Intervertebral disc degenerative change analysis system based on multi-dimensional information

Abstract

The invention relates to an intervertebral disc degenerative change analysis system based on multi-dimensional information. The system comprises a data processing module, a modifiable factor processing module, a circular RNA data processing module, a biomarker processing module, an image processing module, a treatment information processing module, a diagnostic feature anchoring module, a model generation module and a change analysis module. Anchoring is carried out on circular RNA, biomarkers, treatment information and modifiable factors through image information to form an anchoring network, and a reliable and effective basis can be provided for disease course analysis through big data and a large number of research results, so that each doctor has a targeted expectation and understanding on the disease course of a patient in the diagnosis stage; the method has the advantages that an interventional therapy scheme can be arranged in advance or the condition of a patient can be observed in real time, research results can be collected and fed back in time through big data, the information utilization rate is increased, and an information dimension integration scheme for degenerative disease recognition is created, and the accuracy of recognition results is improved.

Description

Technical Field

[0001] The present invention relates to the field of medical analysis technology, and more specifically, to an intervertebral disc degenerative lesion analysis system based on multi-dimensional information. Background Art

[0002] Intervertebral Disc Degeneration (IDD) is a common degenerative disease that affects 266 million people worldwide each year. Characteristics of IDD include accelerated degradation and abnormal synthesis of the extracellular matrix (ECM), reduced water content, reduced disc height, and decreased ability to absorb external mechanical compression. IDD is the main cause of chronic low back pain and spine-related diseases, and has imposed a significant economic and social burden on the world. It has been previously reported that 84% of the world's population will experience low back pain in their lifetime, and 10% of them will suffer from long-term disability. However, current treatment strategies for IDD are mainly limited to symptom intervention, and there is a lack of drugs that can change the course of the disease, so the clinical treatment of IDD and related diseases is severely limited. Therefore, determining the pathophysiological events and molecular mechanisms of IDD is crucial for the development of new treatments.

[0003] At present, the main diagnostic method for intervertebral disc degeneration is through examination. Intervertebral disc degeneration is identified through image analysis, laboratory tests, and mechanical tests to determine the lesion conclusion. However, since the course of the disease is very long at this stage, studies have shown that there are many factors that affect intervertebral disc degeneration. If the degenerative lesions are treated based solely on the diagnostic results, the treatment methods or methods of alleviating the disease are not targeted for individual patients. For different patients, the course, form, and characteristics of degenerative lesions are different, so advance planning of treatment should be adopted. However, this method lacks information sources for collecting and analyzing patient information. For example, the topic studied by the inventors - Causal associations between modifiable risk factors and intervertebral disc degeneration, Wei Guo, PhD, b, c, Bao-Li Li, PhDc, Jian-Yong Zhao, PhD, b, Xiao-Ming Li, PhD, b, Lin-Feng Wang, PhD disclosed the relationship between 33 modifiable risk factors and intervertebral disc degeneration, and Circular RNA GRB10 as a competitive endogenous RNA regulating nucleus pulposus cells death indegenerative intervertebral disk and Circular RNA derived from TIMP2 functions as a competitive endogenous RNA and regulates intervertebral disc degeneration by targeting miR-185-5p and matrix metalloproteinase 2 disclosed the relationship between the action of circular RNA and the mechanism of action of circular RNA and the inhibition of degenerative lesions, and Gene expression profile identifies potential biomarkers for human intervertebral disc Degeneration disclosed the potential biomarkers of human intervertebral disc degeneration identified by gene expression profiles. Therefore, the information in the above multiple dimensions is closely related to degenerative lesions, so a data processing system is needed to analyze degenerative lesions by integrating the above information. Summary of the invention

[0004] In view of this, an object of the present invention is to provide an intervertebral disc degenerative lesion analysis system based on multi-dimensional information.

[0005] In order to solve the above technical problems, the technical solution of the present invention is: a system for analyzing intervertebral disc degeneration based on multidimensional information, a data processing module, a modifiable factor processing module, a circular RNA data processing module, a biomarker processing module, an image processing module, a treatment information processing module, a diagnostic feature anchoring module, a model generation module and a lesion analysis module;

[0006] The data processing module is used to process historical disease course data and extract historical factor data, historical marker data, intervertebral disc image data and treatment information data from each historical disease course data;

[0007] The modifiable factor processing module is used to obtain historical factor data, extract corresponding modifiable factor features from the historical factor data, and associate historical representation features with historical course data based on the extracted modifiable factor features to generate a modifiable factor association link;

[0008] The circular RNA data processing module is used to obtain the inhibition characterization feature corresponding to each circular RNA data from a pre-constructed circular RNA association database, and index the corresponding historical characterization feature from the corresponding historical course data according to the inhibition characterization feature, calculate the inhibition influence deviation, and correct the corresponding inhibition characterization feature;

[0009] The biomarker processing module is used to obtain biomarker data, extract corresponding biomarker features from historical marker data, and associate historical characterization features from historical disease course data based on the extracted biomarker features to generate a marker association link;

[0010] The image processing module is used to process the intervertebral disc image data, and process each intervertebral disc image data into a lesion representation item through a pre-built cluster analysis model, and generate an index pointer of the lesion representation item according to the historical representation feature;

[0011] The treatment information processing module is used to process the treatment information data, and extract the treatment elements in the treatment information feature data through a preset common element processing strategy, and calculate the effect of the historical representation characteristics corresponding to each type of treatment element to generate element influencing variables, and establish an influencing variable link according to the element influencing variables;

[0012] The diagnostic feature anchoring module establishes an anchoring network according to the index pointer, the starting node of the anchoring network is the lesion representation item, and adds the modifiable factor association link, the inhibition representation feature, the marker association link, and the influencing variable link to the anchoring network;

[0013] The model generation module is configured with a preset dynamic anchor configuration strategy, which is used to configure the diagnostic identification conditions and critical mapping functions of the anchor nodes corresponding to each index pointer in the anchor network, and configure the transmission time probability mapping function of the connection between the nodes to generate a diagnostic model;

[0014] The lesion analysis module inputs the patient characterization information into the diagnostic model to obtain the corresponding current critical value and the corresponding disease course map.

