A method for predicting osteoporosis based on an intelligent osteoporosis prediction model

Abstract

The application provides a method for osteoporosis prediction based on an intelligent osteoporosis prediction model, and relates to the technical field of osteoporosis prediction. The method comprises the following steps: obtaining detection calibration sample data, and performing analysis on initial prediction characteristic information to form initial osteoporosis prediction characteristic data; extracting sample data calibrated as normal bone mass to perform analysis on normal prediction characteristic information to form normal bone mass prediction characteristic data; establishing an osteoporosis prediction model according to the initial osteoporosis prediction characteristic data and the normal bone mass prediction characteristic data; performing osteoporosis prediction analysis on a detection target containing an osteoporosis prediction object according to the osteoporosis prediction model to form osteoporosis prediction indication result information; periodically updating the detection calibration sample data, and optimizing and updating the osteoporosis prediction model. The method extracts and analyzes the automatic characteristic information of the correlation factors causing osteoporosis, forms a more systematic and efficient prediction model, and greatly improves the accuracy of osteoporosis prediction.

Description

Technical Field

[0001] This invention relates to the field of osteoporosis prediction technology, and more specifically, to an osteoporosis prediction method based on an intelligent osteoporosis prediction model. Background Technology

[0002] Osteoporosis is a diagnosable and treatable disease [4]. It is a systemic bone disease caused by a decrease in bone density and quality, destruction of bone microstructure, and increased bone fragility due to various reasons, which makes it easy to fracture. Osteoporosis is divided into two main categories: primary and secondary. Osteoporosis can be caused by different factors depending on a person's lifestyle, living environment, physical condition, etc.

[0003] Currently, there is no systematic analytical method for predicting osteoporosis. The simplest approach mainly uses artificial neural networks for self-learning predictive analysis. However, this method cannot fully extract the characteristics of influencing factors, which requires a lot of resources and time. Furthermore, it does not form a correlation analysis of factors related to osteoporosis, but is only based on single-factor probability analysis, resulting in low accuracy of predictive analysis.

[0004] Therefore, designing an osteoporosis prediction method based on an intelligent osteoporosis prediction model, which automatically extracts feature information and performs correlation analysis on the factors that cause osteoporosis, forms a more systematic and efficient prediction model, greatly improving the accuracy of osteoporosis prediction, is an urgent problem to be solved. Summary of the Invention

[0005] The purpose of this invention is to provide a method for predicting osteoporosis based on an intelligent osteoporosis prediction model. This method involves dividing collected data on target bone subjects into normal bone and osteoporosis-labeled subjects to extract predictive features for both normal and osteoporosis-related bone. The predictive features for normal bone and osteoporosis-related bone are then compared and analyzed to extract more accurate predictive features, providing accurate comparative data for subsequent osteoporosis predictions. Furthermore, a reasonable correlation analysis is performed on potential factors leading to osteoporosis to determine the degree of correlation between different factors influencing osteoporosis. This allows for the establishment of a comprehensive predictive model that analyzes different factors when predicting osteoporosis in target subjects, significantly improving the accuracy of comprehensive osteoporosis prediction and making the prediction more systematic. In addition, periodically collecting new big data to update the prediction model fully leverages the advantages of big data to automatically improve prediction accuracy, making the prediction model more accurate and reasonable.

[0006] In a first aspect, the present invention provides a method for predicting osteoporosis based on an intelligent osteoporosis prediction model, comprising: acquiring detection and calibration sample data of objects containing bone for prediction; extracting sample data calibrated as osteoporosis for analysis of initial prediction feature information to form initial osteoporosis prediction feature data; extracting sample data calibrated as normal bone from the detection and calibration sample data for analysis of normal prediction feature information to form normal bone prediction feature data; performing prediction feature information screening and analysis based on the initial osteoporosis prediction feature data and normal bone prediction feature data to establish an osteoporosis prediction model; performing osteoporosis prediction analysis on the target object to be detected containing bone for prediction based on the osteoporosis prediction model to form osteoporosis prediction indication result information; periodically updating the detection and calibration sample data; and optimizing and updating the osteoporosis prediction model based on the updated detection and calibration sample data.

[0007] In this invention, the method divides the collected target bone object data into normal bone and osteoporosis-labeled objects to extract bone prediction feature data for normal bone and osteoporosis prediction feature data for osteoporosis. The normal bone prediction feature data and osteoporosis prediction feature data are then compared and analyzed to extract more accurate feature information for osteoporosis prediction, providing accurate predictive comparison data for subsequent osteoporosis prediction. Simultaneously, a reasonable correlation analysis is performed on possible factors leading to osteoporosis to determine the correlation between different factors and their impact on osteoporosis. This allows for the establishment of a comprehensive predictive model based on different factors when predicting osteoporosis in the target object, significantly improving the accuracy of the comprehensive analysis and making osteoporosis prediction more systematic. Furthermore, periodically collecting new big data to update the predictive model fully leverages the advantages of big data to automatically improve prediction accuracy, making the predictive model more accurate and reasonable.

[0008] One possible implementation involves acquiring detection and calibration sample data of objects with osteoporosis prediction, and extracting sample data labeled as osteoporosis for initial prediction feature information analysis to form initial osteoporosis prediction feature data. This includes: setting osteoporosis prediction factors to form an initial osteoporosis prediction factor set; based on the initial osteoporosis prediction factor set, acquiring sample collection values ​​of all osteoporosis prediction factors for each sample object labeled as osteoporosis to form sample object factor collection values ​​for different osteoporosis prediction factors under the sample object; for each osteoporosis prediction factor, aggregating the sample object factor collection values ​​of all sample objects labeled as osteoporosis to form a prediction factor collection value set; performing feature influence range analysis on each prediction factor collection value set to form an initial prediction feature range set for the factor; and combining the initial prediction feature range sets corresponding to all osteoporosis prediction factors in the initial osteoporosis prediction feature factor set to form the initial osteoporosis prediction feature data.

[0009] In this invention, the diverse factors that cause osteoporosis are considered, and these factors work together to induce osteoporosis. Therefore, when predicting osteoporosis, extracting the level characteristics required to induce osteoporosis from these influencing factors is the foundation for subsequent accurate osteoporosis prediction using this feature information. Considering the large number of factors that induce osteoporosis, the selection of factors can be based on input settings or filtering based on large datasets when extracting factor feature information. It should be noted that for the obtained osteoporosis prediction factors, since the feature information for osteoporosis prediction judgment needs to be extracted to fully determine whether the analyzed osteoporosis prediction factors are truly capable of inducing osteoporosis, the initially determined osteoporosis prediction factors cannot be completely determined to be the factors required for prediction. Generally speaking, height, weight, exercise volume, and nutritional intake that affects bone development are all factors associated with osteoporosis induction. Basically, all factors can be quantified. Therefore, when extracting feature information from sample data, quantified collected values ​​are used as the main form of feature information extraction and analysis. On the one hand, this allows for a full grasp of the level conditions required for factors to induce osteoporosis; on the other hand, it greatly simplifies the complexity of feature information extraction and analysis.

[0010] As one possible implementation, a feature influence range analysis is performed on the collected value set of each predictive factor to form an initial predictive feature range set for the factor. This includes: performing a fitting discretization analysis on the collected value sets of all sample objects in the predictive factor collection value set to form the factor fitting discrete collected value and the corresponding fitting quantity value for each factor fitting discrete collected value; and performing a continuous fitting analysis based on all factor fitting discrete collected values ​​and the corresponding fitting quantity values ​​to form a continuous function f of osteoporosis calibration features corresponding to the osteoporosis predictive factor. n (x), where x is the collected value corresponding to the osteoporosis predictor factor numbered n, and the selected x is based on the continuous function f of the osteoporosis calibration characteristics. n The values ​​obtained by (x) are the corresponding factor quantity values; based on the continuous function f of osteoporosis calibration characteristics. n (x) is used to extract the feature range based on the amount of change, forming the initial prediction feature range set of the factors corresponding to osteoporosis prediction factors.

