Method and system for predicting fracture risk of osteoporosis patient

Abstract

The invention discloses a prediction method and a prediction system for fracture risk of an osteoporosis patient. The method comprises the following steps: firstly, collecting a plurality of parameters such as age, bone mineral density, vitamin D level and the like of the patient; constructing a bone time-varying factor and dynamic prediction model, and determining the influence weight of each parameter on the bone condition at different time nodes; then, a three-dimensional mechanical model is constructed through mechanical simulation of the skeleton microstructure, stress distribution, strain states and structural stability of the skeleton under different loads are simulated, and the model is corrected by fusing parameters to influence weights; then the fracture probability in a future preset time period is calculated; and finally, a result is presented through a nomogram visualization model. The system realizes the above method, can predict the fracture risk more accurately, and provides powerful support for clinical diagnosis and treatment.

Description

Technical Field

[0001] This invention relates to the field of fracture risk prediction, and more particularly to a method and system for predicting fracture risk in patients with osteoporosis. Background Technology

[0002] With the accelerating aging of the population, the incidence of osteoporosis is rising year by year, and fractures caused by it have become a significant threat to the health of the elderly. Osteoporosis patients have increased bone fragility, and even minor external forces can lead to fractures, causing not only limb dysfunction and a decline in quality of life but also increasing the medical burden on families and society. Accurately predicting the fracture risk of osteoporosis patients can provide clinicians with a basis for developing personalized prevention and treatment plans, helping to take early intervention measures and reduce the fracture incidence. Currently, clinical assessment of fracture risk in osteoporosis patients mostly relies on single indicators or simple models. However, because bone condition is dynamically affected by multiple factors, existing assessment methods cannot fully reflect the true risk level of patients, and there is an urgent need for more accurate and systematic prediction methods and systems.

[0003] Existing technologies for predicting fracture risk in osteoporosis patients have two significant drawbacks. Firstly, existing prediction models often neglect the time-varying characteristics of bone condition, failing to fully consider the dynamic changes and interactions of various patient parameters over time. This results in predictions that are difficult to reflect long-term risk trends, particularly in capturing risk changes during disease progression. Secondly, existing technologies lack precision in simulating bone biomechanical properties, often relying on simplified calculations based on only a few parameters such as bone density. They fail to incorporate three-dimensional mechanical modeling of bone microstructure characteristics, making it impossible to accurately simulate the stress-strain state of bone under different loads. Consequently, fracture risk assessment based on mechanical analysis lacks sufficient biomechanical basis, affecting the accuracy and reliability of predictions. Summary of the Invention

[0004] In order to overcome the shortcomings and deficiencies of the existing technology, the present invention provides a method and system for predicting the fracture risk of osteoporosis patients.

[0005] The technical solution adopted in this invention is a method for predicting fracture risk in osteoporosis patients, comprising the following steps: Step S1. Collect different parameters of osteoporosis patients, including patient age, gender, body mass index, bone mineral density test value, history of fracture, smoking status, frequency of drinking, type and duration of taking drugs that affect bone metabolism, serum calcium content, serum phosphorus content, vitamin D level, parathyroid hormone level, and bone turnover marker level. Step S2. Construct a time-varying factor and dynamic prediction model for bone quality based on the parameters. Analyze the changing trend of the parameters over time using this model to determine the influence weight of different parameters on bone quality at different time points. Step S3. Using the mechanical simulation method of bone microstructure, a three-dimensional mechanical model of bone is constructed based on the patient's bone density test value and the bone microstructure characteristics obtained through medical imaging to simulate the stress distribution, strain state and structural stability of bone under different load conditions. Step S4. Input the influence weights of the different parameters obtained in step S2 into the three-dimensional skeletal mechanical model constructed in step S3, correct the mechanical parameters in the model, and obtain the skeletal mechanical simulation results that integrate the patient's dynamic parameters. Step S5. Based on the simulation results of step S4, and combining the time-varying factors of bone quality with the parameter change trends output by the dynamic prediction model, calculate the probability value of bone fracture in the patient within the future preset time period. Step S6. The probability values ​​calculated in step S5 and the degree of influence of different parameters are visualized using a nomograph visualization model to form a nomograph that includes the range of different parameter values, the corresponding fracture risk probability, and the interaction relationship between parameters.

[0006] Furthermore, in the bone time-varying factors and dynamic prediction model constructed in step S2, the influence weights of different parameters on bone status at time t are calculated using the following formula: in, Let be the influence weight of the i-th parameter at time t; The initial weight coefficient for the i-th parameter; Let be the attenuation coefficient of the i-th parameter; Let be the cumulative influence coefficient of the i-th parameter; The cumulative rate coefficient of the i-th parameter; Let be the interaction coefficient between the i-th parameter and the k-th parameter; Let be the standardized value of the k-th parameter at time t; n is the total number of parameters.

[0007] Furthermore, when constructing the three-dimensional mechanical model of the skeleton in step S3, the elastic modulus of the skeleton... Calculated using the following formula: in, BMD is the elastic modulus corresponding to standard bone mineral density; BMD is the patient's bone mineral density measurement value. Standard bone mineral density value; is the bone mineral density influence coefficient; m is the number of skeletal microstructural characteristic parameters; The influence coefficient of the j-th microstructure characteristic parameter; The standardized value of the j-th microstructural feature parameter is given. The microstructural feature parameter includes trabecular thickness, trabecular spacing, number of trabeculae, and cortical thickness.

[0008] Furthermore, in step S4, when correcting the mechanical parameters of the three-dimensional mechanical model of the skeleton, the corrected yield strength of the skeleton... Calculated using the following formula: in, is the initial yield strength; p is the number of parameters involved in the correction. Let be the influence weight of the i-th parameter at time t; This is the correction factor for the yield strength of the i-th parameter; This is the interaction correction coefficient between the i-th parameter and the change in bone mineral density; Let be the change in bone density at time t under the influence of the i-th parameter.

[0009] Furthermore, in step S5, when calculating the probability value of a patient's bones fracturing within a preset future time period, the probability value P is calculated using the following formula: in, q represents the constant term; q represents the number of parameters involved in the calculation. Let be the regression coefficient of the i-th parameter; Let be the influence weight of the i-th parameter at time t; This represents the skeletal biomechanical response value corresponding to the i-th parameter; This is the maximum stress influence coefficient; The maximum stress value obtained from the simulation of the three-dimensional mechanical model of the skeleton.

[0010] Furthermore, in step S6, when constructing the Nomograph visualization model, the risk scores corresponding to different parameters are calculated using the following formula: in, The risk score for the i-th parameter; The conversion coefficient for the score of the i-th parameter; This represents the fracture probability when the i-th parameter acts alone. This is the baseline constant for the score.

[0011] Furthermore, step S3 includes the following sub-steps: S31. The acquired medical images of the patient's bones are segmented to extract the contour and internal structural features of the bone region, determine the boundary between the bone cortex and the trabeculae, distinguish different anatomical parts of the bone, and mark possible bone defects or lesion areas. S32. Based on the segmented image data, a three-dimensional reconstruction algorithm is used to construct a three-dimensional mesh model of the skeleton. The size and accuracy of the mesh unit are set to ensure that the model can accurately reflect the geometric shape of the skeleton. The surface of the model is smoothed to reduce the error caused by image noise. S33. Based on the anatomical features and material properties of the skeleton, assign corresponding mechanical parameters to the three-dimensional mesh model, including the elastic modulus, Poisson's ratio, and shear modulus of the cortex and trabeculae, set the material property differences of different bone regions, and simulate the anisotropic characteristics of the skeleton. S34. Set constraints and load parameters in the three-dimensional mechanical model, constrain the fixed ends of the bones to simulate the skeletal support under physiological conditions, apply loads related to daily activities, including the magnitude and direction of loads when walking and bending over, and determine the point of application and distribution of the loads.

[0012] Furthermore, step S4 includes the following sub-steps: S41. Arrange the influence weights of different parameters obtained in step S2 at different time points according to the time series, establish a dynamic matrix of weight changes with time, and the elements in the matrix correspond to the weight values ​​of different parameters at the corresponding time points. Determine the change law of parameter weights through matrix operations. S42. Based on the weight values ​​in the dynamic matrix, adjust the mechanical properties related to the corresponding parameters in the three-dimensional skeletal mechanical model. For parameters related to bone density, correct the density parameters of the model according to the weight ratio. For parameters related to bone microstructure, adjust the connection mode and distribution density of trabeculae in the model. S43. Perform finite element analysis on the modified three-dimensional mechanical model of the skeleton, calculate the displacement field, stress field and strain field of the model under different load conditions, record the location and stress value of the stress concentration area in the model, and analyze the changes in the mechanical response of the model. S44. Compare the mechanical simulation results of the model before and after correction, evaluate the influence of parameter weights on the model output. If the difference exceeds the preset threshold, readjust the mapping relationship between parameter weights and mechanical parameters until the model output results are stable.