[0015] Further: the modifiable factor processing module includes a factor semantic library, a report analysis unit, a factor feature extraction unit and a factor influence association unit, the factor semantic library is configured with a number of modifiable factor data, each modifiable factor data includes a number of different factor feature words, the report analysis unit retrieves an external analysis report through the factor semantic library to obtain the factor influence representation corresponding to each modifiable factor item and the solid factor influence data corresponding to the factor influence representation, the factor feature extraction unit determines the corresponding modifiable factor item from each historical factor data through the factor semantic library, as well as the factor degree value and factor trust value corresponding to the modifiable factor item to generate the modifiable factor feature; the factor influence association unit obtains the actual impact situation from the historical medical course data through the impact representation corresponding to the modifiable factor item to generate dynamic factor influence data, and generates the model factor data of the associated historical representation feature based on the dynamic factor influence data and the solid factor influence data, and generates the modifiable factor association link based on the model factor data.

[0016] Further: the model factor data includes a factor induction rate value and a factor induction process value, and the expression of the factor induction rate value is:

[0017]

[0018]

[0019] u de =N1 / N de -u D ,

[0020] Among them, U re is the factor induction rate value corresponding to a modifiable factor feature and a historical representation feature, is the preset solid weight value, is the preset static weight value, u ge is the solid-state factor induction rate value in the solid-state factor impact data, r ie is the report confidence value corresponding to the i-th analysis report, u ieis the factor report induction rate corresponding to the i-th analysis report, I1 is the total number of analysis reports that record the modifiable factor characteristics, and n fe is the configuration weight corresponding to the historical representation feature, N ie is the sum of the configuration weights of all historical characterization features mentioned in the i-th analysis report, u de is the dynamic factor induction rate in the dynamic factor impact data, N1 is the number of times the modifiable factor feature causes the corresponding historical representation feature in the historical course data, and N de is the total number of all historical representation features caused by the modifiable factor feature in the historical course instrument, u D is the baseline induction rate value of the historical characterization feature;

[0021] The expression of the factor-induced process value is:

[0022]

[0023]

[0024]

[0025] Among them, S re is the factor induction value corresponding to a modifiable factor feature and a historical representation feature, s ge is the solid-state factor induced value in the solid-state factor impact data, w ie is the factor report impact deviation corresponding to the i-th analysis report, I2 is the total number of analysis reports that record the characteristics of the modifiable factor, and f re () is the time deviation process value mapping function, which reflects the mapping relationship between the time deviation of the factor and the process value impact value, t ie is the actual duration of occurrence of the historical characterization feature recorded in the i-th analysis report, t E is the preset benchmark occurrence duration corresponding to the historical representation feature, s de is the dynamic factor induced value in the dynamic factor impact data, w jr The historical influence degree deviation of the factor corresponding to the historical characterization feature recorded in the jth historical course data, t je It is the actual occurrence duration of the historical characterization feature recorded in the j-th historical medical course data.

[0026] Furthermore: the modifiable factor processing module also includes an item class generation unit, the item class generation unit is configured with a basic item class library and an item class splitting condition, the basic item class library stores a number of modifiable factor items, each modifiable factor item is indexed by a factor feature word, the item class generation unit obtains the number of factor feature words in the analysis report and historical medical data, and determines whether the number of factor feature words meets the corresponding item class splitting condition, and when the item class splitting condition is met, the modifiable factor item is split into two new modifiable factor items.

[0027] Further: the circular RNA data processing module includes an associated suppression retrieval unit, a suppression deviation calculation unit, and a data correction unit;

[0028] The associated inhibition retrieval unit is used to retrieve the corresponding inhibition characterization feature according to the circular RNA data, wherein the inhibition characterization feature includes an inhibition item and an action inhibition function;

[0029] The inhibition deviation calculation unit obtains the actual inhibition function of each historical disease course data, and calculates the inhibition influence deviation according to the actual inhibition function.

[0030] Among them, V da To suppress the influence deviation, t1 is the suppression start time, t2 is the suppression end time, g cq (t) is the preset suppression weight adjustment function, which reflects the relationship between the suppression weight and the time. cv (t) is the action inhibition function, f ev (t) is the actual suppression function;

[0031] The data correction unit is configured with an inhibition correction expression for generating a new action inhibition function, and the inhibition correction expression is f cv (t) = f c ' v (t)+β qj V da [f c ' v (t)-f e ' v (t)], where f c ' v (t) is the action inhibition function at the previous moment, β qj is the preset correction ratio multiplier, f e ' v (t) is the actual inhibition function at the previous moment.

[0032] Further: the biomarker processing module includes an identification and screening unit, a condition index unit, a link construction unit and a marker information library, the marker information library stores a number of related marker sequences, the identification and screening unit is used to identify the related marker sequences in the historical marker data as biomarker features according to the marker information library, the condition index unit is used to retrieve the corresponding abnormal index conditions from the preset abnormal index condition library for the biomarker features, and screen the corresponding historical representation features through the abnormal index conditions, the link construction unit is used to calculate the marker deviation horizontal coordinate and the marker deviation vertical coordinate corresponding to each related marker sequence, and generate a marker association link based on them, and the relative position relationship between the node corresponding to the biomarker feature and the associated node corresponding to the historical representation feature in the marker association link is generated according to the marker deviation horizontal coordinate and the marker deviation vertical coordinate, Among them, E xu is the horizontal axis of the marker deviation, E yu is the vertical coordinate of the marker deviation, I3 is the total number of historical representation features that meet the abnormal index condition, χ1 is the preset marker degree deviation weight, χ2 is the preset marker time deviation weight, k ie is the difference between the degree value of the historical characterization feature that meets the abnormal index condition and the degree value of the benchmark marker, t iw It is the difference between the representation time of the i-th historical representation feature that meets the abnormal index condition and the representation time of the benchmark marker.

