Effect evaluation system for rehabilitation training of lower limbs of children

Abstract

The invention relates to the technical field of training effect evaluation, and provides an effect evaluation system for rehabilitation training of lower limbs of children, which comprises a processor, and a multi-modal sensor module, a data processing module, a training effect evaluation module and a prediction module which are connected with the processor. The multi-mode sensor module monitors and collects physiological data in the rehabilitation process of the child femoral distal closed fracture. The data processing module receives and processes sensor data and extracts features; the training effect evaluation module comprises a growth plate state feature evaluation unit, a skeleton growth feature evaluation unit, a physiological feature evaluation unit and a training effect evaluation unit, and evaluation is performed through a preset feature evaluation formula; and the prediction module uses a random forest model, inputs real-time feature data and predicts a training effect, including data collection, data cleaning and preprocessing, model initialization and training, model evaluation and optimization and evaluation value prediction, so as to evaluate and predict the training effect of the child patient with the distal closed femoral fracture.

Description

Technical Field

[0001] The present invention relates to the technical field of training effect evaluation, and more specifically, to an effect evaluation system for children's lower limb rehabilitation training. Background Art

[0002] The uniqueness of the skeletal system of children lies in its rapid growth and development process, in which the growth plate plays a vital role. The growth plate, also known as the epiphysis or epiphysis, is a special cartilage tissue located at both ends of the long bone. It is a key structure for bone growth in children and adolescents and plays a vital role in the longitudinal growth of bones. The main function of the growth plate is to promote the longitudinal growth of bones. Through the continuous proliferation, differentiation and matrix production of chondrocytes, the growth plate enables the bones to continue to grow in length. This process is regulated by many factors, including endocrine factors such as growth hormone, thyroid hormone, sex hormone, and external factors such as mechanical stress and nutritional status; in addition, the growth plate is also involved in bone shaping, so that the bones maintain the appropriate shape during growth. Because children are in a critical period of growth and development, their bone structure and density will change significantly, and the growth plate itself is also susceptible to injury, especially in sports injuries in children and adolescents. If the growth plate injury is not handled properly, it may lead to abnormal or deformed bone growth. Therefore, during the rehabilitation training of children's leg fractures, special attention must be paid to the potential impact on the growth plate and overall bone development. Improper training programs may interfere with the normal function of the growth plate and affect bone growth and development.

[0003] In the existing public literature, in document 1 (Pseudo-online Multimodal Interface Based on Movement Prediction for Lower Limbs Rehabilitation.2017), Botelho et al. proposed a lower limb rehabilitation system based on a multimodal interface, which detects lower limb movement intention by using EEG, EMG and inertial sensor data. However, although the system can effectively predict movement intention, it has certain deficiencies in the quantitative evaluation of specific rehabilitation effects. It fails to pay attention to the pain points of children's growth and development, and their bone structure and density will change significantly. It ignores key factors such as growth plate status and bone growth, resulting in incomplete evaluation results. At the same time, it lacks foresight and cannot make full use of existing information to enhance the understanding of the rehabilitation process. In document 2 (A wearable multimode system with soft sensors for lower limb activity evaluation and rehabilitation.2018), Bao et al. developed a wearable multimodal system that can obtain movement and physiological data in lower limb rehabilitation training, but its data processing is relatively simple, especially in the real-time evaluation of bone growth and physiological characteristics recovery, lacking precise quantitative indicators and comprehensive evaluation models.

[0004] In order to solve the above problems, a technical solution is now provided. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides an effect evaluation system for children's lower limb rehabilitation training, which collects growth plate, bone and physiological characteristic data through multimodal sensors, combines data processing modules and random forest models, evaluates rehabilitation effects in real time, and predicts rehabilitation progress during the training cycle. The problem of the lack of comprehensive evaluation of children's bone growth, especially the state of growth plates, in the prior art is solved, thereby providing a basis for the formulation of personalized rehabilitation training programs.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A system for evaluating the effect of rehabilitation training for children's lower limbs, comprising a processor and a multimodal sensor module connected to the processor, a data processing module, a training effect evaluation module and a prediction module, wherein the multimodal sensor module is connected to the data processing module, the data processing module is respectively connected to the training effect evaluation module and the prediction module, and the training effect evaluation module is connected to the prediction module; wherein the training effect evaluation module comprises a growth plate state feature evaluation unit, a bone growth feature evaluation unit, a physiological feature evaluation unit and a training effect evaluation unit, and is used to extract growth plate state features, bone growth features and physiological features from processed data and evaluate them through a preset feature evaluation formula; wherein the training effect evaluation unit is connected to the growth plate state feature evaluation unit, the bone growth feature evaluation unit and the physiological feature evaluation unit, and the growth plate state feature evaluation value, the bone growth feature evaluation value and the physiological feature evaluation value are used as input values ​​of a training effect evaluation model in the training effect evaluation unit, and the effect of the rehabilitation training cycle of children with closed distal femoral fractures is evaluated, and the formula of the training effect evaluation model is:

[0008]

[0009] In the formula, TEI is the training effect evaluation index, GPSF is the growth plate state characteristic evaluation value, BGSF is the bone growth characteristic evaluation value, and RCF is the physiological characteristic evaluation value;

[0010] The real-time growth plate status characteristics, bone growth characteristics and physiological characteristics are input into the constructed random forest module, and the model outputs the predicted value of the training effect evaluation index for children with closed distal femoral fractures.

[0011] As a further solution of the present invention, a multimodal sensing module is used to monitor and collect physiological data during the rehabilitation process of children with closed distal femoral fractures. Sensors are distributed in the preset affected area and surrounding areas to collect bone structure dimensions, growth plate force distribution and blood flow, femoral bone structure and density, joint mobility and muscle strength.

[0012] As a further solution of the present invention, the data processing module is used to receive sensor data transmitted by the multimodal sensing module through wireless communication technology, and use a bandpass filter to remove high-frequency noise and low-frequency drift waves; calibrate the original data according to the calibration curve of each sensor; extract growth plate state characteristics, bone growth characteristics and physiological characteristics from the filtered and calibrated data, the growth plate state characteristics include the total thickness of the growth plate, the elastic modulus and stress at different depths of the growth plate, the blood flow through the growth plate and the cross-sectional area of ​​the growth plate; the bone growth characteristics include the femur length and bone density at different positions of the femur; the physiological characteristics include callus volume, knee joint range of motion and quadriceps strength.

[0013] As a further solution of the present invention, the growth plate state characteristic evaluation unit evaluates the growth plate state characteristic according to the stress-elasticity distribution integral from the epiphyseal end to the diaphyseal end, the growth dynamics and the blood supply inside the growth plate, wherein the formula for calculating the growth plate state evaluation value is:

[0014]

[0015] Where GPSF is the growth plate status characteristic evaluation value, h is the total thickness of the growth plate, z is the depth from the epiphyseal end to the diaphyseal end, E(z) is the elastic modulus at depth z, σ(z) is the stress at depth z, t is the duration of the training cycle, Q is the blood flow through the growth plate, A is the cross-sectional area of ​​the growth plate, and k1 and k2 are weight coefficients.

[0016] As a further solution of the present invention, the bone growth characteristic evaluation unit calculates the bone growth characteristic evaluation value according to the bone density distribution integral on the femur length, the femur length growth rate and the bone density change rate, and evaluates the bone growth characteristics, wherein the formula for calculating the bone growth characteristic evaluation value is:

[0017]

[0018] Where BGSF is the bone growth characteristic assessment value, L is the femur length, x is the position from the proximal to the distal end of the femur, BMD(x) is the bone density at position x, and t is the time of the training cycle.

[0019] As a further solution of the present invention, the physiological characteristic evaluation unit evaluates the physiological characteristics according to the callus volume, joint mobility and muscle strength recovery, wherein the formula for calculating the physiological characteristic evaluation value is:

[0020]

[0021] In the formula, RCF is the physiological characteristic evaluation value, CV is the callus volume, and CV max is the expected maximum callus volume, ROM is the range of motion of the affected knee, and ROM normal is the range of motion of the healthy knee, MF is the strength of the quadriceps femoris on the affected side, and MF normal It is the strength of the quadriceps on the healthy side.