[0013] Furthermore, step S5 includes the following sub-steps: S51. Extract calibrated mechanical indices from the skeletal biomechanical simulation results of the fused patient dynamic parameters in step S4, including maximum stress value, maximum strain value, stiffness coefficient of bone structure, and toughness index. Compare these indices with preset thresholds to preliminarily determine the load-bearing capacity of the bone. S52. Combining the time-varying factors of bone quality with the change curves of different parameters in the future preset time period output by the dynamic prediction model, predict the value range of different parameters in the time period, calculate the parameter change rate and cumulative change, and determine the overall trend of the influence of parameter changes on bone condition. S53. Establish a correlation model between mechanical indicators and parameter change trends, take parameter change trends as input variables and mechanical indicators as output variables, and use this model to predict the changes in bone mechanical indicators within a preset time period in the future, and obtain the dynamic prediction curve of mechanical indicators. S54. Based on the dynamic prediction curve of the biomechanical index and referring to the critical value of the biomechanical index when fractures occur in historical case data, calculate the number of times and the duration of the patient's bone biomechanical index exceeding the critical value within a preset time period in the future, and thus obtain the probability value of fracture risk.

[0014] Predictive systems for fracture risk in patients with osteoporosis include: The multi-dimensional parameter acquisition and storage unit is used to collect and store different parameters of osteoporosis patients. This unit is connected to medical testing equipment to receive test data and classify and store it. The bone time-varying model construction and analysis unit is connected to the parameter multi-dimensional acquisition and storage unit. It calls the stored parameter data to construct bone time-varying factors and dynamic prediction models, and analyzes the time change trend and influence weight of the parameters. The skeletal biomechanics simulation unit is connected to the multi-dimensional parameter acquisition and storage unit and the bone time-varying model construction and analysis unit, respectively. It receives bone density and bone microstructure data to construct a three-dimensional skeletal biomechanics model and uses parameter influence weights to correct the model's biomechanical parameters. The fracture risk probability calculation unit is connected to the skeletal biomechanics simulation unit and the bone time-varying model construction and analysis unit. Based on the corrected skeletal biomechanics simulation results and parameter change trends, it calculates the fracture risk probability value within a future preset time period. The nomograph visualization unit is connected to the fracture risk probability calculation unit and the multi-dimensional parameter acquisition and storage unit. It receives fracture risk probability values ​​and different parameter data, and generates a nomograph including the degree of parameter influence through the nomograph visualization model. The system control and coordination unit is connected to the multi-dimensional parameter acquisition and storage unit, the bone time-varying model construction and analysis unit, the skeletal biomechanics simulation unit, the fracture risk probability calculation unit, and the nomograph visualization unit, respectively, to control the operation process of different units and coordinate the data transmission and interaction between different units.

[0015] Beneficial Effects: This invention proposes a method and system for predicting fracture risk in osteoporosis patients. By collecting multiple patient parameters and combining them with time-varying bone factors and a dynamic prediction model, it fully considers the dynamic changes and interactions of various parameters over time. This allows for accurate capture of risk changes during disease progression, overcoming the problem of existing technologies neglecting the time-varying characteristics of bone, and enabling the prediction results to reflect long-term risk trends. Simultaneously, by constructing a three-dimensional mechanical model using mechanical simulation of bone microstructure and correcting the mechanical parameters with patient parameters, it can accurately simulate the stress-strain state of bone under different loads, providing sufficient biomechanical basis for fracture risk assessment and solving the deficiency of inaccurate simulation of bone mechanical properties in existing technologies. Furthermore, the results are presented through a nomograph visualization model, clearly showing the degree of influence of each parameter, facilitating clinicians to develop personalized prevention and treatment plans, effectively reducing the fracture incidence rate and alleviating the medical burden on families and society. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method steps of the present invention; Figure 2 This is a diagram showing the system unit composition of the present invention. Detailed Implementation

[0017] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] like Figure 1 As shown, the method for predicting fracture risk in patients with osteoporosis includes the following steps: Step S1. Collect different parameters of osteoporosis patients, including patient age, gender, body mass index, bone mineral density test value, history of fracture, smoking status, frequency of drinking, type and duration of taking drugs that affect bone metabolism, serum calcium content, serum phosphorus content, vitamin D level, parathyroid hormone level, and bone turnover marker level. Specifically, the core of step S1 lies in the comprehensive and accurate collection of various parameters from osteoporosis patients. These parameters form the basis for all subsequent analyses and predictions, and their completeness and accuracy directly affect the reliability of the entire prediction method. The collected parameters cover multiple dimensions, including the patient's basic physiological information, lifestyle habits, and clinical test indicators. These parameters together constitute a multi-factor system for assessing the patient's bone status and fracture risk. Age is an important basic parameter. With increasing age, the body's bone metabolism gradually becomes unbalanced, bone loss accelerates, and the risk of fracture increases significantly. Gender factors cannot be ignored either; after menopause, due to the decline in estrogen levels, women have a much higher incidence of osteoporosis and fractures than men. Body mass index (BMI) is related to the load borne by bones. Too low a BMI will result in insufficient mechanical stimulation of the bones, which is not conducive to maintaining bone density.

[0019] In the specific implementation process, the patient's basic information, including fixed parameters such as age and gender, is first retrieved through the patient's electronic medical record system. Body mass index (BMI) is calculated by measuring the patient's height and weight, with height measured to the nearest 0.1 cm and weight to the nearest 0.1 kg. The calculation formula is weight (kg) divided by the square of height (m). Bone mineral density (BMD) is measured using a dual-energy X-ray absorptiometry (DXA) system, primarily targeting fracture-prone areas such as the lumbar spine and hip. Results are expressed in grams per square centimeter, accurate to three decimal places. Past fracture history is determined through patient interviews and review of medical records, including the time, location, and cause of fractures. Smoking status is recorded, including the number of years the patient has smoked and the number of cigarettes smoked daily. Alcohol consumption frequency is categorized as non-drinking, occasional drinking (less than 4 times per month), and frequent drinking (4 times or more per month). The type and duration of medications affecting bone metabolism, such as glucocorticoids, are recorded, including the drug name, daily dosage, and duration of continuous use. Serum calcium and phosphorus levels are obtained through blood biochemistry tests, with normal reference ranges of 2.25-2.58 mmol / L and 0.96-1.62 mmol / L, respectively. Vitamin D levels are measured in nanograms per milliliter (Ng / mL) through blood tests, with a level below 20 Ng / mL indicating vitamin D deficiency. Parathyroid hormone levels are also measured in blood tests, with a normal range of 15-65 picograms per milliliter. Bone turnover markers such as type I collagen carboxy-terminal peptide (CTTP) and osteocalcin are also measured in blood or urine.

[0020] Step S2. Construct a time-varying factor and dynamic prediction model for bone quality based on the parameters. Analyze the changing trend of the parameters over time using this model to determine the influence weight of different parameters on bone quality at different time points. Specifically, step S2 aims to construct a time-varying factor and dynamic prediction model for bone quality based on the collected parameters. This model's function is to deeply analyze the changing trends of various parameters over time and quantify the influence weight of each parameter on bone condition at different time points. Bone condition is a dynamic process, influenced by the combined effects of multiple parameters, and the degree of influence of each parameter changes over time. For example, age has a cumulative effect on bone quality, and its weight may gradually increase with time; while the effect of certain medications on bone quality may be more significant in the early stages of medication, and its influence weight may decrease as the duration of medication increases. By constructing this model, static parameter data can be transformed into dynamic influence weights, providing dynamic parameter basis for subsequent skeletal biomechanical simulations and fracture risk calculations.