[0033] Further: the image processing module includes a sample generation unit, an image processing unit, and an image association unit. The sample generation unit is configured with a preset semantic recognition model for identifying the diagnostic result features corresponding to each lesion image, and normalizing the diagnostic result features to obtain a simplified result representation item corresponding to the lesion image. The image processing unit obtains the difference parameter data of the lesion image through a preset image analysis algorithm, and obtains the difference range group and the representation result features corresponding to the difference range group through a preset clustering analysis algorithm based on the difference parameter data. Each representation result feature corresponds to at least one historical representation feature; the clustering analysis model is configured to obtain the corresponding image difference parameter through the intervertebral disc image data, and obtain the corresponding image difference identification value and the representation result feature according to the difference range group into which each image difference parameter falls. If the sum of the image difference identification values ​​corresponding to the representation result feature is greater than the preset representation identification benchmark, the corresponding lesion representation item is output according to the representation result feature. The image association unit is used to establish an index pointer of the historical representation feature to be configured on the lesion representation item.

[0034] Further: the common factor processing strategy splits the treatment information feature data into several different treatment factor features, and groups the treatment factor features according to the similarity of the treatment-related information in the historical course data to generate corresponding treatment factors, obtains the historical representation features that meet the treatment factors to calculate the factor influencing variables, and where Y re is the factor influencing variable, z re is the preset estimated impact value, z id is the corresponding factor influence value in the i-th historical medical course data, and I4 is the total number of historical medical course data that meet the treatment factor.

[0035] Furthermore: the diagnostic feature anchoring module is configured with a factor management unit, the factor management unit is configured with a dynamic difference association list, the dynamic difference association list is configured with anchor difference parameters corresponding to different network node type items, and the anchor difference parameters are used to adjust the positional relationship between modifiable factor features, inhibition characterization features, biomarker features and factor influencing variables in the anchoring network.

[0036] Further: the dynamic anchor configuration strategy includes

[0037] Step S1, selecting corresponding diagnostic identification sub-features from a preset diagnostic identification database through lesion characterization items, and configuring identification judgment thresholds to generate diagnostic identification conditions;

[0038] Step S2: establishing a node association link of the starting node in the anchor network, and sequentially acquiring corresponding historical representation features according to the node association link;

[0039] Step S3, obtaining modifiable factor association links, inhibition characteristics, marker association links, and influencing variable links according to historical characterization features, and extracting critical key parameters and transmission time parameters;

[0040] Step S4: construct the criticality mapping function and the transfer time mapping function according to the criticality key parameters and the transfer time parameters.

[0041] The technical effects of the present invention are mainly reflected in the following aspects: through such a setting, circular RNA, biomarkers, treatment information, and modifiable factors are anchored with image information to form an anchor network. On the one hand, big data and a large number of research results can provide a reliable and effective basis for disease course analysis, so that each doctor can have a targeted expectation and understanding of the patient's disease course at the diagnosis stage, so as to arrange interventional treatment plans in advance or observe the patient's condition in real time. On the second hand, big data can be used as a basis for timely collection and feedback of research results to improve information utilization. On the third hand, a solution for integrating information dimensions for the identification of degenerative lesions is created to improve the accuracy of identification results. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 : The system architecture principle diagram of the intervertebral disc degeneration analysis system based on multi-dimensional information of the present invention;

[0043] Figure 2 : Schematic diagram of the modifiable factor processing module architecture of the present invention;

[0044] Figure 3 : Schematic diagram of the architecture of the circular RNA data processing module of the present invention;

[0045] Figure 4 : Schematic diagram of the biomarker processing module of the present invention;

[0046] Figure 5 : Schematic diagram of the architecture of the image processing module of the present invention.

[0047] Attached figure numbers: 100, data processing module; 200, modifiable factor processing module; 201, factor semantic library; 210, report analysis unit; 220, factor feature extraction unit; 230, factor influence association unit; 300, circular RNA data processing module; 310, association inhibition retrieval unit; 320, inhibition deviation calculation unit; 330, data correction unit; 400, biomarker processing module; 410, identification and screening unit; 420, condition indexing unit; 430, link construction unit; 440, marker information library; 500, image processing module; 510, sample generation unit; 520, image processing unit; 530, image association unit; 600, treatment information processing module; 700, diagnostic feature anchoring module; 800, model generation module; 900, lesion analysis module. DETAILED DESCRIPTION

[0048] The specific implementation modes of the present invention are further described below in conjunction with the accompanying drawings to make the technical solutions of the present invention easier to understand and grasp.

[0049] Reference Figure 1As shown, a system for analyzing intervertebral disc degeneration based on multidimensional information includes a data processing module 100, a modifiable factor processing module 200, a circular RNA data processing module 300100, a biomarker processing module 400, an image processing module 500, a treatment information processing module 600, a diagnostic feature anchoring module 700, a model generation module 800, and a lesion analysis module 900; the core of the present invention is to analyze external research articles and historical disease course data to obtain corresponding data, and then normalize the data, quantify the normalized data by quantization rules, and then anchor it in the basic data, so as to generate corresponding analysis results according to the data currently input by the user, and predict the subsequent situation for probability prediction, or analyze intervention factors, and treat in advance. The following is a detailed description of each module of the present invention.