[0022] As a further solution of the present invention, constructing a random forest model comprises the following steps:

[0023] Step S1, collecting original characteristic data of historical children, including total thickness of growth plate, elastic modulus and stress at different depths of growth plate, blood flow through growth plate and cross-sectional area of ​​growth plate, femur length and bone density at different positions of femur, callus volume, knee joint range of motion and quadriceps strength, and collecting corresponding training effect evaluation index;

[0024] Step S2: clean the collected raw data, fill missing values ​​with the median, and delete outliers; standardize all continuous features; and split the processed data set into a training set and a test set in a ratio of 8:2;

[0025] Step S3, initialize the random forest model, set the number of trees and the maximum depth; use the training set data to train the model, take all the original features as input, and predict the training effect evaluation index;

[0026] Step S4, evaluate the model performance using the test set data; calculate the mean square error and R-squared score; analyze the feature importance and determine the features that have the greatest impact on the prediction results; optimize the model hyperparameters using cross-validation and grid search methods;

[0027] Step S5, inputting the original characteristic data of the new patient; using the constructed random forest model to predict the patient's training effect evaluation index; and outputting the predicted value of the training effect evaluation index.

[0028] Compared with the prior art, the beneficial effects of the effect evaluation system for children's lower limb rehabilitation training of the present invention are:

[0029] The present invention uses a multimodal sensor module to monitor key physiological data in the rehabilitation process of children with closed distal femoral fractures in real time, and can more comprehensively evaluate the growth plate status, bone growth and physiological characteristics, solving the problem of ignoring the characteristics of children's bone growth, especially the impact of growth plate status on rehabilitation effects in the prior art. In addition, especially for the characteristics of children's rapid growth and development, an evaluation model related to bone development is constructed, which can dynamically track the changes in bones and growth plates during the rehabilitation process, overcoming the limitations of the prior art in lacking quantitative evaluation of children's bone development, density changes, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 The present invention is a schematic structural diagram of an effect evaluation system for children's lower limb rehabilitation training. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] Example 1

[0033] A system for evaluating the effect of children's lower limb rehabilitation training comprises a processor and a multimodal sensor module connected to the processor, a data processing module, a training effect evaluation module and a prediction module.

[0034] The multimodal sensor module in the embodiment of the present invention is connected to the data processing module, the data processing module is respectively connected to the training effect evaluation module and the prediction module, and the training effect evaluation module is connected to the prediction module.

[0035] The processor in the embodiment of the present invention adopts the high-performance embedded processor ARM Cortex-A72 as the computing unit of the system, providing a powerful and efficient computing platform for the training effect evaluation system of children's leg fractures. As the core of the system, the processor not only has excellent performance and low power consumption characteristics, but also can coordinate and manage the operation of the entire system. Its high-performance computing capability enables the system to process a large amount of raw data from the multimodal sensor module in real time, and quickly execute the filtering, calibration and feature extraction algorithms in the data processing module. In the training effect evaluation module, the processor can accurately calculate the growth plate state feature evaluation value, the bone growth feature evaluation value and the physiological feature evaluation value, ensuring the accuracy of the evaluation results. In addition, in the prediction module, the powerful algorithm execution capability of ARM Cortex-A72 enables it to efficiently run complex random forest algorithms and complete the training, evaluation and prediction tasks of the model. This integrated solution not only simplifies the system architecture and improves the overall efficiency, but also provides good flexibility and scalability for the system. The real-time performance and rapid response capability of the processor enable medical personnel to obtain evaluation results and predictions in a timely manner, which helps to adjust the rehabilitation training program in a timely manner. In general, the application of the ARM Cortex-A72 processor provides a solid hardware foundation for the present invention and is the key to achieving high performance, multi-function and accurate evaluation of the system, which greatly improves the accuracy and efficiency of the evaluation of the training effect of children's leg fractures.

[0036] The multimodal sensing module is used to monitor and collect physiological data during the rehabilitation process of children with closed distal femoral fractures. Sensors are distributed in the preset affected area and surrounding areas to collect bone structure dimensions, growth plate force distribution and blood flow, femoral bone structure and density, joint range of motion, and muscle strength.