[0021] In practice, the collected parameters are first processed as time series data, gathering parameter data from different time points in the patient's past. If some data is missing, it can be supplemented using appropriate interpolation methods to ensure that each parameter has data records from at least three different time points. Then, a multivariate time series analysis method is used, with each parameter as an independent variable and bone mineral density change as the dependent variable, to construct a regression model. In the model, a time variable is introduced as a moderating factor to analyze the interaction between each parameter and time, thereby determining the influence weight of each parameter at different time points. For example, for the age parameter, by analyzing the rate of change in bone mineral density in patients of different age groups, the change in the influence weight of each year-old increase in bone quality is determined; for the vitamin D level parameter, based on the correlation between vitamin D levels and bone mineral density at different time points, the influence weight of each 1 ng / mL change in vitamin D level on bone quality at different time points is calculated. Finally, the constructed model is validated using cross-validation, with a subset of patient data as the validation set to assess the accuracy of the model in predicting the influence weights of each parameter. If the model error exceeds a preset threshold, the model structure and parameters are adjusted until the model reaches the expected accuracy requirements.

[0022] Step S3. Using the mechanical simulation method of bone microstructure, a three-dimensional mechanical model of bone is constructed based on the patient's bone density test value and the bone microstructure characteristics obtained through medical imaging to simulate the stress distribution, strain state and structural stability of bone under different load conditions. Specifically, the key to step S3 is to utilize mechanical simulation methods of bone microstructure to construct a three-dimensional mechanical model that reflects the true mechanical properties of the patient's bones. This model is an important tool for assessing the stress state and structural stability of bones under different loads. The mechanical properties of bones are not only related to bone density but also closely related to their microstructure, such as the thickness, spacing, and number of trabeculae and the thickness of the cortex. These microstructural features directly affect the load-bearing capacity and fracture resistance of bones. By constructing a three-dimensional mechanical model, the loads on bones from different daily activities (such as walking, bending, and weight-bearing) can be simulated, the stress distribution and strain state within the bone can be analyzed, and areas of stress concentration can be identified. These areas are often high-risk sites for fractures, providing a mechanical basis for subsequent fracture risk assessment.

[0023] In the specific implementation process, the patient's skeletal medical imaging data is first acquired, typically using high-resolution computed tomography (HRCT) or magnetic resonance imaging (MRI) techniques. The scanning range covers the skeletal areas to be evaluated (such as the lumbar spine and femoral neck), with the slice thickness controlled between 0.5 and 1 mm to ensure clear visualization of the bone's microstructural features. Next, the image data undergoes preprocessing, including image noise reduction and grayscale correction, to improve image quality. Then, image segmentation algorithms are used to extract the bone contours from the images, distinguishing between the cortical and trabecular bone regions, and measuring microstructural parameters such as the thickness, spacing, and number of trabecular bone, as well as the thickness of the cortical bone. The measurement accuracy of trabecular bone thickness reaches the micrometer level. Based on the segmented image data, a 3D mesh model of the bone is constructed using 3D reconstruction software. The size of the mesh cells is set according to the size and accuracy requirements of the bone region, generally between 0.1 and 1 mm, ensuring that the model accurately reflects the geometric morphology of the bone. Next, material properties were assigned to the model. Based on bone density measurements and microstructural parameters, mechanical parameters such as the elastic modulus and Poisson's ratio of the cortical and trabecular bone were determined. For example, the elastic modulus of the cortical bone is typically between 10 and 20 GPa, while the elastic modulus of the trabecular bone is relatively lower, between 0.5 and 2 GPa. Finally, boundary conditions and load parameters were set for the model to simulate different physiological loads. For instance, the load on the hip joint during walking is approximately 2-3 times the body weight, and the load on the lumbar spine during bending is approximately 5-6 times the body weight. Finite element analysis software was used to calculate the stress distribution, strain state, and structural stability of the model under different loads, outputting key mechanical indicators such as maximum stress and maximum strain.

[0024] Step S4. Input the influence weights of the different parameters obtained in step S2 into the three-dimensional skeletal mechanical model constructed in step S3, correct the mechanical parameters in the model, and obtain the skeletal mechanical simulation results that integrate the patient's dynamic parameters. Specifically, the main task of step S4 is to integrate the influence weights of the parameters obtained in step S2 into the three-dimensional skeletal mechanical model constructed in step S3, thereby correcting the mechanical parameters in the model and obtaining skeletal mechanical simulation results that can reflect the dynamic changes in the patient's parameters. Since the patient's various parameters are constantly changing, these changes affect bone quality and thus alter the mechanical properties of the bone. Therefore, a mechanical model constructed solely based on initial parameters cannot accurately reflect the patient's current and future skeletal mechanical state. By introducing the influence weights of the parameters into the model correction process, the model can dynamically respond to parameter changes, improving the timeliness and accuracy of the mechanical simulation results and providing more reliable mechanical data for subsequent fracture risk calculations.

[0025] In practice, the mapping relationship between the influence weights of each parameter in step S2 and the skeletal mechanical parameters is first clarified. For example, the influence weight of bone mineral density is positively correlated with the elastic modulus of bone; that is, the greater the influence weight of bone mineral density, the higher the corresponding correction value of the elastic modulus of bone. The influence weight of vitamin D level is related to the toughness parameter of bone; as the influence weight of vitamin D level increases, the correction value of the toughness parameter of bone will also increase accordingly. Then, based on these mapping relationships, the influence weights of each parameter at the current time point are converted into corresponding mechanical parameter correction coefficients. For each mechanical parameter (such as elastic modulus, Poisson's ratio, yield strength, etc.), the correction amount is calculated according to its corresponding parameter influence weight. The calculation formula for the correction amount is determined based on the correlation between the parameter and the mechanical parameter. For example, the correction amount of elastic modulus is equal to the initial elastic modulus multiplied by the product of the influence weight of bone mineral density and a specific coefficient. The calculated correction amount is applied to the three-dimensional mechanical model to update the mechanical parameters of the model. Afterward, the corrected model is recalculated to simulate the stress distribution and strain state of bone under the same load conditions, and the results are compared with those before correction to analyze the differences before and after correction and evaluate the degree of influence of parameter influence weights on the model results. If the difference does not meet expectations, the mapping relationship or the calculation method of the correction amount should be readjusted until the model results can accurately reflect the impact of parameter changes on bone mechanical properties.

[0026] Step S5. Based on the simulation results of step S4, and combining the time-varying factors of bone quality with the parameter change trends output by the dynamic prediction model, calculate the probability value of bone fracture in the patient within the future preset time period. Specifically, the core of step S5 is to calculate the probability of a patient's bone fracture within a predetermined time period based on the corrected bone biomechanical simulation results and the parameter change trends of time-varying bone factors and dynamic prediction models. This step is a crucial link connecting biomechanical analysis and risk prediction. By comprehensively considering the current mechanical state of the bone and the impact of future parameter changes on bone quality, it is possible to more scientifically assess the fracture risk faced by the patient in the future. The predetermined time period can be set according to clinical needs, such as 1 year, 3 years, 5 years, etc. Different time periods correspond to different risk assessment needs. Short-term risk assessment helps to formulate short-term intervention measures, while long-term risk assessment can provide a reference for the patient's long-term health management.

[0027] In the specific implementation process, the first step is to determine the preset time period, typically 1 year, 3 years, or 5 years as the evaluation cycle. Then, based on the parameter change trends output by the bone time-varying factors and dynamic prediction model in step S2, the values ​​of various parameters within the preset time period are predicted. Parameter prediction employs time-series prediction methods, establishing a prediction model for each parameter based on historical parameter data and trends, such as linear regression or exponential smoothing models, to predict the parameter value at each future time point. The predicted parameter values ​​are input into the bone time-varying factors and dynamic prediction model to calculate the changes in the influence weights of each parameter within the preset time period. Simultaneously, based on the corrected three-dimensional skeletal mechanical model from step S4, the maximum stress and maximum strain of the current bone under different loads are obtained. Combining the changes in the influence weights of future parameters, the changing trends of skeletal mechanical indicators within the preset time period are predicted. For example, based on the predicted value of bone mineral density and its influence weight, the changes in the future bone elastic modulus are calculated, thereby predicting the changes in the maximum stress. Referring to a large amount of clinical case data, critical values ​​for mechanical indicators at different bone sites are determined; for example, the probability of fracture increases significantly when the maximum stress exceeds a certain value. The predicted future biomechanical parameters are compared with critical values ​​to calculate the number of times, duration, and magnitude of exceedances within a pre-defined future time period. Based on these calculations, a fracture risk probability calculation model is constructed using statistical methods, such as logistic regression. This model takes parameters related to biomechanical parameters exceeding the critical value as input and outputs the fracture probability value within the pre-defined future time period. For example, the probability of fracture can be calculated using logistic regression by using the number of times the maximum stress exceeds the critical value within the next year and the duration of each exceedance as independent variables. Finally, the calculated probability values ​​are validated by comparing them with actual fracture case data, and the model parameters are adjusted to ensure the accuracy of the probability values.