[0050] The data processing module 100 is used to process the historical course data, and extract the historical factor data, historical marker data, intervertebral disc image data and treatment information data from each historical course data; that is, the historical course data is split according to the content, and there will be the following situations: 1. If there is no data of the corresponding type, then the data of this type is left blank as empty data. 2. If there are multiple data of the corresponding content, they are sorted according to time, a data sequence is constructed, and they are processed in sequence, and a timestamp is set. 3. Each data will be associated with the historical representation feature, and the content of the historical representation feature is to identify the corresponding text content through the semantic recognition large model to select the corresponding historical representation feature. The historical representation feature can have the following content to distinguish lumbar disc herniation (rupture of the annulus fibrosus), lumbar disc herniation (protrusion of the nucleus pulposus compresses the nerve), lumbar spinal stenosis, lumbar spondylolisthesis, lumbar instability, discogenic low back pain, facet joint degenerative low back pain, degenerative lumbar scoliosis, lumbar degenerative spondylolisthesis, endplate inflammation, and lumbar osteoporotic fracture. The above categories can also be graded and sub-itemed according to the degree and characteristics of specific historical characteristics. Historical factor data can be smoking (OR = 1.221, 95% CI: 1.068-1.396), drinking (OR = 1.208, 95% CI: 1.056-1.328) and height (OR = 1.149, 95% CI: 1.072-1.231) genetic tendency is associated with an increased risk of IVDD. In addition, education level (OR = 0.573, 95% CI: 0.502-0.654) and family income (OR = 0.614, 95% CI: 0.445-0.847) have a protective effect on IVDD. Insomnia (OR = 1.799, 95% CI: 1.162-2.783), hypertension (OR = 2.113, 95% CI: 1.132-3.944) and type 2 diabetes (OR = 1.069, 95% CI: 1.024-1.115) were three important risk factors causally associated with IVDD. In addition, we also demonstrated that elevated triglyceride levels (OR = 1.080, 95% CI: 1.013-1.151), fasting blood glucose (OR = 1.189, 95% CI: 1.007-1.405) and HbA1c levels (OR = 1.308, 95% CI: 1.017-1.683) were present. The historical marker data can be 93 and 114 DEGs in the NP and AF regions. The treatment information data can be the name of the treatment and the corresponding drug. Through semantic model recognition and extraction, the data can be normalized.

[0051] The modifiable factor processing module 200 is used to obtain historical factor data, and extract the corresponding modifiable factor features from the historical factor data, and associate the historical representation features from the historical course data based on the extracted modifiable factor features to generate a modifiable factor association link; the modifiable factor association link is a connection link between the historical representation features and the modifiable factors. This link is not a simple connection. First, the features and representations are associated by extracting the modifiable factor features. Since each patient with a modifiable factor in the historical data has a corresponding course of disease feature, these historical representation features are associated as associated items of the modifiable factor feature. The screenshot scheme is as follows:

[0052] The modifiable factor processing module 200 includes a factor semantic library 201, a report analysis unit 210, a factor feature extraction unit 220 and a factor impact association unit 230.

[0053] The factor semantic library 201 is configured with a number of modifiable factor data, each of which includes a number of different factor feature words. The modifiable factor data are first divided by the corresponding degree of words. For example, smoking conditions also have different degrees, and different degrees may have different impacts. Therefore, it is necessary to establish this module to quickly determine the degree words, so as to quickly identify which specific unified degree the corresponding factor feature words appear to belong to.

[0054] The report analysis unit 210 retrieves the external analysis report through the factor semantic library 201 to obtain the factor influence representation corresponding to each modifiable factor item and the solid factor influence data corresponding to the factor influence representation. The solid factor influence data corresponding to each factor influence representation in theory can be retrieved through the analysis report. The solid factor is the corresponding relationship obtained through the analysis report. For example, the analysis of whether the corresponding degree of a modifiable factor will aggravate a certain condition.

[0055] The factor feature extraction unit 220 determines the corresponding modifiable factor item from each historical factor data through the factor semantic library 201, as well as the factor degree value and factor trust value corresponding to the modifiable factor item to generate the modifiable factor feature; that is, the content of the corresponding modifiable factor, that is, the modifiable factor item, is matched through the corresponding semantic library, and then the corresponding degree value and trust value can generate the modifiable factor feature.

[0056] The factor impact association unit 230 obtains the actual impact from the historical course data through the impact representation corresponding to the modifiable factor item to generate dynamic factor impact data, and generates the model factor data of the associated historical representation feature according to the dynamic factor impact data and the solid factor impact data, and generates the modifiable factor association link according to the model factor data. Then each corresponding modifiable factor can determine the actual patient condition from the corresponding historical data, and the corresponding dynamic factor impact data can be determined through the same analysis, so as to obtain the comprehensive impact of the modifiable factor on the condition, thereby marking the corresponding modifiable factor link.

[0057] Specifically, the model factor data includes a factor induction rate value and a factor induction process value, and the expression of the factor induction rate value is:

[0058]

[0059]

[0060] u de =N1 / N de -u D ,

[0061] Among them, U re is the factor induction rate value corresponding to a modifiable factor feature and a historical representation feature, reflecting the probability that the modifiable factor feature induces the historical representation feature. is the preset solid weight value, is the preset static weight value, u ge is the solid-state factor induction rate value in the solid-state factor impact data, that is, the probability that the modifiable factor feature in the analysis report induces the historical representation feature, r ie is the report confidence value corresponding to the i-th analysis report, which reflects the credibility of the analysis report. The confidence value is the credibility and evaluation of the report through the dimensions of the analysis report source, researcher, etc., u ie is the factor report induction rate corresponding to the i-th analysis report, reflecting the probability that the modifiable factor feature recorded in the analysis report induces the historical representation feature, I1 is the total number of analysis reports that record the modifiable factor feature, n fe is the configuration weight corresponding to the historical representation feature. The weight corresponding to each historical representation feature is different and can be obtained by looking up the table. N ie is the sum of the configuration weights of all historical characterization features mentioned in the i-th analysis report, u de is the dynamic factor induction rate in the dynamic factor impact data, reflecting the probability that the modifiable factor feature in the historical data induces the historical representation feature. N1 is the number of times the modifiable factor feature causes the corresponding historical representation feature in the historical course data. N deis the total number of all historical representation features caused by the modifiable factor feature in the historical course instrument, u D is the baseline induction rate value of the historical characterization feature; this value is a preset value.