[0037] The multimodal sensing module in the embodiment of the present invention includes a high-precision ultrasonic or optical sensor for measuring the size of the bone structure, which is placed along the outer side of the thigh from the hip joint to the knee joint to regularly measure the length and diameter changes of the femur; a thin film pressure sensor covering the distal growth plate area of ​​the femur, fixed by a special medical tape, to monitor the force distribution of the growth plate in daily activities in real time; a non-invasive near-infrared spectroscopy sensor placed on the skin surface above the growth plate area to monitor the blood flow changes in deep tissues; a portable ultrasonic bone densitometer to regularly measure the bone density of different parts of the femur; a flexible angle sensor fixed around the knee joint to measure the flexion and extension angle and range of motion of the knee joint; and a portable electromyography sensor for measuring the quadriceps femoris activity potential to evaluate the recovery of muscle strength. These sensors use wireless transmission technology to transmit the collected data to the data processing module of the system in real time. The sampling frequency and working mode can be adjusted according to the rehabilitation stage and doctor's advice to ensure the comprehensiveness and accuracy of the data. Through this multimodal sensing system, the present invention can comprehensively and continuously obtain key physiological data in the rehabilitation process of children's closed distal femoral fractures, including bone structure dimensions, growth plate force distribution and blood flow, femoral bone structure and density, joint mobility and muscle strength, providing reliable basic support for subsequent training effect evaluation and rehabilitation progress prediction.

[0038] The data processing module is used to receive sensor data transmitted by the multimodal sensing module through wireless communication technology, and use a bandpass filter to remove high-frequency noise and low-frequency drift waves; calibrate the original data according to the calibration curve of each sensor; extract growth plate state characteristics, bone growth characteristics and physiological characteristics from the filtered and calibrated data, the growth plate state characteristics include the total thickness of the growth plate, the elastic modulus and stress at different depths of the growth plate, the blood flow through the growth plate and the cross-sectional area of ​​the growth plate; the bone growth characteristics include the femur length and the bone density at different positions of the femur; the physiological characteristics include the callus volume, knee joint range of motion and quadriceps strength.

[0039] The data processing module in the embodiment of the present invention processes, analyzes and extracts features from the raw data collected by the multimodal sensing module, thereby providing a reliable data basis for subsequent rehabilitation evaluation and prediction. It comprehensively reflects the rehabilitation status of children's closed distal femoral fractures from multiple dimensions, not only focusing on fracture healing itself, but also considering the recovery of growth plate function, bone growth and related physiological functions, providing doctors with a more comprehensive and accurate basis for rehabilitation evaluation. At the same time, through standardized data processing procedures, the module can effectively reduce data errors, improve the consistency and reliability of feature extraction, and lay a solid data foundation for subsequent rehabilitation progress prediction and personalized rehabilitation program formulation.

[0040] The training effect evaluation module includes a growth plate state characteristic evaluation unit, a bone growth characteristic evaluation unit, a physiological characteristic evaluation unit and a training effect evaluation unit, which are used to extract growth plate state characteristics, bone growth characteristics and physiological characteristics from the processed data and evaluate them through a preset characteristic evaluation formula.

[0041] The training effect evaluation unit in the embodiment of the present invention is connected to the growth plate state characteristic evaluation unit, the bone growth characteristic evaluation unit and the physiological characteristic evaluation unit.

[0042] The growth plate state characteristic evaluation unit in the embodiment of the present invention evaluates the growth plate state characteristic according to the stress-elastic distribution integral from the epiphyseal end to the diaphyseal end, the growth dynamics and the blood supply inside the growth plate, wherein the formula for calculating the growth plate state evaluation value is:

[0043]

[0044] Where GPSF is the growth plate status characteristic evaluation value, h is the total thickness of the growth plate, z is the depth from the epiphyseal end to the diaphyseal end, E(z) is the elastic modulus at depth z, σ(z) is the stress at depth z, t is the duration of the training cycle, Q is the blood flow through the growth plate, A is the cross-sectional area of ​​the growth plate, and k1 and k2 are weight coefficients.

[0045] The formula for calculating the growth plate status evaluation value in the embodiment of the present invention integrates the mechanical properties, growth dynamics and blood supply of the growth plate. The core of the formula is the integral term Where h represents the total thickness of the growth plate, z is the depth variable from the epiphysis to the diaphysis, σ(z) and E(z) represent the stress and elastic modulus at depth z, respectively. This integral comprehensively reflects the mechanical state and potential growth capacity of the growth plate by calculating the cumulative ratio of stress to elastic modulus over the entire thickness of the growth plate. The term introduces the time dimension, where represents the rate of change of the thickness of the growth plate, that is, the growth rate, which makes the evaluation no longer limited to the static state, but can capture the dynamic changes of the growth plate. The term takes into account the blood supply factor, Q represents the blood flow, A is the cross-sectional area of ​​the growth plate, and their ratio reflects the blood perfusion per unit area. k1 and k2 serve as weight coefficients, giving the formula the flexibility to fine-tune according to clinical experience or specific circumstances. The design of the entire formula reflects a deep understanding of the physiology and biomechanics of the growth plate, taking into account not only the static structural characteristics but also the dynamic growth and metabolic factors. By integrating these key parameters into the growth plate status characteristic evaluation value, the formula facilitates the comparison of different patients or different periods.

