Gout computer-aided judgment method and device based on logistic regression and medium
By processing uric acid metabolism fluid samples using a logistic regression-based method, a gout auxiliary diagnosis model was constructed, which solved the problem of insufficient accuracy in gout diagnosis and achieved efficient and accurate diagnosis under multiple indicators, making it suitable for automated auxiliary diagnosis in primary hospitals.
Patent Information
- Application Number
- CN202511515178.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-22
- Publication Date
- 2026-02-13
AI Technical Summary
Current technologies for assisting in the diagnosis of gout lack sufficient accuracy. Diagnosis based on a single indicator is prone to missed diagnoses and misdiagnoses. Furthermore, the lack of objective and unified quantitative standards makes it difficult to widely promote these technologies in primary hospitals.
A logistic regression-based approach was adopted. By processing uric acid metabolism fluid samples using tandem mass spectrometry, a specific metabolite structure matrix was constructed. Elastic network regularization was performed to determine the logistic regression model coefficients. Nonlinear feature analysis of pathological interaction terms was conducted to construct a gout-aided diagnosis model. The probability of gout attacks was calculated by combining the weight configuration of multiple metabolites and the Sigmoid function, and the diagnostic consistency was verified.
It improves the accuracy and efficiency of gout diagnosis, realizes the efficiency of logistic regression weight configuration under multiple indicators, enhances the objectivity and consistency of doctors' diagnoses, and is suitable for automated auxiliary diagnosis in primary hospitals.
Smart Images
Figure CN121528490A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of medical artificial intelligence, and in particular to a gout computer-aided judgment method based on logistic regression, a device and a medium. BACKGROUND
[0002] Gout is a common metabolic disease, and the accuracy and timeliness of its diagnosis are crucial for patient treatment. Currently, clinical diagnosis mainly relies on serum uric acid (SUA) level detection and clinical symptoms. However, many patients with hyperuricemia do not develop gout, and the SUA level of some patients with acute gout may be normal. The diagnosis of gout based on a single indicator and a small number of indicators is prone to insufficient diagnostic accuracy, and misdiagnosis and missed diagnosis may occur.
[0003] Simply relying on doctors to manually interpret multiple indicators is not only inefficient, but also lacks objective and unified quantitative standards, making it difficult to be widely promoted in clinical settings, especially in primary hospitals. Therefore, there is an urgent need for a technical solution that can comprehensively utilize the above-mentioned multiple metabolite indicators to achieve automated and accurate analysis, thereby assisting doctors in making faster and more accurate gout diagnoses. SUMMARY
[0004] The embodiments of the present application provide a gout computer-aided judgment method based on logistic regression, a device and a medium, which solve the technical problem of insufficient accuracy of existing gout auxiliary diagnosis.
[0005] In a first aspect, the embodiments of the present application provide a gout computer-aided judgment method based on logistic regression, characterized in that the method comprises: obtaining a uric acid metabolite body fluid sample and performing tandem mass spectrometry on the uric acid metabolite body fluid sample to obtain a specific metabolite structured matrix; performing elastic network regularization processing on the specific metabolite structured matrix to determine logistic regression model coefficients; based on the logistic regression model coefficients, constructing an initial logistic regression model and performing nonlinear feature analysis of pathological interaction terms on the initial logistic regression model to obtain a gout auxiliary judgment model; inputting the specific metabolite structured matrix into the gout auxiliary judgment model to determine a gout attack probability parameter; and obtaining a gout disease auxiliary diagnosis result through diagnostic consistency verification based on the gout attack probability parameter.
[0006] In an implementation of the present application, the uric acid metabolism body fluid sample is processed by tandem mass spectrometry to obtain a specific metabolite structured matrix, specifically including: performing liquid chromatography separation on the uric acid metabolism body fluid sample to obtain separated specific metabolites; configuring parameters for multiple reaction monitoring scanning of the separated specific metabolites to determine key source parameters; based on the key source parameters, performing characteristic optimization on the separated specific metabolites to obtain characteristic sub-ions; wherein the characteristic optimization includes: declustering voltage and collision energy optimization; based on the characteristic sub-ions, determining the specific metabolite structured matrix by chromatographic peak integration operation.