[0011] In this invention, it is understood that for the initial osteoporosis predictor factors to be analyzed, characteristic information for predicting osteoporosis is obtained through data collection based on large datasets. Although the collected values ​​corresponding to these osteoporosis predictor factors are discrete data, they can represent continuous data characteristics under large datasets. The fluctuation characteristics or distribution characteristics of this continuous data can fully reflect the value level at which the initial osteoporosis predictor factors induce osteoporosis. Therefore, after obtaining discrete collected values, the continuous variation law of these discrete collected values ​​is fitted to obtain characteristic information, so as to fully and completely extract the characteristic range data corresponding to the initial osteoporosis predictor factors. Different initial osteoporosis predictor factors have different degrees of influence in inducing osteoporosis, and therefore the continuous characteristic curves corresponding to different initial osteoporosis predictor factors are also different. Therefore, the continuous characteristic curve of each initial osteoporosis predictor factor is fitted separately to accurately grasp the characteristics of each initial osteoporosis predictor factor in inducing osteoporosis.

[0012] As one possible approach, based on the discrete collected values ​​and corresponding fitted values ​​of all factors, continuous fitting analysis is performed to form a continuous function f of osteoporosis calibration characteristics corresponding to the osteoporosis prediction factors. n (x), including: in a fitting coordinate system where the discrete collected values ​​of the fitted factors are on the horizontal axis and the fitted quantity values ​​are on the vertical axis, marking all the discrete collected values ​​of the fitted factors and the corresponding fitted quantity values ​​as discrete points in the fitting coordinate system; setting a clustering distance threshold α, and performing clustering analysis on all the fitted discrete points in the fitting coordinate system according to the clustering distance threshold α to form clustered fitting points; and performing function fitting based on all the clustered fitting points formed by the clustering analysis and the remaining fitted discrete points to form a continuous function f of osteoporosis calibration characteristics corresponding to the osteoporosis prediction factors. n (x).

[0013] In this invention, there are various methods for fitting characteristic curves using discrete collected values. This invention provides a more reasonable fitting method, which mainly involves performing effective and reasonable clustering analysis on the original collected values ​​in the early stages of fitting, forming discrete breakpoint data that more intuitively reflects the changing characteristics during fitting. Considering that each collected value corresponds to a different sample object, arbitrary collected values ​​cannot be directly simplified and deleted, because each collected value is valid. It is understandable that for the initial osteoporosis predictor, the numerical level that can induce osteoporosis may have multiple different numerical ranges. Therefore, the collected value information for the sample objects labeled as having osteoporosis will show a large number of discrete points concentrated within certain numerical ranges and exhibiting a certain degree of fluctuation. Fitting continuous characteristic curves using discrete points aims to express the fluctuation characteristics within this clustered range. Therefore, during analysis, it's considered that some highly concentrated discrete points are not conducive to characteristic curve fitting. Clustering analysis is then performed to cluster these highly concentrated discrete points into equivalent fitting points. On the one hand, the fitting points fully capture the distribution or fluctuation trend of these highly concentrated discrete points; on the other hand, it facilitates the rapid and efficient formation of more intuitive characteristic curves during fitting, which is beneficial for the reasonable extraction of feature ranges. The clustering distance threshold can be set based on the actual concentration of discrete points or determined based on big data analysis.

[0014] As one possible implementation, a clustering distance threshold α is set, and clustering analysis is performed on all fitted discrete points in the fitted coordinate system based on the clustering distance threshold α to form clustered fitted points. This includes: extracting arbitrary fitted discrete points as object discrete points; and collecting clustered discrete points within a circular area centered on the object discrete point and with the clustering distance threshold α as the radius, in the following manner: if no other fitted discrete points exist within the circular area, the object discrete point is determined to be an isolated fitted discrete point, and any fitted discrete points other than the object discrete point are extracted to continue establishing a circular area for clustering analysis; if other fitted discrete points exist within the circular area, these other fitted discrete points are marked. For the cluster fitting discrete points, continue to perform cluster analysis by establishing a circular range with a radius equal to the clustering distance threshold α, centered on each cluster fitting discrete point, until no new cluster fitting discrete points are obtained. The set of all obtained cluster fitting discrete points and object discrete points forms a cluster fitting discrete point set. Continue to extract any fitting discrete points other than those included in the cluster fitting discrete point set and establish circular ranges for further cluster analysis. For each obtained cluster fitting discrete point set, perform cluster fitting in the following form to form the corresponding cluster fitting points: obtain the position information of all fitting discrete points in the cluster fitting discrete point set in the fitting coordinate system to form a cluster fitting discrete point position information set. Where i is the index of the cluster fitting discrete point set, and k is the index of the fitted discrete point in the cluster fitting discrete point set with index i; based on the cluster fitting discrete point location information set, the corresponding cluster fitting point B is determined. i Cluster fitting point B i Satisfy the total cluster distance Minimum, and cluster fit point B i The ordinate position is taken as an integer, where: the total cluster distance S k This represents the distance from the cluster fitting point to the fitted discrete point numbered k.

[0015] In this invention, when performing cluster analysis on discrete points, it is considered that there may be multiple characteristic ranges that the initial osteoporosis predictor factors can induce osteoporosis, and therefore, there may be multiple clusters of discrete points. Therefore, during cluster analysis, the first step is to determine how many clustered ranges of discrete points require cluster analysis. Typically, discrete points clustered within the same range are in areas of high concentration where the distance between adjacent discrete points is less than a given clustering distance threshold. These highly concentrated discrete points form continuous regions because they all satisfy the condition that the distance between surrounding discrete points is less than the clustering distance threshold. By performing a diffusion-style distance judgment on adjacent discrete points, the information of discrete points requiring clustering can be determined. During clustering, since each collected value corresponds to a different sample object and is valid, duplicate sample objects with the same collected value cannot be deleted. After all, the recurrence of data with the same collected value during clustering has a real and effective weight deviation effect on the position of the fitted points formed during clustering, thus affecting the fitting of the characteristic curve. Therefore, it is necessary to consider the collected values ​​and their corresponding quantities during cluster analysis. Using the collected values ​​and quantities as the elements of the cluster analysis, and performing minimum distance fitting under these two element characteristics, can form practically effective fitting points equivalent to these discrete cluster points. It should be noted that since the quantities corresponding to the collected values ​​are sample objects, to ensure the practical effectiveness of the fitting points, the quantity element must be an integer when performing cluster analysis.

[0016] As one possible implementation, based on the continuous function f of osteoporosis calibration characteristics. n (x), perform feature range extraction based on the amount of change to form the initial prediction feature range set of the osteoporosis prediction factors, including: determining the continuous function f of the osteoporosis calibration features. n (x) The minimum boundary value X at the x-coordinate position of the fitted coordinate system. min and maximum boundary value X max And based on the minimum boundary value X min and maximum boundary value X max Determine the average level Y ave ,in: Based on the average level Y ave Determine the continuous function f, which is a characteristic feature of osteoporosis calibration. n (x) Exceeds the average level Y ave The range of all x-axis values ​​forms the initial predictive feature range of the factors. t represents the index of the initial predictive feature range corresponding to the osteoporosis predictor factor numbered n; the set includes the initial predictive feature ranges of all factors. The initial predictive feature range set of forming factors.

[0017] In this invention, after obtaining the continuous characteristic curve corresponding to the initial osteoporosis predictor, it is necessary to correctly extract the characteristic range for predicting osteoporosis from the continuous characteristic curve. Considering that the data basis of the continuous characteristic curve is the data of sample objects labeled as osteoporosis, these raw data may themselves have reached the numerical level that induces osteoporosis. Therefore, a more obvious characteristic range can be obtained through the analysis of the average level. By performing integral processing on the continuous characteristic curve over the value range, an average level line is determined, and then the endpoint values ​​of the range on the continuous characteristic curve that exceed the average level line are defined, thereby obtaining the numerical range with obvious predictive characteristics corresponding to the initial osteoporosis predictor.