[0046] The bone growth characteristic evaluation unit in the embodiment of the present invention calculates the bone growth characteristic evaluation value according to the bone density distribution integral on the femur length, the femur length growth rate and the bone density change rate, and evaluates the bone growth characteristics, wherein the formula for calculating the bone growth characteristic evaluation value is:

[0047]

[0048] Where BGSF is the bone growth characteristic assessment value, L is the femur length, x is the position from the proximal to the distal end of the femur, BMD(x) is the bone density at position x, and t is the time of the training cycle.

[0049] The formula for calculating the bone growth characteristic evaluation value in the embodiment of the present invention combines three key factors: bone density distribution, bone length growth rate and dynamic changes in bone density. The first term of the formula is an integral, where L represents the total length of the femur, x is the position variable from the proximal to the distal end of the femur, and BMD(x) represents the bone density at position x. This integral comprehensively reflects the overall mineralization status and strength distribution of the bone by calculating the accumulation of bone density over the entire length of the femur. The term introduces a time dimension, representing the growth rate of femur length, and directly quantifies the longitudinal growth of the bone. is another integral that calculates the rate of change of bone density over time accumulated over the entire length of the femur, reflecting the dynamic changes in bone density. The combination of these three items enables the bone growth characteristic assessment value to capture multiple aspects of bone growth: static bone density distribution, dynamic length growth, and temporal changes in bone density. The design of the formula reflects a deep understanding of the physiology of bone growth, taking into account not only the current state of the bone, but also its growth dynamics. By integrating these key parameters into a comprehensive indicator, the formula provides a basis for evaluating the bone growth status of different patients or the same patient at different times, enabling doctors to more accurately grasp the bone growth during fracture healing, including changes in bone density distribution and length growth, thereby providing an important reference for formulating and adjusting personalized rehabilitation plans. Especially in the rehabilitation process of closed distal femoral fractures in children, this assessment indicator can help doctors monitor the quality of bone healing, predict potential growth problems, and adjust treatment strategies in a timely manner, ultimately improving patients' prognosis and quality of life.

[0050] The physiological characteristic evaluation unit in the embodiment of the present invention evaluates the physiological characteristics according to the callus volume, joint range of motion and muscle strength recovery, wherein the formula for calculating the physiological characteristic evaluation value is:

[0051]

[0052] In the formula, RCF is the physiological characteristic evaluation value, CV is the callus volume, and CV max is the expected maximum callus volume, ROM is the range of motion of the affected knee, and ROM mormal is the range of motion of the healthy knee, MF is the strength of the quadriceps femoris on the affected side, and MF normal It is the strength of the quadriceps on the healthy side.

[0053] The formula for calculating the physiological characteristic evaluation value in the embodiment of the present invention integrates three key aspects: callus formation, joint function and muscle strength. In the formula, RCF represents the physiological characteristic evaluation value, which is a quantitative expression of the overall recovery status. CV represents the volume of the callus, which reflects the degree and quality of fracture healing, while CV max The maximum callus volume during the first period was used as the reference standard. ROM represents the range of motion of the affected knee joint. normal is the range of motion of the healthy knee joint, and the ratio of the two is calculated by the logarithmic function This design can not only reflect the degree of joint function recovery, but also stabilize when it is close to normal. MF represents the strength of the quadriceps femoris on the affected side. normal is the strength of the quadriceps on the healthy side, and their ratio is expressed in the inverse form Appearing in the denominator, the cubic design highlights the importance of muscle recovery. Taking into account callus formation and joint function, the denominator The maximum callus volume and muscle strength recovery are combined to form a dynamic balance. It not only considers the absolute value of each indicator, but also provides a more relative evaluation by comparing it with normal or expected values. As rehabilitation progresses, the callus volume increases, the joint range of motion improves, and muscle strength is restored, the physiological characteristic evaluation value will increase accordingly. It enables doctors to more accurately grasp multiple key aspects of fracture healing, including bone tissue repair, functional recovery, and muscle strength reconstruction, thereby providing a reference for formulating and adjusting personalized rehabilitation plans.