[0062] The expression of the factor-induced process value is:

[0063]

[0064]

[0065]

[0066] Among them, S re The factor-induced process value corresponding to a modifiable factor feature and a historical representation feature. In addition to the probability, a process value is also set. The concept of the process value is to reflect that the factor may change the corresponding disease occurrence. ge is the solid-state factor induced course value in the solid-state factor impact data. The solid-state factor induced course value reflects the accelerating effect of the corresponding solid-state factor on the course of the disease. ie is the factor report influence degree deviation corresponding to the i-th analysis report, that is, the deviation of the results caused by other factors that affect the actual course of the disease in the analysis report, I2 is the total number of analysis reports that record the characteristics of the modifiable factor, and f re () is the time deviation process value mapping function, which reflects the mapping relationship between the time deviation of the factor and the process value impact value. Different time periods correspond to different impact values. For example, the ratio of accelerating the course of disease by one day to accelerating the course of disease by one month is not a proportional relationship, but a different functional relationship based on the impact of the length of time. ie is the actual duration of occurrence of the historical characterization feature recorded in the i-th analysis report, t E is the preset benchmark occurrence duration corresponding to the historical representation feature, s de is the dynamic factor induced course value in the dynamic factor impact data, that is, it represents the average situation of the changes in the modifiable factors and the corresponding course of disease recorded in the historical data, w jr The historical influence degree deviation of the factor corresponding to the historical characterization feature recorded in the jth historical course data, t je It is the actual occurrence duration of the historical characterization feature recorded in the j-th historical medical course data.

[0067] The modifiable factor processing module 200 also includes an item class generation unit, which is configured with a basic item class library and an item class splitting condition. The basic item class library stores a number of modifiable factor items, each modifiable factor item is indexed by a factor feature word, and the item class generation unit obtains the number of factor feature words in the analysis report and the historical course data, and determines whether the number of factor feature words meets the corresponding item class splitting condition. When the item class splitting condition is met, the modifiable factor item is split into two new modifiable factor items. Through the setting of the item class generation unit and the item class splitting condition, a new modifiable factor item can be formed by splitting the item class. For example, smoking will be divided according to different frequency ranges, but if the number is too large within a divided range and the item class splitting condition is met, then it will be split again according to this condition to form a new modifiable factor item.

[0068] The circular RNA data processing module 300100 is used to obtain the inhibitory characterization features corresponding to each circular RNA data from the pre-constructed circular RNA association database, and index the corresponding historical characterization features from the corresponding historical course data according to the inhibitory characterization features, calculate the inhibitory influence deviation, and correct the corresponding inhibitory characterization features; this is mainly to analyze the impact of circular RNA on the disease, mainly through the inhibitory characterization features to establish a connection, the inhibitory characterization features are the inhibition of the secretion of a certain pheromone, thereby affecting the course of the disease, specifically through the following scheme. The circular RNA data processing module 300100 includes an associated inhibition retrieval unit 310, an inhibition deviation calculation unit 320, and a data correction unit 330;

[0069] The associated inhibition retrieval unit 310 is used to retrieve the corresponding inhibition characterization feature according to the circular RNA data, wherein the inhibition characterization feature includes an inhibition item and an action inhibition function, such as the type of the inhibiting pheromone. The action inhibition function represents a function of how the inhibition amount theoretically changes over time.

[0070] The inhibition deviation calculation unit 320 obtains the actual inhibition function of each historical disease course data, and calculates the inhibition influence deviation according to the actual inhibition function,

[0071] Among them, V da To suppress the influence deviation, t1 is the suppression start time, t2 is the suppression end time, g cq (t) is the preset inhibition weight adjustment function. According to this change, the corresponding weight change relationship can be generated. Because different inhibition times have different actual effects, a suppression weight adjustment function is constructed to calculate the suppression influence deviation, reflecting the relationship between the suppression weight and the time. cv (t) is the action inhibition function, f ev(t) is the actual inhibition function; the actual inhibition function represents the function of the actual inhibition amount changing with time.

[0072] The data correction unit is configured with an inhibition correction expression for generating a new action inhibition function, and the inhibition correction expression is f cv (t) = f c ' v (t)+β qj V da [f c ' v (t)-f e ' v (t)], where f c ' v (t) is the action inhibition function at the previous moment, β qj is the preset correction ratio multiplier, f e ' v (t) is the actual inhibition function at the previous moment. By continuously correcting the inhibition function through actual historical data, the expression function of the corresponding circular RNA impact relationship can be obtained and associated with the corresponding representation situation.

[0073] The biomarker processing module is used to obtain biomarker data, extract corresponding biomarker features from historical marker data, and associate historical characterization features from historical disease course data based on the extracted biomarker features to generate a marker association link; the biomarker is specifically a gene fragment, and the corresponding gene fragment is known to affect the course and expression of degenerative lesions, so the biological characterization features are extracted and processed.

[0074] The biomarker processing module includes an identification and screening unit 410, a condition index unit 420, a link construction unit 430 and a marker information library 440. The marker information library 440 stores a number of related marker sequences. The identification and screening unit 410 is used to identify the related marker sequences in the historical marker data as biomarker features according to the marker information library 440. The condition index unit 420 is used to retrieve the corresponding abnormal index conditions from the preset abnormal index condition library for the biomarker features, and screen the corresponding historical representation features through the abnormal index conditions. The link construction unit 430 is used to calculate the marker deviation abscissa and marker deviation ordinate corresponding to each related marker sequence, so as to generate a marker association link based on them. The relative position relationship between the node corresponding to the biomarker feature and the associated node corresponding to the historical representation feature in the marker association link is generated according to the marker deviation abscissa and marker deviation ordinate. Among them, E xu is the horizontal axis of the marker deviation, E yuis the vertical coordinate of the marker deviation, I3 is the total number of historical representation features that meet the abnormal index condition, χ1 is the preset marker degree deviation weight, χ2 is the preset marker time deviation weight, k ie is the difference between the degree value of the historical characterization feature that meets the abnormal index condition and the degree value of the benchmark marker, t iw is the difference between the representation time of the i-th historical representation feature that meets the abnormal index condition and the representation time of the benchmark marker. By taking the influence of time as the horizontal axis deviation and the influence of degree as the vertical axis deviation, the corresponding deviation relationship can be mathematically restored.