[0018] One possible approach involves extracting normal bone quality sample data from the calibration sample data and analyzing its predictive features to form normal bone quality predictive feature data. This includes: obtaining normal collection values ​​of all osteoporosis predictive factors for each sample object calibrated as normal bone quality based on the initial osteoporosis predictive factor set, forming normal collection values ​​of object factors for different osteoporosis predictive factors under the sample object; for each osteoporosis predictive factor, aggregating the normal collection values ​​of object factors under all sample objects calibrated as normal bone quality to form a factor normal collection value set; performing feature fitting analysis on discrete values ​​of each factor normal collection value set to form a continuous function of factor normal features; extracting the normal feature range based on the continuous function of factor normal features to form a set of factor normal feature ranges; and combining the factor normal feature range sets corresponding to all osteoporosis predictive factors in the initial osteoporosis predictive feature factor set to form normal bone quality predictive feature data.

[0019] In this invention, it should be noted that after analyzing and extracting the predictive feature ranges of osteoporosis from the initial predictive factors for osteoporosis in samples labeled as osteoporosis, these feature ranges cannot be directly identified as the feature ranges that can induce osteoporosis. After all, the initial predictive factors for osteoporosis are only factors that may induce osteoporosis identified in the initial stage of analysis. Therefore, it is necessary to collect the feature information of the same factors from normal sample data to obtain the normal feature ranges of these initial predictive factors at normal levels. This allows for subsequent comparison between the normal feature ranges and the predicted feature ranges for osteoporosis to determine whether the factors that can truly induce osteoporosis or those with a significant impact on inducing osteoporosis are actually the factors. Of course, whether the determined factor is the one that can induce osteoporosis or the one with a significant impact depends on the setting of conditional quantities such as cluster distance threshold and endpoint values ​​of the feature range during feature range extraction. These conditional quantities are determined according to the actual analytical needs. In addition, the normal characteristic range of factors can be obtained in the same way as the initial predicted characteristic range of factors extracted from sample objects labeled as osteoporosis. This will not be elaborated here. Of course, new methods of analysis and processing can also be used according to the actual situation.

[0020] As one possible implementation, based on initial osteoporosis prediction feature data and normal bone quality prediction feature data, a prediction feature information screening and analysis is performed to establish an osteoporosis prediction model. This includes: for the same initial osteoporosis prediction feature factors, comparing and analyzing the initial osteoporosis prediction feature data and normal bone quality prediction feature data to determine the associated osteoporosis prediction feature factors and the corresponding prediction feature range data; for each associated osteoporosis prediction feature factor, establishing a prediction condition function C corresponding to the associated osteoporosis prediction feature factor based on the corresponding prediction feature range data. v (e), where v represents the number of different associated osteoporosis predictive feature factors, and: when the obtained value e of an associated osteoporosis predictive feature factor does not belong to any predictive feature range in the corresponding predictive feature range data, then C v (e) = 0; when the obtained value e of the associated osteoporosis predictive feature factor belongs to any predictive feature range in the corresponding predictive feature range data, then C v (e) = 1; Based on the sample size of the data labeled as osteoporosis, correlation analysis was performed on different associated osteoporosis predictive characteristic factors to determine the correlation factor β corresponding to each associated osteoporosis predictive characteristic factor. v According to the prediction condition function C v (e) and the corresponding correlation factor β v The following osteoporosis prediction model is established: P = C1(e)*β1 + C2(e)*β2 + … + C v (e)*β v P represents the predicted probability of osteoporosis.

[0021] In this invention, after obtaining the initial predicted feature ranges of the initial osteoporosis predictive factors for sample data labeled as osteoporosis and the normal feature ranges of the initial osteoporosis predictive factors for normal bone sample data, a comparative analysis can be performed to determine the factors that truly induce osteoporosis, and then determine the reasonable predictive feature ranges corresponding to these factors. It should be noted that these factors do not act alone to induce osteoporosis, but rather exert a systematic effect through a combination of factors. Therefore, a single factor reaching a feature range level does not necessarily, or is highly unlikely, induce osteoporosis. Instead, it is necessary to comprehensively consider whether all related osteoporosis predictive feature factors reach the value levels of their feature ranges to analyze and judge whether osteoporosis is induced. Based on this, after determining the related osteoporosis predictive feature factors through the analysis of the initial predicted feature ranges and normal feature ranges of the factors, a model for osteoporosis prediction can be established to provide probabilistic guidance data for osteoporosis prediction. Here, under the prediction model, each associated osteoporosis prediction feature factor reaching the feature range level can increase the probability of inducing osteoporosis. Therefore, for each associated osteoporosis prediction feature factor, given the feature range, a conditional expression for the prediction model can be established to provide accurate correlation probability when conducting analysis and judgment, making the prediction results of the prediction model more reasonable and accurate.

[0022] As one possible approach, for the same initial osteoporosis prediction feature factors, the initial osteoporosis prediction feature data and normal bone quality prediction feature data are compared and analyzed to determine the associated osteoporosis prediction feature factors and the prediction feature range data corresponding to the associated osteoporosis prediction feature factors, including: the initial prediction feature range of all factors corresponding to the initial osteoporosis prediction feature factors. The initial predicted feature range of each factor is compared with the normal feature ranges of all factors in the corresponding normal feature range set as follows: If the normal feature range set of factors includes the initial predicted feature range of each factor corresponding to the initial osteoporosis prediction feature factor... Then the initial osteoporosis prediction feature factor is determined to be an unrelated predictor factor; if the normal feature range of the factor in the factor normal feature range set matches the initial prediction feature range of each factor corresponding to the initial osteoporosis prediction feature factor... If there is no overlap, then the initial osteoporosis prediction feature factor is determined as the associated osteoporosis prediction feature factor, and the initial prediction feature range of all factors is determined as the corresponding prediction feature range; if there exists a factor normal feature range set that includes part of the initial prediction feature range of the initial osteoporosis prediction feature factor, then... Then, the initial osteoporosis prediction feature factors are determined as associated osteoporosis prediction feature factors, and the initial prediction feature ranges of factors that do not have an inclusion relationship are determined as the corresponding prediction feature ranges; if there exists a factor normal feature range in the factor normal feature range set that corresponds to the initial prediction feature range of the initial osteoporosis prediction feature factor... If there is an intersection, the initial osteoporosis prediction feature factor is determined as the associated osteoporosis prediction feature factor, and the range formed after excluding the intersection part of the initial prediction feature range of the factor is determined as the corresponding prediction feature range; combined with all the prediction feature ranges under the associated osteoporosis prediction feature factor, the corresponding prediction feature range data is formed.

[0023] In this invention, the determination of whether an initial osteoporosis predictor is a relevant osteoporosis predictor feature factor required by the prediction model, and the comparison of the accurate feature range information of the relevant osteoporosis predictor feature factor, are performed through a comparative analysis of the initial predictive feature range and the normal feature range of the factor. That is, if the initial osteoporosis predictor is not a relevant osteoporosis predictor feature factor required by the prediction model, then the feature range extracted for the sample data labeled as osteoporosis is not a true feature range; these feature ranges are actually feature ranges obtained from normal data. Therefore, the initial predictive feature range and the normal feature range of such an initial osteoporosis predictor have a complete inclusion relationship, and the normal feature range must contain the initial predictive feature range, since the sample data of normal bone tissue is large and accurate. If the initial osteoporosis predictor is a relevant osteoporosis predictor feature factor required by the prediction model, then there is no complete inclusion relationship between the initial predictive feature range and the normal feature range; instead, there may be partial overlap, no overlap, or only some feature ranges being included. Through range comparison analysis, the relevant osteoporosis predictor feature factor can be quickly and accurately determined. Meanwhile, after identifying the predictive features of osteoporosis, the initial predictive feature range of the factors that do not overlap with the normal feature range of the factors is determined as the final feature range. This feature range is compared and processed to greatly improve the accuracy, making the analysis of the prediction model more accurate and reasonable.