[0028] Step S6. The probability values ​​calculated in step S5 and the degree of influence of different parameters are visualized using a nomograph visualization model to form a nomograph that includes the range of different parameter values, the corresponding fracture risk probability, and the interaction relationship between parameters.

[0029] Specifically, the main function of step S6 is to use a nomograph visualization model to visualize the fracture risk probability values ​​and the influence of each parameter, forming an intuitive and easy-to-understand nomograph. As an effective visualization tool, the nomograph can present complex multi-factor analysis results graphically, clearly showing the value range of each parameter, the corresponding fracture risk probability, and the interaction relationships between parameters, facilitating clinicians' quick understanding and application of the prediction results. Compared with traditional numerical tables or textual descriptions, nomographs have advantages such as strong intuitiveness, large information content, and ease of interpretation, helping doctors better communicate with patients, enabling patients to understand their fracture risk status and influencing factors, and improving patient adherence to treatment and preventive measures.

[0030] In the specific implementation process, the fracture risk probability values ​​and the influence data of each parameter calculated in step S5 are first organized, including the value range of each parameter, the corresponding fracture risk probability under different values, and the interaction coefficient between parameters. Then, the coordinate axes and scale of the nomograph are determined. Typically, the horizontal axis represents each parameter, with each parameter occupying a certain length, determined by the degree of influence of the parameter; the greater the influence, the longer the corresponding length. The vertical axis represents the fracture risk probability, with a scale from 0 to 100%, and intervals set according to accuracy requirements, such as every 5% or 10%. Based on the parameter value range and the corresponding fracture risk probability, the scale line and corresponding risk probability point for each parameter are plotted on the nomograph. For example, for the age parameter, the value range of 20-80 years corresponds to different risk probability points; connecting these points forms the risk curve for the age parameter. The interaction relationship between parameters is represented by adding connecting lines or annotations to the nomograph. For example, when both body mass index and bone density are low, the fracture risk probability will significantly increase; this synergistic effect can be shown through special marker lines. After the nomograph is completed, it is validated by inviting clinicians to evaluate its clarity, accuracy, and usability. Adjustments and optimizations are made based on feedback to ensure the nomograph accurately conveys predictive information and meets clinical application needs. Finally, the completed nomograph is exported to commonly used image formats such as PNG and JPEG for easy storage and viewing in electronic medical record systems.

[0031] Preferably, in the bone time-varying factors and dynamic prediction model constructed in step S2, the influence weights of different parameters on bone status at time t are calculated using the following formula: in, Let be the influence weight of the i-th parameter at time t; The initial weight coefficient for the i-th parameter; Let be the attenuation coefficient of the i-th parameter; Let be the cumulative influence coefficient of the i-th parameter; The cumulative rate coefficient of the i-th parameter; Let be the interaction coefficient between the i-th parameter and the k-th parameter; Let be the standardized value of the k-th parameter at time t; n is the total number of parameters.

[0032] Specifically, this paper describes the calculation method of the influence weights of each parameter on bone status at time t in the time-varying factors and dynamic prediction model of bone quality constructed in step S2. By introducing multiple coefficients and parameter interaction terms, the influence of different parameters on bone status at different time points is quantified more accurately, thereby improving the model's ability to capture dynamic changes. The initial weight coefficient reflects the basic influence of the parameter on bone quality in the initial state. The initial weights of different parameters differ; for example, the initial weight coefficient of bone mineral density is usually higher than that of lifestyle parameters such as smoking status. The decay coefficient reflects the rate at which the parameter's influence weakens over time. For example, some short-term medications have a large decay coefficient, and their influence weight decreases rapidly over time. The cumulative influence coefficient and cumulative rate coefficient describe the characteristics of the parameter's influence accumulating over time. For example, the cumulative influence coefficient of age is large, and the cumulative rate coefficient is relatively stable, causing its influence weight to gradually increase over time. The interaction influence coefficient measures the interaction between different parameters. For example, there is a certain interaction between body mass index (BMI) and bone mineral density (BMD). When BMI is low, the influence weight of BMD on bone status will increase accordingly. In the implementation process, the first step is to determine the value range of each coefficient based on a large amount of clinical data. The initial weight coefficient is generally between 0.1 and 0.8, the attenuation coefficient and cumulative rate coefficient are mostly between 0 and 0.5, and the interaction influence coefficient is between -0.3 and 0.3 depending on the correlation strength between the parameters. Then, standardized parameter values ​​are collected from patients at different time points, and these are substituted into the formula to calculate the influence weight of each parameter at the corresponding time point. During the calculation, the accuracy of the standardized parameter values ​​must be ensured. The Z-score standardization method is typically used to convert parameter values ​​into standardized data with a mean of 0 and a standard deviation of 1 to eliminate the influence of different parameter dimensions. Finally, the calculated influence weights are compared and verified with actual changes in bone quality. The model is continuously optimized by adjusting the coefficient values ​​until the influence weights calculated by the model accurately reflect the dynamic impact of the parameters on bone quality.

[0033] Preferably, when constructing the three-dimensional mechanical model of the skeleton in step S3, the elastic modulus of the skeleton is... Calculated using the following formula: in, BMD is the elastic modulus corresponding to standard bone mineral density; BMD is the patient's bone mineral density measurement value. Standard bone mineral density value; is the bone mineral density influence coefficient; m is the number of skeletal microstructural characteristic parameters; The influence coefficient of the j-th microstructure characteristic parameter; The standardized value of the j-th microstructural feature parameter is given. The microstructural feature parameter includes trabecular thickness, trabecular spacing, number of trabeculae, and cortical thickness.

[0034] Specifically, the calculation of the elastic modulus of bone in step S3, when constructing the three-dimensional mechanical model of the skeleton, is described. The purpose is to more accurately determine the elastic modulus of bone by comprehensively considering bone density and bone microstructure characteristics, providing a reliable material property basis for subsequent mechanical simulations. The elastic modulus corresponding to standard bone density is a benchmark value, usually determined based on test data from a large number of healthy individuals. The standard elastic modulus varies for different bone locations; for example, the standard elastic modulus for the lumbar spine is approximately 12 GPa, while that for the hip is approximately 15 GPa. The bone density influence coefficient is used to quantify the relationship between bone density and elastic modulus. The value of this coefficient is related to the bone location, generally between 1.5 and 3.0. For every 0.1 g / cm² increase in bone density, the elastic modulus increases accordingly. The number of microstructure characteristics is determined based on the complexity of the bone microstructure. For key locations such as the lumbar spine and hip, 4-6 microstructure characteristics are typically selected, including trabecular thickness, trabecular spacing, number of trabeculae, cortical thickness, and trabecular connectivity density. The influence coefficients of each microstructural feature parameter are determined based on their importance to bone biomechanical properties. The influence coefficient of cortical bone thickness is relatively large, generally between 0.05 and 0.2, while the influence coefficient of trabecular spacing is relatively small, mostly between 0.01 and 0.08. The standardized values ​​of the microstructural feature parameters are also processed using the Z-score method, placing them within the range of -2 to 2 to facilitate comparison and calculation between different parameters. In the implementation process, the patient's bone mineral density (BMD) values ​​are first obtained using dual-energy X-ray absorptiometry (DXA) and compared with standard BMD values ​​to calculate the ratio. Simultaneously, high-resolution medical images are used to extract bone microstructural feature parameters, which are then standardized. Then, these values ​​are substituted into the formula to calculate the elastic modulus of the bone. During the calculation, attention must be paid to the parameter differences in different bone locations; the values ​​of the standard elastic modulus, bone mineral density influence coefficient, and microstructural feature parameter influence coefficient are determined separately for different locations such as the lumbar spine and hip. After the calculation is completed, the obtained elastic modulus is compared with the elastic modulus of the bone sample measured by mechanical experiments. If the error exceeds 5%, the values ​​of each influence coefficient are adjusted and the calculation is repeated until the error is within an acceptable range, so as to ensure that the calculation accuracy of the elastic modulus meets the requirements of mechanical simulation.

[0035] Preferably, when correcting the mechanical parameters of the three-dimensional mechanical model of the skeleton in step S4, the corrected bone yield strength Calculated using the following formula: in, is the initial yield strength; p is the number of parameters involved in the correction. Let be the influence weight of the i-th parameter at time t; This is the correction factor for the yield strength of the i-th parameter; This is the interaction correction coefficient between the i-th parameter and the change in bone mineral density; Let be the change in bone density at time t under the influence of the i-th parameter.