[0075] The image processing module 500 is used to process intervertebral disc image data, and processes each intervertebral disc image data into a lesion representation item through a pre-constructed cluster analysis model, and generates an index pointer of the lesion representation item according to the historical representation characteristics; first, the intervertebral disc image data is extracted according to the scale and the corresponding size characteristics, and the image is digitized and formatted to reflect a certain disease condition through a standard size format, so that the newly generated image can be quickly classified for disease condition analysis, thereby judging the current disease condition.

[0076] The image processing module 500 includes a sample generating unit 510, an image processing unit 520, and an image associating unit 530.

[0077] The sample generation unit 510 is configured with a preset semantic recognition model for identifying the diagnostic result features corresponding to each lesion image, and normalizing the diagnostic result features to obtain a simplified result representation item corresponding to the lesion image. First, the sample generation unit 510 is used to generate corresponding samples, associate the diagnostic results with identifiable features in the image, and extract the key image area. Then, after normalization, the data basis of the image as a diagnostic result can be obtained, and this data basis, that is, the simplified result representation item, is used as the basis for identifying the disease through the image.

[0078] The image processing unit 520 obtains difference parameter data of the lesion image through a preset image analysis algorithm, and obtains a difference range group and a characterization result feature corresponding to the difference range group through a preset cluster analysis algorithm based on the difference parameter data, and each characterization result feature corresponds to at least one historical characterization feature; the cluster analysis model is configured to obtain corresponding image difference parameters through intervertebral disc image data, and the difference parameters, that is, reflect the corresponding size relationship by generating image difference parameters through the marked size and the corresponding size relationship, and obtain the corresponding image difference identification value and the characterization result feature according to the difference range group into which each image difference parameter falls. If the sum of the image difference identification values ​​corresponding to the characterization result feature is greater than the preset characterization identification benchmark, the corresponding lesion characterization item is output according to the characterization result feature. The purpose of the image processing unit 520 is to establish a difference range group and obtain the corresponding lesion characterization item through cluster analysis.

[0079] The image association unit 530 is used to establish an index pointer of the historical representation feature to be configured in the lesion representation item.

[0080] The treatment information processing module 600 is used to process the treatment information data, and extract the treatment elements in the treatment information feature data through a pre-set common element processing strategy, and calculate the effect of the historical representation characteristics corresponding to each type of treatment element to generate the element influencing variable, and establish the influencing variable link according to the element influencing variable; the common element processing strategy splits the treatment information feature data into a number of different treatment element characteristics, and groups the treatment element characteristics according to the similarity of the treatment-related information in the historical course data to generate the corresponding treatment element, obtains the historical representation characteristics that meet the treatment element to calculate the element influencing variable, and where Y re is the factor influencing variable, z re is the preset estimated impact value, z id is the corresponding factor influence value in the i-th historical course data, and I4 is the total number of historical course data that meet the treatment factor. The impact of treatment on the outcome is calculated by adding the estimated impact and the actual impact to obtain the corresponding factor influence value, thereby obtaining the relationship between the treatment factor and the historical course data.

[0081] The diagnostic feature anchoring module 700 establishes an anchoring network according to the index pointer, the starting node of the anchoring network is the lesion representation item, and modifiable factor association links, inhibition representation features, marker association links, and influencing variable links are added to the anchoring network; the diagnostic feature anchoring module 700 is configured with a factor management unit, the factor management unit is configured with a dynamic difference association list, the dynamic difference association list is configured with anchor difference parameters corresponding to different network node type items, and the anchor difference parameters are used to adjust the positional relationship between modifiable factor features, inhibition representation features, biomarker features, and element influencing variables in the anchoring network. In this way, first of all, all information can be associated with historical representation features, and the association method is digitized and quantified, and the historical representation features are connected through the development of the course of disease, so it is very convenient to calculate the possible probability influence, time influence, and degree influence of each course of disease, or directly add intervention factors to check the influence of adding a certain treatment method on the above relationship, so as to obtain more accurate diagnosis and analysis results.

[0082] The model generation module 800 is configured with a preset dynamic anchor configuration strategy, which is used to configure the diagnostic identification conditions and critical mapping functions of the anchor nodes corresponding to each index pointer in the anchor network, and configure the transmission time probability mapping function of the connection between the nodes to generate a diagnostic model;

[0083] The dynamic anchor configuration strategy includes

[0084] Step S1, using the lesion representation item to obtain the corresponding diagnostic identification sub-feature from the preset diagnostic identification database, and configuring the identification judgment threshold to generate a diagnostic identification condition; the diagnostic identification sub-feature is indexed by the lesion representation item, and the corresponding identification judgment threshold will be retrieved to generate the corresponding diagnostic identification condition. When each input information including image information is matched, an identification sub-value will be assigned. When the sum of the identification sub-values ​​is greater than the identification judgment threshold, the diagnostic identification condition is deemed to be met.

[0085] Step S2: Establish a node association link of the starting node in the anchor network, and obtain the corresponding historical representation features in sequence according to the node association link; this is a link established according to the course of the disease.

[0086] Step S3, according to the historical characterization features, obtain the modifiable factor association link, the inhibition characterization feature, the marker association link, the influencing variable link and extract the critical key parameters and the transfer time parameters; since the data relationship is known, the data value corresponding to each node can extract the corresponding critical key parameters and the transfer time parameters. The critical key parameters refer to all data related to the degree value, and the transfer time parameters refer to the data related to the time value. For example, when the modifiable factors are known, the circular RNA intervention is known, and the treatment information is known, the critical key parameters and the transfer time parameters corresponding to these three information features can be obtained.