[0007] In an implementation of the present application, the specific metabolite structured matrix is processed by elastic network regularization to determine the logistic regression model coefficient, specifically including: constructing a regularization loss function, and based on the regularization loss function, calculating the optimal model coefficient; based on the optimal model coefficient, determining the logistic regression loss by logistic regression analysis; performing K-fold cross-validation on the specific metabolite structured matrix to obtain the optimal regularization strength parameter; based on the optimal regularization strength parameter, constructing an elastic network regularization penalty term; based on the logistic regression loss and the elastic network regularization penalty term, determining the logistic regression model coefficient by characteristic metabolite coefficient selection.
[0008] In an implementation of the present application, the initial logistic regression model is analyzed for non-linear features of pathological interaction terms to obtain a gout auxiliary judgment model, specifically including: constructing features of uric acid metabolites on the initial logistic regression model to determine uric acid metabolite feature polynomial terms; performing metabolic influence driving analysis on the uric acid metabolite feature polynomial terms to obtain interaction term data sets of influence relationships between uric acid metabolites; based on the interaction term data sets, determining metabolic influence compliance terms by cross-validation judgment; updating the metabolic influence compliance terms to the initial logistic regression model to obtain the gout auxiliary judgment model.
[0009] In an implementation of the present application, the specific metabolite structured matrix is input into the gout auxiliary judgment model to determine a gout attack probability parameter, specifically including: configuring weights for metabolite concentrations of the specific metabolite structured matrix, and performing weighted calculation on the metabolite concentrations to determine a linear combination score of uric acid metabolites; wherein the metabolite terms of the weight configuration include: serine, piperidine acid, trigonelline, L-isoleucine, uracil, xanthine, glycocholic acid, urate, L-pyroglutamic acid, betaine; performing Sigmoid function calculation on the linear combination score to obtain the gout attack probability parameter.
[0010] In an implementation form of the present application, the gout symptom auxiliary diagnosis result is obtained through the diagnosis consistency verification according to the gout attack probability parameter, and specifically includes: obtaining the gout diagnosis data of the physician, and performing Kappa test on the gout attack probability parameter and the physician diagnosis data to determine a diagnosis consistency parameter; wherein the diagnosis consistency parameter includes: positive consistency rate and negative consistency rate; and performing consistency rate threshold determination on the diagnosis consistency parameter to obtain the gout symptom auxiliary diagnosis result.
[0011] In an implementation form of the present application, after the gout symptom auxiliary diagnosis result is obtained through the diagnosis consistency verification according to the gout attack probability parameter, the method further includes: performing time series analysis on the gout symptom auxiliary diagnosis result to determine the dynamic change trend of the gout attack of the patient; and performing acute attack state analysis on the dynamic change trend of the gout attack of the patient to obtain the metabolite interaction item of the gout acute attack; wherein the metabolite interaction item of the gout acute attack includes: the interaction item of uric acid and short-chain fatty acid, and the interaction item of xanthine and oxidative stress; and the gout attack early warning period of the patient is determined through the baseline change gout acute attack early warning based on the metabolite interaction item of the gout acute attack.
[0012] In an implementation form of the present application, the dynamic change trend of the gout attack of the patient is analyzed to obtain the metabolite interaction item of the gout acute attack, and specifically includes: constructing a feature interaction item according to the dynamic change trend, and determining the interaction state of the constructed feature interaction item to determine a feature interaction coefficient; and the metabolite interaction item of the gout acute attack is obtained through the gout metabolic baseline level analysis of the patient according to the feature interaction coefficient; wherein the baseline period of the baseline level analysis is one week.
[0013] In a second aspect, the embodiments of the present application further provide a gout computer-aided judgment device based on logistic regression, characterized in that the device comprises: at least one processor; and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform a gout computer-aided judgment method based on logistic regression.
[0014] In a third aspect, the embodiments of the present application further provide a computer storage medium storing computer executable instructions, characterized in that the computer executable instructions are executed to implement a gout computer-aided judgment method based on logistic regression.