[0036] Specifically, the description of the calculated bone yield strength in step S4, which modifies the mechanical parameters of the three-dimensional skeletal biomechanical model, focuses on combining the influence weights of the parameters obtained in step S2 with changes in bone mineral density (BMD) to dynamically correct the bone yield strength. This makes the corrected yield strength more reflective of the patient's current skeletal biomechanical properties. The initial yield strength is a baseline value determined based on standard bone samples. The initial yield strength varies for different bone locations; for example, the initial yield strength of the femoral neck is approximately 120 MPa, and that of the lumbar spine is approximately 100 MPa. The number of parameters involved in the correction is determined based on their significance in influencing the yield strength, typically selecting 5-8 key parameters such as BMD, vitamin D level, and parathyroid hormone level. The correction coefficient for each parameter is determined based on the correlation between the parameter and the yield strength. The correction coefficient for BMD is relatively large, generally between 0.1 and 0.5, while the correction coefficient for vitamin D level is relatively small, mostly between 0.05 and 0.2. The cross-correction coefficient between parameters and changes in bone mineral density (BMD) describes the degree to which a parameter's influence changes with BMD. For example, when BMD decreases, the cross-correction coefficient for certain drug parameters increases accordingly to reflect their protective effect on yield strength. This coefficient typically ranges from 0 to 0.3. The change in BMD refers to the change in a patient's BMD over a specific time period, calculated as the difference between two consecutive BMD measurements, accurate to three decimal places (g / cm²). In practice, the influence weight of each parameter at the current time point is first obtained, calculated using a formula. Then, recent BMD data are collected, the change in BMD is calculated, and standardized. Next, the initial yield strength, the influence weights of each parameter, the correction coefficient, the cross-correction coefficient, and the change in BMD are substituted into the formula to calculate the corrected bone yield strength. During the calculation, it is crucial to ensure the accuracy of each parameter's value. For negative changes in BMD (i.e., decreased BMD), special attention must be paid to the cross-correction coefficient to accurately reflect the parameter's corrective effect on yield strength when BMD decreases. Finally, the corrected yield strength was applied to the three-dimensional mechanical model of the skeleton for mechanical simulation. The simulation results were compared with actual bone mechanical test data. The calculation of yield strength was optimized by adjusting the correction coefficient and the interactive correction coefficient, so that the mechanical response of the model is more in line with the actual situation.

[0037] Preferably, in step S5, when calculating the probability value of a patient's bone fracture within a preset future time period, the probability value P is calculated using the following formula: in, q represents the constant term; q represents the number of parameters involved in the calculation. Let be the regression coefficient of the i-th parameter; Let be the influence weight of the i-th parameter at time t; This represents the skeletal biomechanical response value corresponding to the i-th parameter; This is the maximum stress influence coefficient; The maximum stress value obtained from the simulation of the three-dimensional mechanical model of the skeleton.

[0038] Specifically, step S5 describes the method for calculating the probability of a patient's bones fracturing within a preset time period. This is achieved by introducing a constant term, regression coefficients, parameter influence weights, bone biomechanical response values, and maximum stress values ​​to construct a logistic regression-based probability calculation model, thus quantifying fracture risk more scientifically. The constant term is a baseline value fitted based on a large amount of historical case data; its value is related to the fracture location and population characteristics, generally ranging from -3 to -1. The number of parameters involved in the calculation is determined by their importance to fracture risk, typically including 6-10 key parameters such as bone mineral density, age, and previous fracture history. The regression coefficients of each parameter reflect the strength of its influence on the fracture probability. The regression coefficient for a previous fracture history is relatively large, generally between 0.5 and 1.5, while the regression coefficients for lifestyle parameters such as smoking status are relatively small, mostly between 0.1 and 0.5. The corresponding bone biomechanical response value refers to the specific response exhibited by the bone in the biomechanical simulation under the influence of that parameter, such as strain and displacement values, output by the biomechanical simulation software, accurate to three decimal places. The maximum stress influence coefficient measures the contribution of maximum stress value to fracture probability. Its value ranges from 0.3 to 0.8 depending on the bone location; the higher the maximum stress value, the more significant the coefficient's effect on fracture probability. In implementation, the constant term and regression coefficients for each parameter are first determined by collecting data from a large number of fracture and non-fracture patients and fitting the data using Logistic regression analysis. Then, the influence weights of each parameter and the corresponding skeletal biomechanical response values, as well as the maximum stress value simulated by a three-dimensional skeletal biomechanical model, are obtained. These data are substituted into the formula to calculate the fracture probability value within the future preset time period. During the calculation, attention must be paid to the parameter value range to ensure that the skeletal biomechanical response values ​​and maximum stress values ​​are standardized to eliminate the influence of dimensions. For different preset time periods (e.g., 1 year, 3 years, 5 years), the corresponding constant term and regression coefficients need to be determined separately to adapt to the changing patterns of fracture risk over different time spans. Finally, the calculated fracture probability value is compared and verified with actual fracture occurrences. By adjusting the constant term and regression coefficients, the model is optimized so that the probability value accurately reflects the patient's actual fracture risk, providing a reliable basis for clinical decision-making.

[0039] Preferably, in step S6, when constructing the Nomograph visualization model, the risk scores corresponding to different parameters are calculated using the following formula: in, The risk score for the i-th parameter; The conversion coefficient for the score of the i-th parameter; This represents the fracture probability when the i-th parameter acts alone. This is the baseline constant for the score.

[0040] Specifically, the description of the risk score calculation method for each parameter during the construction of the nomograph visualization model in step S6 aims to convert the fracture probability when a parameter acts alone into an intuitive risk score, so as to clearly show the degree of influence of each parameter on fracture risk on the nomograph. The score conversion coefficient is used to convert probability values ​​into scores. Different parameters have different score conversion coefficients; the greater the influence of a parameter, the larger its score conversion coefficient, generally between 5 and 20. For example, the score conversion coefficient for bone mineral density is usually between 15 and 20, while the score conversion coefficient for alcohol consumption frequency is mostly between 5 and 10. The fracture probability when a parameter acts alone refers to the fracture probability corresponding to a specific value of that parameter while other parameters remain unchanged. It is calculated using the controlled variable method combined with a formula, and its value ranges between 0 and 1. The score baseline constant is an adjustment value set to keep the risk score within a reasonable range (e.g., 0-100 points). Its value is determined according to the number of parameters and the score conversion coefficient, generally between 20 and 50. In the implementation process, the fracture probability of each parameter acting alone under different values ​​is first calculated using the controlled variable method to ensure that other parameters remain at an average level during the calculation. Then, the score conversion coefficient and score baseline constant are determined according to the importance of the parameters, with larger conversion coefficients assigned to key parameters. Next, the fracture probability of each parameter acting alone is substituted into the formula to calculate the corresponding risk score. During the calculation, attention must be paid to the logarithmic transformation of the probability values. When the probability value is close to 0 or 1, an appropriate smoothing method must be used to avoid abnormal fluctuations in the score. Finally, the calculated risk scores are applied to the nomograph plot to ensure that the risk score of each parameter accurately corresponds to the corresponding fracture probability. By adjusting the score conversion coefficient and score baseline constant, the score distribution on the nomograph is made reasonable and easy to interpret, allowing clinicians to quickly determine the corresponding risk score based on the patient's parameter values, and thus assess the overall fracture risk.

[0041] Preferably, step S3 includes the following sub-steps: S31. The acquired medical images of the patient's bones are segmented to extract the contour and internal structural features of the bone region, determine the boundary between the bone cortex and the trabeculae, distinguish different anatomical parts of the bone, and mark possible bone defects or lesion areas. S32. Based on the segmented image data, a three-dimensional reconstruction algorithm is used to construct a three-dimensional mesh model of the skeleton. The size and accuracy of the mesh unit are set to ensure that the model can accurately reflect the geometric shape of the skeleton. The surface of the model is smoothed to reduce the error caused by image noise. S33. Based on the anatomical features and material properties of the skeleton, assign corresponding mechanical parameters to the three-dimensional mesh model, including the elastic modulus, Poisson's ratio, and shear modulus of the cortex and trabeculae, set the material property differences of different bone regions, and simulate the anisotropic characteristics of the skeleton. S34. Set constraints and load parameters in the three-dimensional mechanical model, constrain the fixed ends of the bones to simulate the skeletal support under physiological conditions, apply loads related to daily activities, including the magnitude and direction of loads when walking and bending over, and determine the point of application and distribution of the loads.