[0087] Step S4: construct the critical mapping function and the transfer time mapping function according to the critical key parameters and the transfer time parameters. The critical key parameters and the transfer time parameters, the critical mapping function and the transfer time function are pre-designed and can be changed according to the needs. For example, the critical mapping function can weight the critical impact values ​​of different factors, and the transfer time function can be a linear change function of probability and average impact time.

[0088] The lesion analysis module 900 inputs the patient characterization information into the diagnostic model to obtain the corresponding current critical value and the corresponding disease course map. The map can accurately reflect the degree of each predicted disease, the probability of occurrence, and the predicted time of occurrence, giving doctors and patients sufficient advice to facilitate early intervention treatment.

[0089] Of course, the above are only typical examples of the present invention. In addition, the present invention may also have many other specific implementations. All technical solutions formed by equivalent replacement or equivalent transformation fall within the scope of protection required by the present invention.

Claims

1. A system for analyzing intervertebral disc degeneration based on multidimensional information, characterized in that: Data processing module, modifiable factor processing module, circular RNA data processing module, biomarker processing module, image processing module, treatment information processing module, diagnostic feature anchoring module, model generation module and lesion analysis module; The data processing module is used to process historical disease course data and extract historical factor data, historical marker data, intervertebral disc image data and treatment information data from each historical disease course data; The modifiable factor processing module is used to obtain historical factor data, extract corresponding modifiable factor features from the historical factor data, and associate historical representation features with historical course data based on the extracted modifiable factor features to generate a modifiable factor association link; The circular RNA data processing module is used to obtain the inhibition characterization feature corresponding to each circular RNA data from a pre-constructed circular RNA association database, and index the corresponding historical characterization feature from the corresponding historical course data according to the inhibition characterization feature, calculate the inhibition influence deviation, and correct the corresponding inhibition characterization feature; The biomarker processing module is used to obtain biomarker data, extract corresponding biomarker features from historical marker data, and associate historical characterization features from historical disease course data based on the extracted biomarker features to generate a marker association link; The image processing module is used to process the intervertebral disc image data, and process each intervertebral disc image data into a lesion representation item through a pre-built cluster analysis model, and generate an index pointer of the lesion representation item according to the historical representation feature; The treatment information processing module is used to process the treatment information data, and extract the treatment elements in the treatment information feature data through a preset common element processing strategy, and calculate the effect of the historical representation characteristics corresponding to each type of treatment element to generate element influencing variables, and establish an influencing variable link according to the element influencing variables; The diagnostic feature anchoring module establishes an anchoring network according to the index pointer, the starting node of the anchoring network is the lesion representation item, and adds the modifiable factor association link, the inhibition representation feature, the marker association link, and the influencing variable link to the anchoring network; The model generation module is configured with a preset dynamic anchor configuration strategy, which is used to configure the diagnostic identification conditions and critical mapping functions of the anchor nodes corresponding to each index pointer in the anchor network, and configure the transmission time probability mapping function of the connection between the nodes to generate a diagnostic model; The lesion analysis module inputs the patient characterization information into the diagnostic model to obtain the corresponding current critical value and the corresponding disease course map.

2. The intervertebral disc degeneration analysis system based on multidimensional information as claimed in claim 1, characterized in that: The modifiable factor processing module includes a factor semantic library, a report analysis unit, a factor feature extraction unit and a factor influence association unit. The factor semantic library is configured with a plurality of modifiable factor data, each modifiable factor data includes a plurality of different factor feature words. The report analysis unit retrieves an external analysis report through the factor semantic library to obtain a factor influence representation corresponding to each modifiable factor item and solid factor influence data corresponding to the factor influence representation. The factor feature extraction unit determines a corresponding modifiable factor item from each historical factor data through the factor semantic library, and a factor degree value and a factor trust value corresponding to the modifiable factor item to generate the modifiable factor feature. The factor impact association unit obtains the actual impact situation from the historical medical course data through the impact representation corresponding to the modifiable factor item to generate dynamic factor impact data, and generates model factor data of the associated historical representation characteristics based on the dynamic factor impact data and the solid factor impact data, and generates a modifiable factor association link based on the model factor data.

3. The intervertebral disc degeneration analysis system based on multidimensional information as claimed in claim 2, characterized in that: The model factor data includes a factor induction rate value and a factor induction process value. The expression of the factor induction rate value is: u de =N1 / N de -u D , Among them, U re is the factor induction rate value corresponding to a modifiable factor feature and a historical representation feature, is the preset solid weight value, is the preset static weight value, u ge is the solid-state factor induction rate value in the solid-state factor impact data, r ie is the report confidence value corresponding to the i-th analysis report, u ie is the factor report induction rate corresponding to the i-th analysis report, I1 is the total number of analysis reports that record the modifiable factor characteristics, and n fe is the configuration weight corresponding to the historical representation feature, N ie is the sum of the configuration weights of all historical characterization features mentioned in the i-th analysis report, u de is the dynamic factor induction rate in the dynamic factor impact data, N1 is the number of times the modifiable factor feature causes the corresponding historical representation feature in the historical course data, and N de is the total number of all historical representation features caused by the modifiable factor feature in the historical course instrument, u D is the baseline induction rate value of the historical characterization feature; The expression of the factor-induced process value is: Among them, S re is the factor induction value corresponding to a modifiable factor feature and a historical representation feature, s ge is the solid-state factor induced value in the solid-state factor impact data, w ie is the factor report impact deviation corresponding to the i-th analysis report, I2 is the total number of analysis reports that record the characteristics of the modifiable factor, and f re () is the time deviation process value mapping function, which reflects the mapping relationship between the time deviation of the factor and the process value impact value, t ie is the actual duration of occurrence of the historical characterization feature recorded in the i-th analysis report, t E is the preset benchmark occurrence duration corresponding to the historical representation feature, s de is the dynamic factor induced value in the dynamic factor impact data, w jr The historical influence degree deviation of the factor corresponding to the historical characterization feature recorded in the jth historical course data, t je It is the actual occurrence duration of the historical characterization feature recorded in the j-th historical medical course data.