[0024] As one possible approach, based on the sample size of the data labeled as osteoporosis, correlation analysis is performed on different associated osteoporosis predictive features to determine the correlation factor β corresponding to each associated osteoporosis predictive feature. v This includes: for all sample objects labeled as osteoporosis, determining the number of sample objects whose collected values ​​of the associated osteoporosis predictive feature factors for each associated osteoporosis predictive feature factor belong to the predictive feature range data, and obtaining the associated sample frequency W corresponding to each associated osteoporosis predictive feature factor. v ,in: Qv Q represents the number of sample objects whose acquired predictive feature factor values ​​belong to the predictive feature range data corresponding to the predictive feature range data of the predictive feature factor with number v for osteoporosis. all This represents the total number of sample subjects categorized as osteoporosis; based on the frequency W of each associated sample. v Determine the corresponding correlation factor β v ,in:

[0025] In this invention, the correlation factor reflects the weight ratio of different associated osteoporosis predictive features relative to other associated osteoporosis predictive features inducing osteoporosis. After determining all associated osteoporosis predictive features, frequency statistics can be performed on each associated osteoporosis predictive feature to determine the correlation factor in the sample data labeled as osteoporosis. After all, in the sample data labeled as osteoporosis, if an associated osteoporosis predictive feature has a greater impact on inducing osteoporosis, then the frequency of its collected values ​​falling within the characteristic range will be higher. Therefore, frequency statistics can accurately determine the magnitude of the correlation factor, and normalization processing can make the prediction model results more reasonable and facilitate cross-data comparison.

[0026] The beneficial effects of the osteoporosis prediction method based on an intelligent osteoporosis prediction model provided by this invention are as follows:

[0027] This method divides collected data on target bone subjects into normal bone and osteoporosis-labeled subjects to extract predictive features for both. The predictive features for normal bone and osteoporosis are then compared and analyzed to extract more accurate predictive features, providing precise comparative data for subsequent osteoporosis predictions. Furthermore, it incorporates reasonable correlation analysis with potential factors contributing to osteoporosis, determining the degree of correlation between different factors and their impact on osteoporosis. This allows for the establishment of a comprehensive predictive model that integrates various factors, significantly improving the accuracy of osteoporosis prediction and making the prediction more systematic. Additionally, periodically collecting new big data to update the predictive model leverages the advantages of big data to automatically improve prediction accuracy, making the model more accurate and reasonable. Attached Figure Description

[0028] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments of the present invention will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 A flowchart illustrating the steps of the osteoporosis prediction method based on an intelligent osteoporosis prediction model provided in this embodiment of the invention. Detailed Implementation

[0030] The technical solutions of the present invention will now be described with reference to the accompanying drawings in the embodiments of the present invention.

[0031] Osteoporosis is a diagnosable and treatable disease [4]. It is a systemic bone disease caused by a decrease in bone density and quality, destruction of bone microstructure, and increased bone fragility due to various reasons, which makes it easy to fracture. Osteoporosis is divided into two main categories: primary and secondary. Osteoporosis can be caused by different factors depending on a person's lifestyle, living environment, physical condition, etc.

[0032] Currently, there is no systematic analytical method for predicting osteoporosis. The simplest approach mainly uses artificial neural networks for self-learning predictive analysis. However, this method cannot fully extract the characteristics of influencing factors, which requires a lot of resources and time. Furthermore, it does not form a correlation analysis of factors related to osteoporosis, but is only based on single-factor probability analysis, resulting in low accuracy of predictive analysis.

[0033] refer to Figure 1 This invention provides a method for predicting osteoporosis based on an intelligent osteoporosis prediction model.

[0034] The osteoporosis prediction method based on the intelligent osteoporosis prediction model specifically includes the following steps:

[0035] S1: Obtain the detection and calibration sample data of the bone prediction object, and extract the sample data calibrated as osteoporosis to analyze the initial prediction feature information and form the initial osteoporosis prediction feature data.

[0036] Acquire detection and calibration sample data containing osteoporosis prediction objects, and extract sample data calibrated as osteoporosis for initial prediction feature information analysis to form initial osteoporosis prediction feature data. This includes: setting osteoporosis prediction factors to form an initial osteoporosis prediction factor set; based on the initial osteoporosis prediction factor set, obtaining sample collection values ​​of all osteoporosis prediction factors for each sample object calibrated as osteoporosis to form sample object factor collection values ​​for different osteoporosis prediction factors under the sample object; for each osteoporosis prediction factor, aggregating the sample object factor collection values ​​of all sample objects calibrated as osteoporosis to form a prediction factor collection value set; performing feature influence range analysis on each prediction factor collection value set to form an initial prediction feature range set for the factor; and combining the initial prediction feature range sets corresponding to all osteoporosis prediction factors in the initial osteoporosis prediction feature factor set to form the initial osteoporosis prediction feature data.

[0037] Considering the diverse factors that cause osteoporosis, and the combined effects of these factors inducing it, extracting the required level of these influencing factors to induce osteoporosis is fundamental for accurate prediction. Given the numerous factors that induce osteoporosis, the selection of factors can be based on input settings or filtering based on large datasets. It's important to note that the initial identification of osteoporosis predictive factors cannot be definitively confirmed as truly inducing osteoporosis; therefore, the extracted predictive factors must be fully extracted to determine their predictive value. Generally, height, weight, exercise levels, and nutritional intake affecting bone development are all associated with osteoporosis. Since almost all these factors are quantifiable, using quantified values ​​as the primary form of feature extraction analysis allows for a thorough understanding of the required level of factors to induce osteoporosis and significantly simplifies the complexity of feature extraction analysis.

[0038] Specifically, for each predictive factor's collected value set, a feature influence range analysis is performed to form an initial predictive feature range set for the factor. This includes: performing a fitting discretization analysis on the collected values ​​of all sample objects in the predictive factor's collected value set to form the factor's fitted discrete collected values ​​and the corresponding fitted quantity value for each factor's fitted discrete collected values; and performing a continuous fitting analysis based on all factor fitted discrete collected values ​​and the corresponding fitted quantity values ​​to form a continuous function f for osteoporosis calibration characteristics corresponding to the osteoporosis predictive factor. n (x), where x is the collected value corresponding to the osteoporosis predictor factor numbered n, and the selected x is based on the continuous function f of the osteoporosis calibration characteristics. n The values ​​obtained by (x) are the corresponding factor quantity values; based on the continuous function f of osteoporosis calibration characteristics. n (x) is used to extract the feature range based on the amount of change, forming the initial prediction feature range set of the factors corresponding to osteoporosis prediction factors.

[0039] Understandably, for the initial osteoporosis predictor factors to be analyzed, characteristic information for predicting osteoporosis is obtained by collecting values ​​based on big data. Although the collected values ​​corresponding to these osteoporosis predictor factors are discrete data, they can represent continuous data characteristics under big data. The fluctuation characteristics or distribution characteristics of this continuous data can fully reflect the value level at which the initial osteoporosis predictor factors induce osteoporosis. Therefore, after obtaining discrete collected values, the continuous variation law of these discrete collected values ​​is fitted to obtain characteristic information, so as to fully and completely extract the characteristic range data corresponding to the initial osteoporosis predictor factors. Different initial osteoporosis predictor factors have different degrees of influence in inducing osteoporosis, so the continuous characteristic curves corresponding to different initial osteoporosis predictor factors are also different. Therefore, the continuous characteristic curve of each initial osteoporosis predictor factor is fitted separately to accurately grasp the characteristics of each initial osteoporosis predictor factor in inducing osteoporosis.

[0040] Based on the discrete collected values ​​and corresponding fitted values ​​of all factors, continuous fitting analysis is performed to form a continuous function f of osteoporosis calibration characteristics corresponding to the osteoporosis prediction factors. n (x), including: in a fitting coordinate system where the discrete collected values ​​of the fitted factors are on the horizontal axis and the fitted quantity values ​​are on the vertical axis, marking all the discrete collected values ​​of the fitted factors and the corresponding fitted quantity values ​​as discrete points in the fitting coordinate system; setting a clustering distance threshold α, and performing clustering analysis on all the fitted discrete points in the fitting coordinate system according to the clustering distance threshold α to form clustered fitting points; and performing function fitting based on all the clustered fitting points formed by the clustering analysis and the remaining fitted discrete points to form a continuous function f of osteoporosis calibration characteristics corresponding to the osteoporosis prediction factors. n (x).