[0042] Specifically, step S3, the construction of the 3D skeletal mechanical model, is a series of steps. Its core significance lies in ensuring, through a standardized process, that the model accurately reflects the geometric morphology and mechanical properties of the patient's bones, providing a reliable foundation for subsequent mechanical simulations. In step S31, the segmentation of medical images requires high-precision image segmentation algorithms, such as threshold segmentation combined with region growing, to ensure the accuracy of skeletal region contour extraction. The distinction between the cortical and trabecular bone boundaries needs to reach a precision of 0.1 mm. The marking of bone defects or lesions must be combined with clinical diagnostic information to ensure that no structural abnormalities that might affect skeletal mechanical properties are overlooked. In the 3D reconstruction process of S32, the size of the mesh cells needs to be determined according to the size and complexity of the skeletal region. For structurally complex regions such as the hip, the mesh cell size is typically 0.2-0.5 mm, while for relatively simple regions such as the lumbar vertebrae, it can be set to 0.5-1 mm. Gaussian filtering is used for model surface smoothing, with the filtering radius controlled within the range of 0.5-1 pixels to remove noise while preserving key structural features. When assigning mechanical parameters in S33, the elastic modulus of the cortical bone is set at 15-20 GPa and the Poisson's ratio at 0.2-0.3, while the elastic modulus of the trabecular bone is set at 1-2 GPa and the Poisson's ratio at 0.2-0.25. Differences in material properties across different bone regions are determined by consulting anatomical databases and considering the patient's specific condition to reflect the anisotropic characteristics of the bone. In the constraint and load parameter settings of S34, the fixed-end constraints must simulate the actual connection state of the bone within the body. For example, the constraint on the proximal femur must consider the hip joint's motion characteristics. The load magnitude is calculated based on the patient's weight; the load during walking is typically 2-3 times the body weight, and the load on the lumbar spine during bending is 5-6 times the body weight. The load direction must conform to the principles of human biomechanics, and the point of application should be selected from the main stress-bearing areas of the bone, such as the lateral edge of the femoral neck. The entire implementation process requires multiple verifications, adjusting the model parameters by comparing with bone samples with known mechanical properties until the simulation result error is within 5%.

[0043] Preferably, step S4 includes the following sub-steps: S41. Arrange the influence weights of different parameters obtained in step S2 at different time points according to the time series, establish a dynamic matrix of weight changes with time, and the elements in the matrix correspond to the weight values ​​of different parameters at the corresponding time points. Determine the change law of parameter weights through matrix operations. S42. Based on the weight values ​​in the dynamic matrix, adjust the mechanical properties related to the corresponding parameters in the three-dimensional skeletal mechanical model. For parameters related to bone density, correct the density parameters of the model according to the weight ratio. For parameters related to bone microstructure, adjust the connection mode and distribution density of trabeculae in the model. S43. Perform finite element analysis on the modified three-dimensional mechanical model of the skeleton, calculate the displacement field, stress field and strain field of the model under different load conditions, record the location and stress value of the stress concentration area in the model, and analyze the changes in the mechanical response of the model. S44. Compare the mechanical simulation results of the model before and after correction, evaluate the influence of parameter weights on the model output. If the difference exceeds the preset threshold, readjust the mapping relationship between parameter weights and mechanical parameters until the model output results are stable.

[0044] Specifically, step S4, which integrates dynamic parameters of the patient to correct the skeletal biomechanical model, is significant because it accurately reflects the dynamic changes of parameters in the skeletal biomechanical model through a systematic process, improving the model's adaptability to individual patient conditions. In step S41, when establishing the dynamic matrix, the time series interval is determined based on the rate of parameter change. Parameters with slower changes, such as bone mineral density, can use an interval of 3-6 months, while parameters with faster changes, such as bone turnover markers, use an interval of 1-2 months. Matrix operations employ matrix multiplication and eigenvalue decomposition to extract the main trends in parameter weight changes, ensuring the matrix accurately represents the evolution of weights over time. In step S32, when adjusting biomechanical properties, the weight of bone mineral density-related parameters is adjusted to a ratio of 1:1.2 to 1:1.5 to the density parameter. That is, for every 0.1 increase in weight, the density parameter increases by 12%-15%. Adjusting bone microstructure-related parameters requires changing the node connection strength of the trabecular model; for every 0.1 increase in weight, the connection strength increases by 8%-10%. Simultaneously, the distribution density of trabeculae is adjusted to ensure that structural changes are consistent with parameter influences. The finite element analysis in S43 employs an explicit dynamic algorithm with a calculation step size of 1e-6 seconds to capture instantaneous stress-strain changes. Stress concentration regions are identified using a threshold method, marking areas exceeding 70% of the material's yield strength as high-risk regions. Their location coordinates and stress values ​​are recorded. When analyzing changes in mechanical response, the number and distribution of high-risk regions before and after correction are compared. The results in S44 are compared using root mean square error (RMSE) calculation. A preset threshold is typically set to 5%. If the RMS error of the maximum stress value before and after correction exceeds 5%, the mapping relationship needs to be readjusted. This adjustment is achieved through least squares fitting, using actual mechanical test data as a benchmark to optimize the corresponding ratio of weights to mechanical parameters until the model output stabilizes. An error within 3% for three consecutive calculations is considered acceptable. The entire process requires dynamic updates based on the patient's clinical data to ensure the model remains consistent with the patient's current condition.

[0045] Preferably, step S5 includes the following sub-steps: S51. Extract calibrated mechanical indices from the skeletal biomechanical simulation results of the fused patient dynamic parameters in step S4, including maximum stress value, maximum strain value, stiffness coefficient of bone structure, and toughness index. Compare these indices with preset thresholds to preliminarily determine the load-bearing capacity of the bone. S52. Combining the time-varying factors of bone quality with the change curves of different parameters in the future preset time period output by the dynamic prediction model, predict the value range of different parameters in the time period, calculate the parameter change rate and cumulative change, and determine the overall trend of the influence of parameter changes on bone condition. S53. Establish a correlation model between mechanical indicators and parameter change trends, take parameter change trends as input variables and mechanical indicators as output variables, and use this model to predict the changes in bone mechanical indicators within a preset time period in the future, and obtain the dynamic prediction curve of mechanical indicators. S54. Based on the dynamic prediction curve of the biomechanical index and referring to the critical value of the biomechanical index when fractures occur in historical case data, calculate the number of times and the duration of the patient's bone biomechanical index exceeding the critical value within a preset time period in the future, and thus obtain the probability value of fracture risk.

[0046] Specifically, step S5, the calculation of fracture risk probability, is a detailed breakdown of steps. Its core value lies in combining bone biomechanical properties with parameter change trends through multi-stage analysis and calculation, thereby improving the accuracy and reliability of fracture risk prediction. In S51, when extracting key mechanical indicators, the extraction of maximum stress and maximum strain values ​​must cover all areas of the bone, especially fracture-prone areas such as the femoral neck and lumbar vertebrae. The stiffness coefficient is calculated using the slope of the load-displacement curve, and the toughness index is the area under the curve. These indicators are compared with preset thresholds using statistical methods to calculate the probability of the indicators exceeding the thresholds, thus preliminarily judging the degree of decline in bone load-bearing capacity. In S52, when predicting the range of parameter changes, a time series prediction model such as the ARIMA model is used. The prediction period is consistent with the preset time period. The calculation of the parameter change rate uses linear regression, and the cumulative change is the sum of the changes at each time point within the prediction time period. The determination of the overall impact trend needs to be combined with medical common sense, such as the fact that aging usually leads to a decrease in bone density, while an increase in vitamin D levels shows a positive trend. When establishing the correlation model in S53, multiple linear regression is used, with the parameter change trend as the independent variable and the mechanical index as the dependent variable. Significant influencing factors are screened using stepwise regression. The model's coefficient of determination (R²) must reach above 0.7. The dynamic prediction curve is plotted using interpolation to ensure that the curve is smooth and reflects changes at key time points. In S54, the historical case data referenced must include information on at least 1000 similar patients. The critical value is determined using ROC curve analysis, selecting the point with the largest Youden index as the critical value. When calculating the number of times the critical value is exceeded, the differences in loading conditions must be considered, as the critical value varies under different activity states. The duration is calculated to the day. The final probability value is calculated using a Bayesian algorithm, combining prior probability and likelihood function, with the output probability value accurate to four decimal places. The entire process requires cross-validation to ensure that the prediction accuracy is above 80% in different patient groups.