4. The intervertebral disc degeneration analysis system based on multidimensional information as claimed in claim 2, characterized in that: The modifiable factor processing module also includes an item class generation unit, which is configured with a basic item class library and an item class splitting condition. The basic item class library stores a number of modifiable factor items, each modifiable factor item is indexed by a factor feature word, and the item class generation unit obtains the number of factor feature words in the analysis report and historical medical data, and determines whether the number of factor feature words meets the corresponding item class splitting condition. When the item class splitting condition is met, the modifiable factor item is split into two new modifiable factor items.

5. The intervertebral disc degeneration analysis system based on multidimensional information as claimed in claim 1, characterized in that: The circular RNA data processing module includes an associated inhibition retrieval unit, an inhibition deviation calculation unit, and a data correction unit; The associated inhibition retrieval unit is used to retrieve the corresponding inhibition characterization feature according to the circular RNA data, wherein the inhibition characterization feature includes an inhibition item and an action inhibition function; The inhibition deviation calculation unit obtains the actual inhibition function of each historical disease course data, and calculates the inhibition influence deviation according to the actual inhibition function. Among them, V da To suppress the influence deviation, t1 is the suppression start time, t2 is the suppression end time, g cq (t) is the preset suppression weight adjustment function, which reflects the relationship between the suppression weight and the time. cv (t) is the action inhibition function, f ev (t) is the actual suppression function; The data correction unit is configured with an inhibition correction expression for generating a new action inhibition function, and the inhibition correction expression is f cv (t) = f c ' v (t)+β qj V da [f c ' v (t)-f e ' v (t)], where f c ' v (t) is the action inhibition function at the previous moment, β qj is the preset correction ratio multiplier, f e ' v (t) is the actual inhibition function at the previous moment.

6. The intervertebral disc degeneration analysis system based on multidimensional information as claimed in claim 1, characterized in that: The biomarker processing module includes an identification and screening unit, a condition index unit, a link construction unit and a marker information library, wherein the marker information library stores a number of related marker sequences, the identification and screening unit is used to identify the related marker sequences in the historical marker data as biomarker features according to the marker information library, the condition index unit is used to retrieve the corresponding abnormal index conditions from the preset abnormal index condition library for the biomarker features, and screen the corresponding historical representation features through the abnormal index conditions, the link construction unit is used to calculate the marker deviation abscissa and marker deviation ordinate corresponding to each related marker sequence, and generate a marker association link based on them, and the relative position relationship between the node corresponding to the biomarker feature and the associated node corresponding to the historical representation feature in the marker association link is generated according to the marker deviation abscissa and marker deviation ordinate, Among them, E xu is the horizontal axis of the marker deviation, E yu is the vertical coordinate of the marker deviation, I3 is the total number of historical representation features that meet the abnormal index condition, χ1 is the preset marker degree deviation weight, χ2 is the preset marker time deviation weight, k ie is the difference between the degree value of the historical characterization feature that meets the abnormal index condition and the degree value of the benchmark marker, t iw It is the difference between the representation time of the i-th historical representation feature that meets the abnormal index condition and the representation time of the benchmark marker.

7. The intervertebral disc degeneration analysis system based on multidimensional information as claimed in claim 1, characterized in that: The image processing module includes a sample generation unit, an image processing unit, and an image association unit. The sample generation unit is configured with a preset semantic recognition model for identifying the diagnostic result features corresponding to each lesion image, and normalizing the diagnostic result features to obtain a simplified result representation item corresponding to the lesion image. The image processing unit obtains difference parameter data of the lesion image through a preset image analysis algorithm, and obtains a difference range group and a representation result feature corresponding to the difference range group based on the difference parameter data through a preset cluster analysis algorithm, wherein each representation result feature corresponds to at least one historical representation feature. The cluster analysis model is configured to obtain corresponding image difference parameters through intervertebral disc image data, and obtain corresponding image difference identification values ​​and characterization result features according to the difference range group into which each image difference parameter falls. If the sum of the image difference identification values ​​corresponding to the characterization result features is greater than a preset characterization identification benchmark, the corresponding lesion characterization item is output according to the characterization result features. The image association unit is used to establish an index pointer of the historical characterization feature to be configured in the lesion characterization item.

8. The intervertebral disc degeneration analysis system based on multidimensional information as claimed in claim 1, characterized in that: The common element processing strategy splits the treatment information feature data into several different treatment element features, and groups the treatment element features according to the similarity of the treatment-related information in the historical course data to generate corresponding treatment elements, obtains the historical representation features that meet the treatment elements to calculate the element influencing variables, and where Y re is the factor influencing variable, z re is the preset estimated impact value, z id is the corresponding factor influence value in the i-th historical medical course data, and I4 is the total number of historical medical course data that meet the treatment factor.

9. The intervertebral disc degeneration analysis system based on multidimensional information as claimed in claim 1, characterized in that: The diagnostic feature anchoring module is configured with a factor management unit, the factor management unit is configured with a dynamic difference association linked list, the dynamic difference association linked list is configured with anchor difference parameters corresponding to different network node type items, and the anchor difference parameters are used to adjust the positional relationship between modifiable factor features, inhibition characterization features, biomarker features and factor influencing variables in the anchoring network.

10. The intervertebral disc degeneration analysis system based on multi-dimensional information as claimed in claim 1, characterized in that: The dynamic anchor configuration strategy includes Step S1, selecting corresponding diagnostic identification sub-features from a preset diagnostic identification database through lesion characterization items, and configuring identification judgment thresholds to generate diagnostic identification conditions; Step S2: establishing a node association link of the starting node in the anchor network, and sequentially acquiring corresponding historical representation features according to the node association link; Step S3, obtaining modifiable factor association links, inhibition characteristics, marker association links, and influencing variable links according to historical characterization features, and extracting critical key parameters and transmission time parameters; Step S4: construct the criticality mapping function and the transfer time mapping function according to the criticality key parameters and the transfer time parameters.

Images