[0054] The training effect evaluation unit in the embodiment of the present invention uses the growth plate state characteristic evaluation value, the bone growth characteristic evaluation value and the physiological characteristic evaluation value as the input value of the training effect evaluation model in the training effect evaluation unit to evaluate the effect of the rehabilitation training cycle of children with closed distal femoral fractures. The formula of the training effect evaluation model is:

[0055]

[0056] Wherein, TEI is the training effect evaluation index, GPSF is the growth plate state characteristic evaluation value, BGSF is the bone growth characteristic evaluation value, and RCF is the physiological characteristic evaluation value.

[0057] The training effect evaluation model in the embodiment of the present invention integrates three dimensions: growth plate status, bone growth characteristics and physiological characteristics. In the formula, the training effect evaluation index is a quantitative expression of the overall rehabilitation effect; the growth plate status characteristic evaluation value reflects the health status and functional recovery degree of the growth plate. Through logarithmic processing, it not only retains the influence of the growth plate status characteristic evaluation value, but also avoids its possible over-dominance; the bone growth characteristic evaluation value reflects the overall situation of bone growth, including bone density distribution and length growth; the square root design not only highlights the importance of bone growth, but also smoothes its influence; the physiological characteristic evaluation value comprehensively reflects the formation of callus, joint mobility and muscle strength recovery. The application of this Sigmoid function makes the impact of the physiological characteristic evaluation value on the training effect evaluation index most significant at a medium level, and tends to be stable at extremely high or low values, which is consistent with the nonlinear impact of physiological characteristics on rehabilitation effects. The interaction between growth plate status and bone growth is comprehensively considered, and the denominator introduces the influence of physiological characteristics through the Sigmoid function to form a dynamic balance. Not only the independent contribution of each indicator is considered, but also the complex interaction between them is reflected through mathematical processing. As rehabilitation progresses, growth plate function improves, bone growth normalizes, and physiological function recovers, the training effect evaluation index will rise accordingly, reflecting the evaluation of the rehabilitation effect.

[0058] The real-time growth plate status characteristics, bone growth characteristics and physiological characteristics are input into the constructed random forest module, and the model outputs the predicted value of the training effect evaluation index for children with closed distal femoral fractures.

[0059] The construction of the random forest model in the embodiment of the present invention includes the following steps:

[0060] Step S1, collecting original characteristic data of historical children, including total thickness of growth plate, elastic modulus and stress at different depths of growth plate, blood flow through growth plate and cross-sectional area of ​​growth plate, femur length and bone density at different positions of femur, callus volume, knee joint range of motion and quadriceps strength, and collecting corresponding training effect evaluation index;

[0061] Step S2: clean the collected raw data, fill missing values ​​with the median, and use the interquartile range method to identify and delete outliers; use the Z-score standardization method to standardize all continuous features; split the processed data set into a training set and a test set in a ratio of 8:2, and use a stratified sampling method to ensure that the distribution of the target variable in the two sets is consistent;

[0062] Step S3, initialize the random forest model, set the number of trees to 100, and the maximum depth to 10; then use the training set data to train the model, take all the preprocessed original features as input, and predict the training effect evaluation index; during the training process, use the bootstrap sampling method to create a sub-training set for each decision tree, which helps to improve the generalization ability of the model; for each node split, randomly select sqrt(m) features as candidate features, where m is the total number of features. This randomness helps to reduce the risk of overfitting of the model and improve its adaptability to new data;

[0063] Step S4, evaluate the model performance using the test set data; calculate the mean square error and R square score; analyze the feature importance and determine the features that have the greatest impact on the prediction results; use Gini impurity or mean precision reduction to quantify the importance of each feature, which helps to understand which factors have the greatest impact on the training effect; then, use k-fold cross-validation to evaluate the stability of the model to ensure that the model performs consistently on different subsets; finally, use the grid search method to optimize the model hyperparameters, with the optimization goal of minimizing the mean square error of cross-validation; through this series of evaluation and optimization steps, not only can we fully understand the performance of the model, but also further improve its prediction accuracy and generalization ability;

[0064] Step S5, input the original characteristic data of the new child, and perform the same preprocessing steps on the input data as the training data, including missing value processing and standardization, to ensure that the new data is in the same format as the data during model training; use the constructed random forest model to predict the child's training effect evaluation index; output the predicted value of the training effect evaluation index; and provide a confidence interval for the prediction to reflect the reliability of the prediction.