[0047] The nomograph visualization model in this invention is a tool that intuitively presents the predicted fracture risk of osteoporosis patients and the degree of influence of each parameter. Specifically, it graphically displays the range of values ​​for various parameters, the corresponding fracture risk probability, and the interactions between parameters. To implement this model, firstly, fracture risk probability values ​​and the degree of influence of each parameter need to be compiled, and coordinate axes and scales need to be determined. The horizontal axis represents each parameter, with its length depending on the degree of influence, while the vertical axis represents the fracture risk probability, with scales ranging from 0 to 100%. Next, scale lines and risk points are drawn based on the parameter value ranges and corresponding risk probabilities, connecting them to form a risk curve. Interactions between parameters are indicated by connecting lines or annotations. After completion, clinicians are invited to evaluate and optimize the model, which is then exported to a commonly used image format. The model's function is to visualize complex multifactorial analysis results, facilitating rapid understanding of the predictions by clinicians, while also helping patients understand their own fracture risk and influencing factors, thus improving treatment adherence. Its significance lies in solving the problem of insufficient intuitiveness in traditional numerical tables or textual descriptions, providing a clear reference for the development of personalized prevention and treatment plans in clinical practice, helping to improve the efficiency of medical decision-making and reduce the incidence of fractures.

[0048] The bone time-varying factors and dynamic prediction model in this invention is used to analyze the trends of various parameters in osteoporosis patients over time and determine their weights on bone status at different time points. Specifically, the model comprehensively considers multiple parameters such as patient age, bone mineral density, and lifestyle habits, quantifying the dynamic impact of each parameter on bone quality at different times. In implementation, time-series data for the parameters are first compiled; if missing data is found, it is supplemented through interpolation to ensure that each parameter has data at at least three different time points. Then, a multivariate time-series analysis method is used to construct a regression model with the parameters as independent variables and bone mineral density change as the dependent variable, introducing time variables to analyze interactions and determine the weights of each parameter's influence at different time points. Finally, cross-validation is used to validate the model; if the error exceeds a threshold, the structure and parameters are adjusted. The model's function is to transform static parameter data into dynamic influence weights, providing dynamic parameter basis for bone biomechanical simulation and fracture risk calculation, overcoming the deficiency of existing models that ignore the time-varying characteristics of bone quality. Its significance lies in its ability to accurately capture the changing patterns of fracture risk during disease progression, making the prediction results more consistent with the patient's actual situation, providing reliable support for long-term risk assessment, and helping to improve the accuracy and timeliness of fracture risk prediction.

[0049] The mechanical simulation of skeletal microstructure in this invention is a method for constructing a three-dimensional mechanical model based on the skeletal microstructural features obtained from patient bone density and medical imaging to simulate the stress distribution, strain state, and structural stability of bones under different loads. Specifically, it comprehensively considers factors such as bone density, trabecular bone thickness, and cortical bone thickness to construct a model that reflects the true mechanical properties of bone. In implementation, high-resolution computed tomography (CT) or magnetic resonance imaging (MRI) is used to acquire skeletal medical images with a slice thickness of 0.5-1 mm. After preprocessing, image segmentation algorithms are used to extract the bone contour, distinguishing between the cortex and trabeculae, and measuring microstructural parameters. Then, a three-dimensional mesh model is constructed using 3D reconstruction software, with a mesh element size of 0.1-1 mm. Cortical and trabecular bone are assigned corresponding mechanical parameters such as elastic modulus and Poisson's ratio. Boundary conditions and load parameters are set to simulate different physiological loads, and mechanical indices are calculated using finite element analysis software. The purpose of this simulation is to provide a biomechanical basis for fracture risk assessment and accurately identify high-incidence fracture areas with stress concentration. Its significance lies in solving the problem that existing technologies rely on simplified calculations based on only a few parameters, resulting in inaccurate simulation of mechanical properties. It can more realistically reflect the mechanical state of bones under different loads, laying a solid foundation for improving the accuracy of fracture risk prediction and helping clinicians to more scientifically assess patients' bone load-bearing capacity and fracture resistance.

[0050] like Figure 2 As shown, the system for predicting fracture risk in patients with osteoporosis includes: The multi-dimensional parameter acquisition and storage unit is used to collect and store different parameters of osteoporosis patients. This unit is connected to medical testing equipment to receive test data and classify and store it. The bone time-varying model construction and analysis unit is connected to the parameter multi-dimensional acquisition and storage unit. It calls the stored parameter data to construct bone time-varying factors and dynamic prediction models, and analyzes the time change trend and influence weight of the parameters. The skeletal biomechanics simulation unit is connected to the multi-dimensional parameter acquisition and storage unit and the bone time-varying model construction and analysis unit, respectively. It receives bone density and bone microstructure data to construct a three-dimensional skeletal biomechanics model and uses parameter influence weights to correct the model's biomechanical parameters. The fracture risk probability calculation unit is connected to the skeletal biomechanics simulation unit and the bone time-varying model construction and analysis unit. Based on the corrected skeletal biomechanics simulation results and parameter change trends, it calculates the fracture risk probability value within a future preset time period. The nomograph visualization unit is connected to the fracture risk probability calculation unit and the multi-dimensional parameter acquisition and storage unit. It receives fracture risk probability values ​​and different parameter data, and generates a nomograph including the degree of parameter influence through the nomograph visualization model. The system control and coordination unit is connected to the multi-dimensional parameter acquisition and storage unit, the bone time-varying model construction and analysis unit, the skeletal biomechanics simulation unit, the fracture risk probability calculation unit, and the nomograph visualization unit, respectively, to control the operation process of different units and coordinate the data transmission and interaction between different units.

[0051] This invention presents a method and system for predicting fracture risk in patients with osteoporosis. The primary advantage of this method and system lies in its ability to accurately capture dynamic changes in bone quality, effectively overcoming the limitation of existing technologies that neglect time-varying characteristics. By constructing a dynamic prediction model for time-varying bone factors, the evolution trends and interactions of multiple patient parameters over time are comprehensively analyzed to determine the influence weight of each parameter at different time points. This dynamic analysis approach overcomes the limitations of existing technologies that rely solely on static data, comprehensively reflecting the changing patterns of fracture risk during disease progression. This makes the prediction results more closely reflect the patient's actual situation and provides a reliable basis for long-term risk assessment.

[0052] Secondly, it has significant advantages in simulating bone biomechanical properties, successfully solving the problem of inaccurate simulation in existing technologies. Utilizing mechanical simulation methods of bone microstructure, a three-dimensional mechanical model is constructed based on the patient's bone density and microstructural features obtained from medical imaging, simulating stress distribution and strain states under different loads. Simultaneously, by integrating the patient's dynamic parameters to correct the model's mechanical parameters, the simulation results more closely resemble the biomechanical characteristics of individual bones, laying a solid mechanical foundation for fracture risk assessment and significantly improving the accuracy of the assessment.

[0053] Furthermore, the method and system present the results through a nomograph visualization model, a design that offers significant practical advantages. The nomograph clearly displays the value range of each parameter, the corresponding fracture risk probability, and the interaction relationships between parameters, making complex prediction results intuitive and easy to understand. Clinicians can quickly grasp the degree of influence of each parameter on fracture risk, facilitating the development of personalized prevention and treatment plans. This effectively improves the convenience and targeted nature of clinical applications, thereby helping to reduce the incidence of fractures and alleviate the burden on healthcare.

[0054] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," "link," and "fix" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0055] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for predicting fracture risk in patients with osteoporosis, characterized in that, Includes the following steps: Step S1. Collect different parameters of osteoporosis patients, including patient age, gender, body mass index, bone mineral density test value, history of fracture, smoking status, frequency of drinking, type and duration of taking drugs that affect bone metabolism, serum calcium content, serum phosphorus content, vitamin D level, parathyroid hormone level, and bone turnover marker level. Step S2. Construct a time-varying factor and dynamic prediction model for bone quality based on the parameters. Analyze the changing trend of the parameters over time using this model to determine the influence weight of different parameters on bone quality at different time points. Step S3. Using the mechanical simulation method of bone microstructure, a three-dimensional mechanical model of bone is constructed based on the patient's bone density test value and the bone microstructure characteristics obtained through medical imaging to simulate the stress distribution, strain state and structural stability of bone under different load conditions. Step S4. Input the influence weights of the different parameters obtained in step S2 into the three-dimensional skeletal mechanical model constructed in step S3, correct the mechanical parameters in the model, and obtain the skeletal mechanical simulation results that integrate the patient's dynamic parameters. Step S5. Based on the simulation results of step S4, and combining the time-varying factors of bone quality with the parameter change trends output by the dynamic prediction model, calculate the probability value of bone fracture in the patient within the future preset time period. Step S6. The probability values ​​calculated in step S5 and the degree of influence of different parameters are visualized using a nomograph visualization model to form a nomograph that includes the range of different parameter values, the corresponding fracture risk probability, and the interaction relationship between parameters.