[0041] There are various methods for fitting characteristic curves using discrete collected values. This invention provides a more reasonable fitting method, which mainly involves performing effective and reasonable clustering analysis on the original collected values ​​in the early stage of fitting, forming discrete breakpoint data that more intuitively reflects the changing characteristics during fitting. Considering that each collected value corresponds to a different sample object, arbitrary collected values ​​cannot be directly simplified and deleted, because each collected value is valid. It is understandable that for the initial osteoporosis predictor, the numerical level that can induce osteoporosis may have multiple different numerical ranges. Thus, the collected value information of the sample objects labeled as osteoporosis will show a large number of discrete points concentrated in a certain numerical range and exhibiting a certain degree of fluctuation. Fitting continuous characteristic curves using discrete points aims to express the fluctuation characteristics within this clustered range. Therefore, during analysis, it's considered that some highly concentrated discrete points are not conducive to characteristic curve fitting. Clustering analysis is then performed to cluster these highly concentrated discrete points into equivalent fitting points. On the one hand, the fitting points fully capture the distribution or fluctuation trend of these highly concentrated discrete points; on the other hand, it facilitates the rapid and efficient formation of more intuitive characteristic curves during fitting, which is beneficial for the reasonable extraction of feature ranges. The clustering distance threshold can be set based on the actual concentration of discrete points or determined based on big data analysis.

[0042] A clustering distance threshold α is set, and clustering analysis is performed on all fitted discrete points in the fitted coordinate system based on the clustering distance threshold α to form clustered fitting points. This includes: extracting arbitrary fitted discrete points as object discrete points; and collecting clustered discrete points within a circular area centered on the object discrete point and with a clustering distance threshold α as the radius, in the following manner: if no other fitted discrete points exist within the circular area, the object discrete point is determined to be an isolated fitted discrete point, and any other fitted discrete points besides the object discrete point are extracted to continue establishing a circular area for clustering analysis; if other fitted discrete points exist within the circular area, these other fitted discrete points are marked as clustered fitting points. For scattered points, continue clustering analysis by establishing a circular range with a radius equal to the clustering distance threshold α, centered on each cluster-fitted discrete point, until no new cluster-fitted discrete points are obtained. The resulting set of all cluster-fitted discrete points and object discrete points forms a cluster-fitted discrete point set. Then, extract any fitted discrete points (excluding those already included in the cluster-fitted discrete point set) and continue clustering analysis within the same circular range. For each cluster-fitted discrete point set, perform cluster fitting in the following form to form corresponding cluster-fitted points: obtain the position information of all fitted discrete points in the cluster-fitted discrete point set within the fitting coordinate system, forming a cluster-fitted discrete point position information set. Where i is the index of the cluster fitting discrete point set, and k is the index of the fitted discrete point in the cluster fitting discrete point set with index i; based on the cluster fitting discrete point location information set, the corresponding cluster fitting point B is determined. i Cluster fitting point B i Satisfy the total cluster distance Minimum, and cluster fit point B i The ordinate position is taken as an integer, where: the total cluster distance S k This represents the distance from the cluster fitting point to the fitted discrete point numbered k.

[0043] When performing cluster analysis on discrete points, it's possible that there are multiple characteristic ranges within which initial osteoporosis predictors can induce osteoporosis, resulting in multiple clusters of discrete points. Therefore, the first step in cluster analysis is to determine how many clustered ranges of discrete points require clustering. Typically, discrete points clustered within the same range tend to have higher concentrations, with the distance between adjacent points being less than a given clustering distance threshold. These highly concentrated discrete points form continuous regions because they all satisfy the condition that the distance between surrounding discrete points is less than the clustering distance threshold. By performing a diffusion-style distance assessment of adjacent discrete points, the information of discrete points requiring clustering can be determined. During clustering, since each collected value corresponds to a different sample object and is valid, duplicate sample objects with the same collected value cannot be deleted. After all, the recurrence of identical collected values ​​during clustering has a real and effective weight deviation effect on the position of the fitted points formed during clustering, thus affecting the fitting of the characteristic curve. Therefore, it is necessary to consider the collected values ​​and their corresponding quantities during cluster analysis. Using the collected values ​​and quantities as the elements of the cluster analysis, and performing minimum distance fitting under these two element characteristics, can form practically effective fitting points equivalent to these discrete cluster points. It should be noted that since the quantities corresponding to the collected values ​​are sample objects, to ensure the practical effectiveness of the fitting points, the quantity element must be an integer when performing cluster analysis.

[0044] Based on the continuous function f of osteoporosis calibration characteristics n (x), perform feature range extraction based on the amount of change to form the initial prediction feature range set of the osteoporosis prediction factors, including: determining the continuous function f of the osteoporosis calibration features. n (x) The minimum boundary value X at the x-coordinate position of the fitted coordinate system. min and maximum boundary value X max And based on the minimum boundary value X min and maximum boundary value X max Determine the average level Y ave ,in: Based on the average level Yave Determine the continuous function f, which is a characteristic feature of osteoporosis calibration. n (x) Exceeds the average level Y ave The range of all x-axis values ​​forms the initial predictive feature range of the factors. t represents the index of the initial predictive feature range corresponding to the osteoporosis predictor factor numbered n; the set includes the initial predictive feature ranges of all factors. The initial predictive feature range set of forming factors.

[0045] After obtaining the continuous characteristic curves corresponding to the initial osteoporosis predictor factors, it is necessary to correctly extract the characteristic ranges used to predict osteoporosis from these curves. Considering that the data basis for the continuous characteristic curves is data from samples labeled as osteoporosis, these raw data may already have reached numerical levels that induce osteoporosis. Therefore, a more obvious characteristic range can be obtained through average level analysis. By integrating the continuous characteristic curve over the value range, an average line is determined, and then the endpoints of the range exceeding the average line on the continuous characteristic curve are defined, thus obtaining the numerical range with obvious predictive characteristics corresponding to the initial osteoporosis predictor factors.

[0046] S2: Extract sample data calibrated as normal bone from the test and calibration sample data, analyze the normal predictive feature information, and form normal bone predictive feature data.

[0047] The analysis of normal predictive feature information from sample data calibrated as normal bone mass is performed on the initial osteoporosis prediction factor set to form normal bone mass prediction feature data. This includes: obtaining normal collection values ​​of all osteoporosis prediction factors for each sample object calibrated as normal bone mass, forming normal collection values ​​of object factors for different osteoporosis prediction factors under the sample object; for each osteoporosis prediction factor, a set of normal collection values ​​of object factors under all sample objects calibrated as normal bone mass is collected to form a factor normal collection value set; feature fitting analysis of discrete values ​​is performed on the normal collection value set of each factor to form a continuous function of factor normal features; the normal feature range is extracted based on the continuous function of factor normal features to form a set of factor normal feature ranges; and the normal feature range sets corresponding to all osteoporosis prediction factors in the initial osteoporosis prediction feature factor set are combined to form normal bone mass prediction feature data.

[0048] It should be noted that after analyzing and extracting the predictive feature ranges of osteoporosis from the initial predictive factors for osteoporosis in samples labeled as osteoporosis, these feature ranges cannot be directly assumed to be the same as those that can induce osteoporosis. This is because the initial predictive factors are only those factors initially identified as potentially inducing osteoporosis. Therefore, it is necessary to collect the same feature information from normal sample data to obtain the normal feature ranges of these initial predictive factors at normal levels. This allows for a subsequent comparison between the normal feature ranges and the predicted osteoporosis feature ranges to determine whether the factors truly induce osteoporosis or those with a significant impact on its induction. Whether the identified factor is the one that induces osteoporosis or the one with a significant impact depends on the settings of conditional quantities such as cluster distance thresholds and endpoint values ​​of the feature ranges during feature range extraction. These conditional quantities are determined based on the actual analytical needs. Furthermore, the normal feature ranges of factors can be obtained in the same way as the initial predictive feature ranges extracted from samples labeled as osteoporosis, which will not be elaborated here. Of course, new analytical methods can also be used depending on the specific circumstances.

[0049] S3: Based on the initial osteoporosis prediction feature data and normal bone quality prediction feature data, perform screening and analysis of prediction feature information to establish an osteoporosis prediction model.