[0065] The following is a Python code example that uses a random forest model to obtain the predicted results of the training scenario evaluation index. Please note that this example is only a starting point and may need to be adjusted according to the actual situation and device interface in actual application;

[0066]

[0067]

[0068]

[0069] This code is only used as an example. In actual applications, appropriate modifications and adjustments need to be made according to specific circumstances.

[0070] The prediction module in the embodiment of the present invention proposes a method based on a random forest model to evaluate the effect of rehabilitation training for children with closed distal femoral fractures. The method includes five main steps, which constitute a complete machine learning process; first, historical child data is collected, including multi-dimensional information such as growth plate characteristics, bone growth indicators, and physiological parameters, providing rich training data for the model; secondly, a series of data preprocessing techniques, such as missing value filling, outlier processing, and feature standardization, are used to ensure data quality and consistency; in the model construction stage, the random forest algorithm is used to enhance the generalization ability and robustness of the model by setting appropriate parameters and using bootstrap sampling and other techniques; subsequently, the method uses a variety of evaluation indicators and techniques, such as mean square error, feature importance analysis, and cross-validation to evaluate and optimize model performance; finally, for new children, the method can preprocess input data, use the trained model for prediction, and provide confidence intervals for prediction results, providing reference for clinical decision-making.

[0071] Example 2

[0072] The system first uses a multimodal sensor module to arrange multiple sensors on the affected area and surrounding areas to comprehensively collect key physiological data such as bone structure, growth plate status, blood flow and muscle strength; these raw data are then transmitted to the data processing module for filtering and denoising, data calibration and feature extraction, and converted into growth plate status characteristics, bone growth characteristics and physiological characteristic indicators; then, the training effect evaluation module receives these processed feature data, and conducts in-depth analysis of growth plate status, bone growth and physiological characteristics through three special evaluation units. Each unit uses a specific evaluation formula to comprehensively consider multiple factors to obtain the corresponding evaluation value, and finally calculates the training effect evaluation index through the preset model; the system also includes a prediction module based on the random forest algorithm, which first collects and organizes a large amount of historical child data, and after cleaning and standardization, it is used to train the random forest model, and continuously optimizes the model performance through techniques such as feature importance analysis, cross-validation and hyperparameter optimization; finally, the system can evaluate and predict the data of new children in real time, and quickly calculate the predicted training effect evaluation index by inputting the latest feature data.

[0073] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0074] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A system for evaluating the effect of children's lower limb rehabilitation training, comprising a processor and a multimodal sensor module connected to the processor, a data processing module, a training effect evaluation module and a prediction module, characterized in that: The multimodal sensor module is connected to the data processing module, the data processing module is connected to the training effect evaluation module and the prediction module respectively, and the training effect evaluation module is connected to the prediction module; wherein the training effect evaluation module includes a growth plate state feature evaluation unit, a bone growth feature evaluation unit, a physiological feature evaluation unit and a training effect evaluation unit, which are used to extract growth plate state features, bone growth features and physiological features from the processed data and evaluate them through a preset feature evaluation formula; wherein the training effect evaluation unit is connected to the growth plate state feature evaluation unit, the bone growth feature evaluation unit and the physiological feature evaluation unit, and the growth plate state feature evaluation value, the bone growth feature evaluation value and the physiological feature evaluation value are used as input values ​​of the training effect evaluation model in the training effect evaluation unit to evaluate the effect of the rehabilitation training cycle of children with closed distal femoral fractures, and the formula of the training effect evaluation model is: In the formula, TEI is the training effect evaluation index, GPSF is the growth plate state characteristic evaluation value, BGSF is the bone growth characteristic evaluation value, and RCF is the physiological characteristic evaluation value; The real-time growth plate status characteristics, bone growth characteristics and physiological characteristics are input into the constructed random forest module, and the model outputs the predicted value of the training effect evaluation index for children with closed distal femoral fractures.

