative rehabilitation assessment model based on the set of path node parameters, historical natural walking data, bone scan image data, medical record information data, and a preset Gaussian mixture algorithm includes: Calculating the historical natural walking data based on the Gaussian mixture algorithm to obtain log-likelihood probability values; Generating an action score set based on the log-likelihood probability values and a preset single-action sequence; Calculating the correlation between the historical natural walking data, bone scan image data, and medical record information data, and constructing an adjacency matrix of the correlation graph based on the correlation; Performing feature extraction on the bone scan image data based on a preset gradient Sobel algorithm to obtain bone area features, bone perimeter features, and shape factor features; Constructing a Gaussian mixture model based on the action score set, adjacency matrix of the correlation graph, bone area features, bone perimeter features, and shape factor features; Optimizing the Gaussian mixture model according to the set of path node parameters to obtain a postoperative rehabilitation assessment model.

3. The method for evaluating postoperative rehabilitation of bones according to claim 2, wherein The performing of feature extraction on the bone scan image data based on a preset gradient Sobel algorithm to obtain bone area features, bone perimeter features, and shape factor features includes: Performing gradient calculation on the bone scan image data based on the gradient Sobel algorithm to obtain gradient image data, gradient magnitude, and gradient direction; Performing feature extraction on the gradient image data based on a double-threshold detection method, gradient magnitude, and gradient direction to obtain edge contour features; Performing morphological analysis on the edge contour features to obtain bone area features, bone perimeter features, and shape factor features.

4. The method for evaluating postoperative rehabilitation of bones according to claim 3, wherein The performing of morphological analysis on the edge contour features to obtain bone area features, bone perimeter features, and shape factor features includes: Counting the number of pixel points in the bone scan image according to the edge contour features and gradient magnitude to obtain a set of pixel points; Converting the set of pixel points into bone area features based on a conversion method; Traversing the edge contour features to obtain a set of Euclidean distances, and accumulating the set of Euclidean distances to obtain bone perimeter features; Performing feature analysis on the bone area features and bone perimeter features to obtain shape factor features.

5. The method for evaluating postoperative rehabilitation of bones according to claim 2, characterized in that, The optimizing of the Gaussian mixture model according to the set of path node parameters to obtain a postoperative rehabilitation assessment model includes: Generating an optimization function according to the set of path node parameters; Solve the Gaussian mixture model based on the optimization function to obtain the optimized weights, mean vectors, and covariance matrices; Perform derivative calculations based on the preset zero derivatives, optimized weights, mean vectors, and covariance matrices to obtain optimized model parameters; Optimize the Gaussian mixture model based on the optimized weights, mean vectors, covariance matrices, and optimized model parameters to obtain a postoperative rehabilitation evaluation model.

6. The method for evaluating postoperative rehabilitation of bones according to claim 1, wherein The ant colony search model constructed based on the preset ant colony algorithm, bone scan image data, and rehabilitation evaluation task includes: Calculate the weighted image data based on the bone scan image data and the rehabilitation evaluation task; Perform multi-level coarsening operations on the weighted image data based on the preset order to obtain N-level image parameter data; Construct an ant colony search model based on the N-level image parameter data and the ant colony algorithm.

7. The bone postoperative rehabilitation evaluation method according to claim 1, wherein The iterative search for the support vector machine model based on the ant colony search model to obtain a set of path node parameters includes: Use the numerical differentiation method to perform derivative calculations on the medical record information data to obtain the first derivative and the second derivative; Calculate the weights and bias terms based on the first derivative, the second derivative, and the preset partial derivative; Generate a pheromone concentration calculation formula based on the weights and bias terms. The expression of the pheromone concentration calculation formula is as follows: , where is the pheromone concentration, is the weight, is a variable related to the first derivative and the second derivative, is the bias term; Use the linear discriminant analysis method to reduce the dimension of the medical record information data to obtain the reduced-dimensional case data; Calculate the pheromone concentration based on the pheromone concentration calculation formula for the reduced-dimensional case data; Obtain the penalty parameter and kernel function parameter of the support vector machine model, and set the penalty parameter and kernel function parameter as path nodes; Perform iterative search on the support vector machine model based on the ant colony search model, path nodes, and pheromone concentration to obtain a set of path node parameters.

8. A postoperative rehabilitation evaluation device for bones, characterized in that, Including: The first data acquisition module is used to acquire bone scan image data and rehabilitation evaluation tasks; The first model construction module is used to construct an ant colony search model based on the preset ant colony algorithm, bone scan image data, and rehabilitation evaluation task; The second data acquisition module is used to acquire the patient's historical natural walking data and medical record information data; The second model construction module is used to construct a support vector machine model based on the preset support vector machine algorithm, historical natural walking data, and medical record information data; The parameter set acquisition module is used to perform iterative search on the support vector machine model based on the ant colony search model to obtain a set of path node parameters; The third model construction module is used to construct a postoperative rehabilitation evaluation model based on the set of path node parameters, historical natural walking data, bone scan image data, medical record information data, and the preset Gaussian mixture algorithm; The rehabilitation evaluation module is used to acquire real-time natural walking data and use the postoperative rehabilitation evaluation model for the real-time natural walking data to obtain a rehabilitation evaluation result.

9. A post-operative rehabilitation assessment device for bones, characterized in that, The described bone postoperative rehabilitation evaluation device includes: a memory and at least one processor, and instructions are stored in the memory; At least one of the processors invokes the instructions in the memory so that the bone postoperative rehabilitation evaluation device executes each step of the bone postoperative rehabilitation evaluation method described in any one of claims 1-7.

10. A computer-readable storage medium, on which instructions are stored, characterized in that, When the instruction is executed by a processor, it implements each step of a method for evaluating postoperative rehabilitation of bones as described in any one of claims 1-7.

Images