[0050] Based on initial osteoporosis prediction feature data and normal bone mass prediction feature data, a screening and analysis of prediction feature information is performed to establish an osteoporosis prediction model. This includes: for the same initial osteoporosis prediction feature factors, comparing and analyzing the initial osteoporosis prediction feature data and normal bone mass prediction feature data to determine the associated osteoporosis prediction feature factors and the corresponding prediction feature range data; for each associated osteoporosis prediction feature factor, establishing a prediction condition function C corresponding to the associated osteoporosis prediction feature factor based on the corresponding prediction feature range data. v (e), where v represents the number of different associated osteoporosis predictive feature factors, and: when the obtained value e of an associated osteoporosis predictive feature factor does not belong to any predictive feature range in the corresponding predictive feature range data, then C v (e) = 0; when the obtained value e of the associated osteoporosis predictive feature factor belongs to any predictive feature range in the corresponding predictive feature range data, then C v (e) = 1; Based on the sample size of the data labeled as osteoporosis, correlation analysis was performed on different associated osteoporosis predictive characteristic factors to determine the correlation factor β corresponding to each associated osteoporosis predictive characteristic factor. v According to the prediction condition function C v (e) and the corresponding correlation factor β vThe following osteoporosis prediction model is established: P = C1(e)*β1 + C2(e)*β2 + … + C v (e)*β v P represents the predicted probability of osteoporosis.

[0051] After obtaining the initial predicted characteristic ranges of the osteoporosis predictive factors for sample data labeled as osteoporosis and the normal characteristic ranges of the initial predicted characteristic factors for osteoporosis for normal bone sample data, a comparative analysis can be performed to determine the factors that truly induce osteoporosis, and then determine the reasonable predictive characteristic ranges for these factors. It should be noted that these factors do not act alone to induce osteoporosis, but rather exert a systematic effect through a combination of factors. Therefore, a single factor reaching its characteristic range does not necessarily, or is highly unlikely, induce osteoporosis. Instead, it is necessary to comprehensively consider whether all related osteoporosis predictive characteristic factors reach their characteristic range values ​​to analyze and judge whether osteoporosis is induced. Based on this, after determining the related osteoporosis predictive characteristic factors through the analysis of the initial predicted characteristic ranges and normal characteristic ranges of the factors, a model for osteoporosis prediction can be established to provide probabilistic guidance data for osteoporosis prediction. Here, under the prediction model, each associated osteoporosis prediction feature factor reaching the feature range level can increase the probability of inducing osteoporosis. Therefore, for each associated osteoporosis prediction feature factor, given the feature range, a conditional expression for the prediction model can be established to provide accurate correlation probability when conducting analysis and judgment, making the prediction results of the prediction model more reasonable and accurate.

[0052] Specifically, for the same initial osteoporosis prediction characteristic factors, the initial osteoporosis prediction characteristic data and normal bone mass prediction characteristic data are compared and analyzed to determine the associated osteoporosis prediction characteristic factors and the prediction characteristic range data corresponding to the associated osteoporosis prediction characteristic factors. This includes: the initial prediction characteristic range of all factors corresponding to the initial osteoporosis prediction characteristic factors. The initial predicted feature range of each factor is compared with the normal feature ranges of all factors in the corresponding normal feature range set as follows: If the normal feature range set of factors includes the initial predicted feature range of each factor corresponding to the initial osteoporosis prediction feature factor... Then the initial osteoporosis prediction feature factor is determined to be an unrelated predictor factor; if the normal feature range of the factor in the factor normal feature range set matches the initial prediction feature range of each factor corresponding to the initial osteoporosis prediction feature factor... If there is no overlap, then the initial osteoporosis prediction feature factor is determined as the associated osteoporosis prediction feature factor, and the initial prediction feature range of all factors is determined as the corresponding prediction feature range; if there exists a factor normal feature range set that includes part of the initial prediction feature range of the initial osteoporosis prediction feature factor, then... Then, the initial osteoporosis prediction feature factors are determined as associated osteoporosis prediction feature factors, and the initial prediction feature ranges of factors that do not have an inclusion relationship are determined as the corresponding prediction feature ranges; if there exists a factor normal feature range in the factor normal feature range set that corresponds to the initial prediction feature range of the initial osteoporosis prediction feature factor... If there is an intersection, the initial osteoporosis prediction feature factor is determined as the associated osteoporosis prediction feature factor, and the range formed after excluding the intersection part of the initial prediction feature range of the factor is determined as the corresponding prediction feature range; combined with all the prediction feature ranges under the associated osteoporosis prediction feature factor, the corresponding prediction feature range data is formed.

[0053] Here, to determine whether the initial osteoporosis predictor is the relevant osteoporosis predictor feature factor required by the prediction model, and to compare and determine the accurate feature range information of the relevant osteoporosis predictor feature factor, a comparative analysis of the initial predictive feature range and the normal feature range of the factor is used. That is, if the initial osteoporosis predictor is not the relevant osteoporosis predictor feature factor required by the prediction model, then the feature range extracted for the sample data labeled as osteoporosis is not the true feature range; these feature ranges are actually feature ranges obtained from normal data. Therefore, the initial predictive feature range and the normal feature range of such an initial osteoporosis predictor have a complete inclusion relationship, and the normal feature range must contain the initial predictive feature range, since the sample data of normal bone is large and accurate. If the initial osteoporosis predictor is the relevant osteoporosis predictor feature factor required by the prediction model, then there is no complete inclusion relationship between the initial predictive feature range and the normal feature range; instead, there may be partial overlap, no overlap, or only some feature ranges being included. Through range comparison analysis, the relevant osteoporosis predictor feature factor can be quickly and accurately determined. Meanwhile, after identifying the predictive features of osteoporosis, the initial predictive feature range of the factors that do not overlap with the normal feature range of the factors is determined as the final feature range. This feature range is compared and processed to greatly improve the accuracy, making the analysis of the prediction model more accurate and reasonable.

[0054] Based on the sample size of the data labeled as osteoporosis, correlation analysis was conducted on different associated predictive features of osteoporosis to determine the correlation factor β corresponding to each associated predictive feature of osteoporosis. v This includes: for all sample objects labeled as osteoporosis, determining the number of sample objects whose collected values ​​of the associated osteoporosis predictive feature factors for each associated osteoporosis predictive feature factor belong to the predictive feature range data, and obtaining the associated sample frequency W corresponding to each associated osteoporosis predictive feature factor. v ,in: Q vQ represents the number of sample objects whose acquired predictive feature factor values ​​belong to the predictive feature range data corresponding to the predictive feature range data of the predictive feature factor with number v for osteoporosis. all This represents the total number of sample subjects categorized as osteoporosis; based on the frequency W of each associated sample. v Determine the corresponding correlation factor β v ,in:

[0055] The correlation factor reflects the weight ratio of different associated osteoporosis predictive features relative to other associated osteoporosis predictive features inducing osteoporosis. After identifying all associated osteoporosis predictive features, frequency statistics can be performed on each associated osteoporosis predictive feature to determine the correlation factor within the labeled osteoporosis sample data. After all, in the labeled osteoporosis sample data, if an associated osteoporosis predictive feature has a significant impact on inducing osteoporosis, then the frequency of its collected values ​​falling within the characteristic range will be higher. Therefore, frequency statistics can accurately determine the magnitude of the correlation factor, and normalization processing can make the prediction model results more reasonable and facilitate cross-data comparisons.

[0056] S4: Based on the osteoporosis prediction model, perform osteoporosis prediction analysis on the target object containing the osteoporosis prediction object, and generate osteoporosis prediction indication results information.

[0057] After obtaining the osteoporosis prediction model, it is only necessary to compare and analyze the values ​​of the associated osteoporosis prediction feature factors of the target object containing the osteoporosis prediction object with the feature range to determine the indication result information, that is, the probability of predicting osteoporosis, which can be used as a data reference for osteoporosis prediction analysis.

[0058] S5: Periodically update the detection and calibration sample data, and optimize and update the osteoporosis prediction model based on the updated detection and calibration sample data.

[0059] Since the predictive model is based on the collection of feature information from historical data, the basic data for predictive analysis of osteoporosis in different groups at different times may change. Therefore, it is necessary to update the basic sample data to obtain reasonable and timely feature range information to ensure the accuracy of the predictive analysis results.