2. The effect evaluation system for children's lower limb rehabilitation training according to claim 1, characterized in that: The multimodal sensing module is used to monitor and collect physiological data during the rehabilitation process of children with closed distal femoral fractures. Sensors are distributed in the preset affected area and surrounding areas to collect bone structure dimensions, growth plate force distribution and blood flow, femoral bone structure and density, joint range of motion, and muscle strength.

3. The effect evaluation system for children's lower limb rehabilitation training according to claim 1, characterized in that: The data processing module is used to receive sensor data transmitted by the multimodal sensing module through wireless communication technology, and use a bandpass filter to remove high-frequency noise and low-frequency drift waves; calibrate the original data according to the calibration curve of each sensor; extract growth plate state characteristics, bone growth characteristics and physiological characteristics from the filtered and calibrated data, the growth plate state characteristics include the total thickness of the growth plate, the elastic modulus and stress at different depths of the growth plate, the blood flow through the growth plate and the cross-sectional area of ​​the growth plate; the bone growth characteristics include the femur length and the bone density at different positions of the femur; the physiological characteristics include the callus volume, knee joint range of motion and quadriceps strength.

4. The effect evaluation system for children's lower limb rehabilitation training according to claim 1, characterized in that: The growth plate state characteristic evaluation unit evaluates the growth plate state characteristics according to the stress-elasticity distribution integral from the epiphyseal end to the diaphyseal end, the growth dynamics and the blood supply inside the growth plate, wherein the formula for calculating the growth plate state evaluation value is: Where GPSF is the growth plate status characteristic evaluation value, h is the total thickness of the growth plate, z is the depth from the epiphyseal end to the diaphyseal end, E(z) is the elastic modulus at depth z, σ(z) is the stress at depth z, t is the duration of the training cycle, Q is the blood flow through the growth plate, A is the cross-sectional area of ​​the growth plate, and k1 and k2 are weight coefficients.

5. The effect evaluation system for children's lower limb rehabilitation training according to claim 1, characterized in that: The bone growth characteristic evaluation unit calculates the bone growth characteristic evaluation value according to the bone density distribution integral on the femur length, the femur length growth rate and the bone density change rate, and evaluates the bone growth characteristics, wherein the formula for calculating the bone growth characteristic evaluation value is: Where BGSF is the bone growth characteristic assessment value, L is the femur length, x is the position from the proximal to the distal end of the femur, BMD(x) is the bone density at position x, and t is the time of the training cycle.

6. The effect evaluation system for children's lower limb rehabilitation training according to claim 1, characterized in that: The physiological characteristic evaluation unit evaluates the physiological characteristics according to the callus volume, joint mobility and muscle strength recovery. The formula for calculating the physiological characteristic evaluation value is: In the formula, RCF is the physiological characteristic evaluation value, CV is the callus volume, and CV max is the expected maximum callus volume, ROM is the range of motion of the affected knee, and ROM normal is the range of motion of the healthy knee, MF is the strength of the quadriceps femoris on the affected side, and MF normal It is the strength of the quadriceps on the healthy side.

7. The effect evaluation system for children's lower limb rehabilitation training according to claim 1, characterized in that: Building a random forest model involves the following steps: Step S1, collecting original characteristic data of historical children, including total thickness of growth plate, elastic modulus and stress at different depths of growth plate, blood flow through growth plate and cross-sectional area of ​​growth plate, femur length and bone density at different positions of femur, callus volume, knee joint range of motion and quadriceps strength, and collecting corresponding training effect evaluation index; Step S2: clean the collected raw data, fill missing values ​​with the median, and delete outliers; standardize all continuous features; and split the processed data set into a training set and a test set in a ratio of 8:2; Step S3, initialize the random forest model, set the number of trees and the maximum depth; use the training set data to train the model, take all the original features as input, and predict the training effect evaluation index; Step S4, evaluate the model performance using the test set data; calculate the mean square error and R-squared score; analyze the feature importance and determine the features that have the greatest impact on the prediction results; optimize the model hyperparameters using cross-validation and grid search methods; Step S5, inputting the original characteristic data of the new patient; using the constructed random forest model to predict the patient's training effect evaluation index; and outputting the predicted value of the training effect evaluation index.

Images