2. The method for predicting fracture risk in osteoporosis patients according to claim 1, characterized in that, In the time-varying factors and dynamic prediction model of bone quality constructed in step S2, the influence weights of different parameters on bone status at time t are calculated using the following formula: in, Let be the influence weight of the i-th parameter at time t; The initial weight coefficient for the i-th parameter; Let be the attenuation coefficient of the i-th parameter; Let be the cumulative influence coefficient of the i-th parameter; The cumulative rate coefficient of the i-th parameter; Let be the interaction coefficient between the i-th parameter and the k-th parameter; Let be the standardized value of the k-th parameter at time t; n is the total number of parameters.

3. The method for predicting fracture risk in osteoporosis patients according to claim 1, characterized in that, In step S3, when constructing the three-dimensional mechanical model of the skeleton, the elastic modulus of the skeleton is... Calculated using the following formula: in, BMD is the elastic modulus corresponding to standard bone mineral density; BMD is the patient's bone mineral density measurement value. Standard bone mineral density value; is the bone mineral density influence coefficient; m is the number of skeletal microstructural characteristic parameters; The influence coefficient of the j-th microstructure characteristic parameter; The standardized value of the j-th microstructural feature parameter is given. The microstructural feature parameter includes trabecular thickness, trabecular spacing, number of trabeculae, and cortical thickness.

4. The method for predicting fracture risk in osteoporosis patients according to claim 1, characterized in that, In step S4, when correcting the mechanical parameters of the three-dimensional skeletal mechanical model, the corrected bone yield strength... Calculated using the following formula: in, is the initial yield strength; p is the number of parameters involved in the correction. Let be the influence weight of the i-th parameter at time t; This is the correction factor for the yield strength of the i-th parameter; This is the interaction correction coefficient between the i-th parameter and the change in bone mineral density; Let be the change in bone density at time t under the influence of the i-th parameter.

5. The method for predicting fracture risk in osteoporosis patients according to claim 1, characterized in that, In step S5, when calculating the probability of a patient's bones fracturing within a predetermined time period, the probability value P is calculated using the following formula: in, q represents the constant term; q represents the number of parameters involved in the calculation. Let be the regression coefficient of the i-th parameter; Let be the influence weight of the i-th parameter at time t; This represents the skeletal biomechanical response value corresponding to the i-th parameter; This is the maximum stress influence coefficient; The maximum stress value obtained from the simulation of the three-dimensional mechanical model of the skeleton.

6. The method for predicting fracture risk in osteoporosis patients according to claim 1, characterized in that, In step S6, when constructing the Nomograph visualization model, the risk scores corresponding to different parameters are calculated using the following formula: in, The risk score for the i-th parameter; The conversion coefficient for the score of the i-th parameter; This represents the fracture probability when the i-th parameter acts alone. This is the baseline constant for the score.

7. The method for predicting fracture risk in osteoporosis patients according to claim 1, characterized in that, Step S3 includes the following sub-steps: S31. The acquired medical images of the patient's bones are segmented to extract the contour and internal structural features of the bone region, determine the boundary between the bone cortex and the trabeculae, distinguish different anatomical parts of the bone, and mark possible bone defects or lesion areas. S32. Based on the segmented image data, a three-dimensional reconstruction algorithm is used to construct a three-dimensional mesh model of the skeleton. The size and accuracy of the mesh unit are set to ensure that the model can accurately reflect the geometric shape of the skeleton. The surface of the model is smoothed to reduce the error caused by image noise. S33. Based on the anatomical features and material properties of the skeleton, assign corresponding mechanical parameters to the three-dimensional mesh model, including the elastic modulus, Poisson's ratio, and shear modulus of the cortex and trabeculae, set the material property differences of different bone regions, and simulate the anisotropic characteristics of the skeleton. S34. Set constraints and load parameters in the three-dimensional mechanical model, constrain the fixed ends of the bones to simulate the skeletal support under physiological conditions, apply loads related to daily activities, including the magnitude and direction of loads when walking and bending over, and determine the point of application and distribution of the loads.

8. The method for predicting fracture risk in osteoporosis patients according to claim 1, characterized in that, Step S4 includes the following sub-steps: S41. Arrange the influence weights of different parameters obtained in step S2 at different time points according to the time series, establish a dynamic matrix of weight changes with time, and the elements in the matrix correspond to the weight values ​​of different parameters at the corresponding time points. Determine the change law of parameter weights through matrix operations. S42. Based on the weight values ​​in the dynamic matrix, adjust the mechanical properties related to the corresponding parameters in the three-dimensional skeletal mechanical model. For parameters related to bone density, correct the density parameters of the model according to the weight ratio. For parameters related to bone microstructure, adjust the connection mode and distribution density of trabeculae in the model. S43. Perform finite element analysis on the modified three-dimensional mechanical model of the skeleton, calculate the displacement field, stress field and strain field of the model under different load conditions, record the location and stress value of the stress concentration area in the model, and analyze the changes in the mechanical response of the model. S44. Compare the mechanical simulation results of the model before and after correction, evaluate the influence of parameter weights on the model output. If the difference exceeds the preset threshold, readjust the mapping relationship between parameter weights and mechanical parameters until the model output results are stable.

9. The method for predicting fracture risk in osteoporosis patients according to claim 1, characterized in that, Step S5 It includes the following steps: S51. Extract calibrated mechanical indices from the skeletal biomechanical simulation results of the fused patient dynamic parameters in step S4, including maximum stress value, maximum strain value, stiffness coefficient of bone structure, and toughness index. Compare these indices with preset thresholds to preliminarily determine the load-bearing capacity of the bone. S52. Combining the time-varying factors of bone quality with the change curves of different parameters in the future preset time period output by the dynamic prediction model, predict the value range of different parameters in the time period, calculate the parameter change rate and cumulative change, and determine the overall trend of the influence of parameter changes on bone condition. S53. Establish a correlation model between mechanical indicators and parameter change trends, take parameter change trends as input variables and mechanical indicators as output variables, and use this model to predict the changes in bone mechanical indicators within a preset time period in the future, and obtain the dynamic prediction curve of mechanical indicators. S54. Based on the dynamic prediction curve of the biomechanical index and referring to the critical value of the biomechanical index when fractures occur in historical case data, calculate the number of times and the duration of the patient's bone biomechanical index exceeding the critical value within a preset time period in the future, and thus obtain the probability value of fracture risk.

10. A system for predicting fracture risk in patients with osteoporosis, characterized in that, include: The multi-dimensional parameter acquisition and storage unit is used to collect and store different parameters of osteoporosis patients. This unit is connected to medical testing equipment to receive test data and classify and store it. The bone time-varying model construction and analysis unit is connected to the parameter multi-dimensional acquisition and storage unit. It calls the stored parameter data to construct bone time-varying factors and dynamic prediction models, and analyzes the time change trend and influence weight of the parameters. The skeletal biomechanics simulation unit is connected to the multi-dimensional parameter acquisition and storage unit and the bone time-varying model construction and analysis unit, respectively. It receives bone density and bone microstructure data to construct a three-dimensional skeletal biomechanics model and uses parameter influence weights to correct the model's biomechanical parameters. The fracture risk probability calculation unit is connected to the skeletal biomechanics simulation unit and the bone time-varying model construction and analysis unit. Based on the corrected skeletal biomechanics simulation results and parameter change trends, it calculates the fracture risk probability value within a future preset time period. The nomograph visualization unit is connected to the fracture risk probability calculation unit and the multi-dimensional parameter acquisition and storage unit. It receives fracture risk probability values ​​and different parameter data, and generates a nomograph including the degree of parameter influence through the nomograph visualization model. The system control and coordination unit is connected to the multi-dimensional parameter acquisition and storage unit, the bone time-varying model construction and analysis unit, the skeletal biomechanics simulation unit, the fracture risk probability calculation unit, and the nomograph visualization unit, respectively, to control the operation process of different units and coordinate the data transmission and interaction between different units.

Images