[0060] In summary, the beneficial effects of the osteoporosis prediction method based on the intelligent osteoporosis prediction model provided in this embodiment of the invention are as follows:

[0061] This method divides collected data on target bone subjects into normal bone and osteoporosis-labeled subjects to extract predictive features for both. The predictive features for normal bone and osteoporosis are then compared and analyzed to extract more accurate predictive features, providing precise comparative data for subsequent osteoporosis predictions. Furthermore, it incorporates reasonable correlation analysis with potential factors contributing to osteoporosis, determining the degree of correlation between different factors and their impact on osteoporosis. This allows for the establishment of a comprehensive predictive model that integrates various factors, significantly improving the accuracy of osteoporosis prediction and making the prediction more systematic. Additionally, periodically collecting new big data to update the predictive model leverages the advantages of big data to automatically improve prediction accuracy, making the model more accurate and reasonable.

[0062] In the embodiments of this application, "instruction" can include direct and indirect instructions, as well as explicit and implicit instructions. The information indicated by a certain piece of information is called the information to be instructed. In the specific implementation process, there are many ways to instruct the information to be instructed, such as, but not limited to, directly instructing the information to be instructed, such as the information to be instructed itself or its index. It can also indirectly instruct the information to be instructed by instructing other information, where there is a relationship between the other information and the information to be instructed. It can also instruct only a part of the information to be instructed, while the other parts are known or pre-agreed upon. For example, the instruction of specific information can be achieved by using a pre-agreed (e.g., protocol-defined) arrangement of various pieces of information, thereby reducing instruction overhead to some extent. At the same time, common parts of various pieces of information can be identified and uniformly indicated to reduce the instruction overhead caused by individually indicating the same information.

[0063] Furthermore, the specific indication method can also be any existing indication method, such as, but not limited to, the above-mentioned indication methods and their various combinations. Specific details of various indication methods can be found in existing technologies, and will not be repeated here. As described above, for example, when multiple pieces of information of the same type need to be indicated, the indication methods for different pieces of information may differ. In the specific implementation process, the required indication method can be selected according to specific needs. This application embodiment does not limit the selected indication method; therefore, the indication methods involved in this application embodiment should be understood to cover various methods that enable the party to be indicated to obtain the information to be indicated.

[0064] It should be understood that the information to be indicated can be sent as a whole or divided into multiple sub-information messages sent separately, and the sending period and / or timing of these sub-information messages can be the same or different. The specific sending method is not limited in this application embodiment. The sending period and / or timing of these sub-information messages can be predefined, for example, according to a protocol, or configured by the sending device by sending configuration information to the receiving device.

[0065] "Predefined" or "pre-configured" can be achieved by pre-saving corresponding codes, tables, or other means that can be used to indicate relevant information in the device. This application does not limit the specific implementation method. "Saving" can refer to saving in one or more memories. These memories can be separate installations or integrated into the encoder, decoder, processor, or communication device. Alternatively, some memories can be separately installed, while others are integrated into the decoder, processor, or communication device. The type of memory can be any form of storage medium, and this application does not limit this.

[0066] The “protocol” mentioned in this application embodiment may refer to a protocol family in the field of communication, a standard protocol with a similar protocol family frame structure, or a related protocol applied to future communication systems. This application embodiment does not specifically limit this.

[0067] In the embodiments of this application, descriptions such as "when," "under the circumstances," "if," and "if" all refer to the device making corresponding processing under certain objective circumstances, and are not limited to a specific time. They do not require the device to make a judgment action during implementation, nor do they imply any other limitations.

[0068] In the description of the embodiments of this application, unless otherwise stated, " / " indicates that the objects before and after are in an "or" relationship. For example, A / B can represent A or B. "And / or" in the embodiments of this application is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone, where A and B can be singular or plural. Furthermore, in the description of the embodiments of this application, unless otherwise stated, "multiple" refers to two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple. Additionally, to facilitate a clear description of the technical solutions of the embodiments of this application, the terms "first" and "second" are used in the embodiments of this application to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first," "second," etc., do not limit the quantity or order of execution, and that "first," "second," etc., are not necessarily different. Furthermore, in the embodiments of this application, words such as "exemplary" or "for example" are used to indicate that something is being used as an example, illustration, or description. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner for ease of understanding.

[0069] It should be understood that the processor in the embodiments of this application can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0070] It should also be understood that the memory in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DRRAM).

[0071] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0072] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0073] In this application, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or multiple items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0074] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0075] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0076] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0077] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0078] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0079] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0080] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0081] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for predicting osteoporosis based on an intelligent osteoporosis prediction model, characterized in that, include: Acquire detection and calibration sample data of objects containing osteoporosis, and extract sample data calibrated as osteoporosis for initial prediction feature information analysis to form initial osteoporosis prediction feature data; The normal bone quality prediction feature information is analyzed by extracting sample data calibrated as normal bone quality from the detection and calibration sample data to form normal bone quality prediction feature data. Based on the initial osteoporosis prediction feature data and the normal bone quality prediction feature data, the prediction feature information is screened and analyzed to establish an osteoporosis prediction model; Based on the osteoporosis prediction model, osteoporosis prediction analysis is performed on the target to be detected containing osteoporosis prediction objects to generate osteoporosis prediction indication results information; The detection and calibration sample data are periodically updated, and the osteoporosis prediction model is optimized and updated based on the updated detection and calibration sample data.

2. The osteoporosis prediction method based on an intelligent osteoporosis prediction model according to claim 1, characterized in that, The process of acquiring detection and calibration sample data containing bone-predicting objects, and extracting sample data labeled as osteoporosis for initial prediction feature information analysis to form initial osteoporosis prediction feature data, includes: Define osteoporosis predictors to form an initial set of osteoporosis predictors; Based on the initial set of osteoporosis prediction factors, sample collection values ​​of all osteoporosis prediction factors are obtained for each sample object labeled as osteoporosis, forming sample object factor collection values ​​of different osteoporosis prediction factors under the sample object; For each osteoporosis predictor, the collected values ​​of the factor for all sample objects labeled as osteoporosis are aggregated to form a predictor factor collection value set; Perform feature influence range analysis on the collected value set of each of the predicted factors to form an initial predicted feature range set for the factors; The initial osteoporosis prediction feature data is formed by combining the initial prediction feature range set of all osteoporosis prediction factors in the initial osteoporosis prediction feature factor set.

3. The osteoporosis prediction method based on an intelligent osteoporosis prediction model according to claim 2, characterized in that, The step of performing feature influence range analysis on the collected value set of each of the predicted factors to form an initial predicted feature range set for the factors includes: The sample object factor collection values ​​in the collection set of the prediction factors are subjected to fitting discretization analysis to form factor fitting discrete collection values ​​and fitting quantity values ​​corresponding to each factor fitting discrete collection value; Based on the discrete collected values ​​and corresponding fitted quantity values ​​of all the factors, continuous fitting analysis is performed to form the continuous function f of osteoporosis calibration characteristics corresponding to the osteoporosis prediction factors. n (x), where x is the factor collection value corresponding to the osteoporosis prediction factor numbered n, and the selected x is based on the osteoporosis calibration feature continuous function f. n The value obtained by (x) is the corresponding factor quantity value; Based on the continuous function f of the osteoporosis calibration characteristics n (x), perform feature range extraction based on the amount of change to form the initial prediction feature range set of the osteoporosis prediction factor.

4. The osteoporosis prediction method based on an intelligent osteoporosis prediction model according to claim 3, characterized in that, The process involves fitting discrete collected values ​​and corresponding fitted quantity values ​​based on all the factors, performing continuous fitting analysis to form a continuous function f of osteoporosis calibration characteristics corresponding to the osteoporosis prediction factors. n (x), including: In a fitting coordinate system where the discrete collected values ​​of the factors are fitted on the horizontal axis and the number of fitted values ​​are on the vertical axis, all the discrete collected values ​​of the factors and the corresponding number of fitted values ​​are marked in the fitting coordinate system as discrete points. Set a clustering distance threshold α, and perform clustering analysis on all the fitted discrete points in the fitted coordinate system according to the clustering distance threshold α to form clustered fitting points; Based on all the cluster fitting points formed by cluster analysis and the remaining fitting discrete points, a function fitting is performed to form a continuous function f of osteoporosis calibration characteristics corresponding to the osteoporosis predictor factors. n (x).

5. The osteoporosis prediction method based on an intelligent osteoporosis prediction model according to claim 4, characterized in that, The process of setting a clustering distance threshold α and performing clustering analysis on all the fitted discrete points in the fitted coordinate system based on the clustering distance threshold α to form clustered fitted points includes: Extract any of the fitted discrete points and determine them as object discrete points. Within a circular area centered on the object discrete point and with the clustering distance threshold α as the radius, perform the following form of cluster discrete point collection: If there are no other fitted discrete points within the circular range, the object discrete point is determined to be an isolated fitted discrete point, and any fitted discrete point other than the object discrete point is extracted to continue to establish a circular range for cluster analysis. If other fitted discrete points exist within the circular range, these other fitted discrete points are marked as cluster fitted discrete points. Cluster analysis continues to be performed by establishing a circular range with a radius equal to the clustering distance threshold α, centered on each of the cluster fitted discrete points, until no new cluster fitted discrete points are obtained. All the obtained cluster fitted discrete points and the set of object discrete points are combined to form a cluster fitted discrete point set. Furthermore, any fitted discrete points other than those included in the cluster fitted discrete point set are extracted to establish a circular range for cluster analysis. For each of the obtained cluster fitting discrete point sets, perform cluster fitting in the following form to form the corresponding cluster fitting points: Obtain the position information of all fitted discrete points in the cluster fitting discrete point set in the fitting coordinate system, forming a cluster fitting discrete point position information set. Where i is the number of the cluster fitting discrete point set, and k is the number of the fitted discrete point in the cluster fitting discrete point set with number i. Based on the cluster fitting discrete point location information set, determine the corresponding cluster fitting point B. i The cluster fitting point B i Satisfy the total cluster distance The minimum, and the cluster fitting point B i The ordinate position is taken as an integer, where: Total cluster distance S k This represents the distance from the cluster fitting point to the fitted discrete point numbered k.

6. The osteoporosis prediction method based on an intelligent osteoporosis prediction model according to claim 5, characterized in that, The continuous function f based on the osteoporosis calibration characteristics n (x), perform feature range extraction based on the amount of change to form the initial prediction feature range set of the osteoporosis prediction factors, including: Determine the continuous function f of the osteoporosis calibration characteristics. n (x) The minimum boundary value X at the x-coordinate position of the fitted coordinate system. min and maximum boundary value X max And according to the minimum boundary value X min and maximum boundary value X max Determine the average level Y ave ,in: According to the average level Y ave The continuous function f of the osteoporosis calibration feature was determined. n (x) exceeds the average level Y ave The range of all x-axis values ​​forms the initial predictive feature range of the factors. t represents the number of the initial predictive feature range of the osteoporosis predictor factor with number n; Initial prediction feature range based on all the factors mentioned The initial predictive feature range set of the factors is formed.

7. The osteoporosis prediction method based on an intelligent osteoporosis prediction model according to claim 6, characterized in that, The step of extracting sample data labeled as normal bone from the detection and calibration sample data and analyzing its normal predictive feature information to form normal bone predictive feature data includes: Based on the initial osteoporosis prediction factor set, normal collection values ​​of all osteoporosis prediction factors are obtained for each sample object labeled as normal bone quality, forming normal collection values ​​of object factors for different osteoporosis prediction factors under the sample object; For each osteoporosis predictor, the normal collection values ​​of the factor for all sample objects labeled as having normal bone quality are collected to form a factor normal collection value set; For each of the normally collected values ​​of the aforementioned factors, a feature fitting analysis of discrete values ​​is performed to form a continuous function of the normal characteristics of the factors. Based on the continuous function of the normal characteristics of the factors, the range of normal characteristics is extracted to form a set of normal characteristic ranges of the factors; The normal bone density prediction feature data is formed by combining the normal feature range set of all the osteoporosis prediction factors corresponding to the initial osteoporosis prediction feature factor set.

8. The osteoporosis prediction method based on an intelligent osteoporosis prediction model according to claim 7, characterized in that, The step of filtering and analyzing predictive feature information based on the initial osteoporosis prediction feature data and the normal bone quality prediction feature data to establish an osteoporosis prediction model includes: For the same initial osteoporosis prediction feature factors, the initial osteoporosis prediction feature data and the normal bone quality prediction feature data are compared and analyzed to determine the associated osteoporosis prediction feature factors and the prediction feature range data corresponding to the associated osteoporosis prediction factors; For each associated osteoporosis predictor feature, a prediction condition function C is established based on the prediction feature range data corresponding to the associated osteoporosis predictor feature. v (e), where v represents the number of the different associated osteoporosis predictive feature factors, wherein: When the obtained value e of the associated osteoporosis prediction feature factor does not belong to any prediction feature range in the corresponding prediction feature range data, then C v (e) = 0; When the obtained value e of the associated osteoporosis prediction feature factor belongs to any prediction feature range in the corresponding prediction feature range data, then C v (e) = 1; Based on the sample size of the data labeled as osteoporosis, correlation analysis was performed on different associated osteoporosis predictive characteristic factors to determine the correlation factor β corresponding to each associated osteoporosis predictive characteristic factor. v ; According to the prediction condition function C v (e) and the corresponding correlation factor β v The following osteoporosis prediction model was established: P = C1(e)*β1 + C2(e)*β2 + … + C v (e)*β v P represents the predicted probability of osteoporosis.

9. The osteoporosis prediction method based on an intelligent osteoporosis prediction model according to claim 8, characterized in that, For the same initial osteoporosis prediction feature factors, the initial osteoporosis prediction feature data and the normal bone quality prediction feature data are compared and analyzed to determine the associated osteoporosis prediction feature factors and the prediction feature range data corresponding to the associated osteoporosis prediction factors, including: The initial prediction feature range for all factors corresponding to the initial osteoporosis prediction feature factors. The initial predicted feature range of each factor is compared with the normal feature ranges of all factors in the corresponding normal feature range set as follows: If the normal feature range of the factors in the set of normal feature ranges includes the initial prediction feature range of each factor corresponding to the initial osteoporosis prediction feature factor. Then the initial osteoporosis prediction feature factors are determined to be non-correlated prediction factors; If the normal feature range of the factors in the set of normal feature ranges of the factors corresponds to the initial prediction feature range of each factor corresponding to the initial osteoporosis prediction feature factor... If there is no intersection between them, then the initial osteoporosis prediction feature factor is determined as the associated osteoporosis prediction feature factor, and the initial prediction feature range of all the factors is determined as the corresponding prediction feature range; If the normal feature range of the factors in the set of normal feature ranges contains a portion of the initial osteoporosis prediction feature range, then the initial prediction feature range of the factors... Then the initial osteoporosis prediction feature factor is determined to be the associated osteoporosis prediction feature factor, and the initial prediction feature range of the factor that does not have an inclusion relationship is determined to be the corresponding prediction feature range; If there exists a normal characteristic range of the factor in the set of normal characteristic ranges and the initial prediction characteristic range of the factor corresponding to the initial osteoporosis prediction characteristic factor. If there is an intersection, the initial osteoporosis prediction feature factor is determined to be a related osteoporosis prediction feature factor, and the range formed after excluding the intersection part of the initial prediction feature range of the factor is determined as the corresponding prediction feature range. By combining all the predicted feature ranges under the aforementioned associated osteoporosis prediction feature factors, the corresponding predicted feature range data is formed.

10. The osteoporosis prediction method based on an intelligent osteoporosis prediction model according to claim 9, characterized in that, Based on the sample size of the data labeled as osteoporosis, correlation analysis is performed on different associated osteoporosis predictive characteristic factors to determine the correlation factor β corresponding to each associated osteoporosis predictive characteristic factor. v ,include: For all sample objects labeled as osteoporosis, determine the number of sample objects whose collected values ​​of the associated osteoporosis predictive feature factors belong to the predicted feature range data for each associated osteoporosis predictive feature factor, and obtain the associated sample frequency W for each associated osteoporosis predictive feature factor. v ,in: Q v Q represents the number of sample objects whose acquired associated predictive feature factor values ​​belong to the predictive feature range data corresponding to the associated osteoporosis predictive feature factor numbered v. all This represents the total number of sample subjects categorized as having osteoporosis. Based on the frequency W of each of the associated samples v Determine the corresponding correlation factor β v ,in:

Images