[0015] The embodiment of the present application provides a gout computer-aided judgment method, device and medium based on a logistic regression, the method comprises the following steps: BRIEF DESCRIPTION OF DRAWINGS
[0016] The drawings described herein are used to provide further understanding of the present application, and form a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application, and do not constitute an improper limitation on the present application. In the drawings: Figure 1 A gout computer-aided judgment method based on a logistic regression is provided for the embodiment of the present application. Figure 2 A gout computer-aided judgment device based on a logistic regression is provided for the embodiment of the present application. DETAILED DESCRIPTION
[0017] In order to make the purpose, technical scheme and advantages of the present application clearer, the technical scheme of the present application will be described clearly and completely in combination with the specific embodiments of the present application and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.
[0018] The embodiment of the present application provides a gout computer-aided judgment method, device and medium based on a logistic regression, the method comprises the following steps:
[0019] The technical scheme provided by the embodiment of the present application will be described in detail below with reference to the drawings.
[0020] Figure 1 A gout computer-aided judgment method based on a logistic regression is provided for the embodiment of the present application. As shown in the figure, Figure 1 The gout computer-aided judgment method based on a logistic regression provided by the embodiment of the present application specifically comprises the following steps: Step 101, obtaining a uric acid metabolism body fluid sample, and performing tandem mass spectrometry on the uric acid metabolism body fluid sample to obtain a specific metabolite structured matrix.
[0021] For example, in the prior art, the selection of metabolites is 7 (serine, piperidine acid, trigonelline, myristic acid / tetradecanoic acid, uracil, arachidonic acid, glycocholic acid), in order to improve the accuracy of gout judgment, 10 metabolites are used for analysis in the present application. In order to separate the 10 metabolites (gold standard) used in the present application from other substances, the present application processes the uric acid metabolism body fluid sample by tandem mass spectrometry to obtain a specific metabolite structured matrix, which is used to represent the absolute concentration of all target metabolites in each sample.
[0022] Specifically, the uric acid metabolism body fluid sample is processed by tandem mass spectrometry to obtain a specific metabolite structured matrix, which includes: performing liquid chromatography separation on the uric acid metabolism body fluid sample to obtain separated specific metabolites; configuring parameters for multi-reaction monitoring scanning of the separated specific metabolites to determine key source parameters; based on the key source parameters, performing characteristic optimization on the separated specific metabolites to obtain characteristic sub-ions; wherein the characteristic optimization includes: declustering voltage and collision energy optimization; according to the characteristic sub-ions, determining the specific metabolite structured matrix by chromatographic peak integration operation.
[0023] In one embodiment, the serum sample is used as the sample, and the deproteinated supernatant sample can be obtained through sample preparation and pretreatment.
[0024] First, the supernatant sample is subjected to chromatography separation, and the chromatography column is a HILIC chromatography column; in the mobile phase, the A phase can be ammonium formate aqueous solution, and the B phase can be acetonitrile. The supernatant sample is subjected to gradient elution, maintaining 95% B phase initially, linearly decreasing to 50% B phase within 10 minutes, maintaining for 2 minutes, and then returning to the initial condition within 10 seconds.
[0025] The HILIC mode can effectively retain and separate high-polarity metabolites, and the gradient elution program optimizes the peak shape and separation degree of metabolites of different polarities, which is beneficial to distinguish 10 different metabolites. The 10 different metabolites used in the present application are: serine, piperidine acid, trigonelline, L-isoleucine, uracil, xanthine, glycocholic acid, urate, L-pyroglutamic acid, and betaine.
[0026] Then, the ion source of the mass spectrometry detection part is selected as an electrospray ion source (ESI), and the scanning mode is selected as a multi-reaction monitoring (MRM) mode. In order to improve the detection efficiency, the ionization polarity analysis of the sample adopts a positive and negative ion switching mode, and all target metabolites are detected simultaneously in one injection.
[0027] Further, it is necessary to set the key source parameters in mass spectrometry detection in advance according to actual needs, and the key source parameters include but are not limited to: ion source temperature, atomizing gas (GS1), auxiliary gas (GS2), air curtain gas, spray voltage (IS).
[0028] Finally, the declustering voltage and collision energy of 10 metabolites and internal standards are optimized respectively to ensure that each target parent ion can be efficiently fragmented and produce characteristic daughter ions, and the MRM chromatographic peaks collected are integrated to obtain a specific metabolite structured matrix. Taking a representative urate as an example, its MRM ion pair is in negative ion mode, and the values of its specific declustering voltage and collision energy need to be optimized to achieve the highest sensitivity.
[0029] Step 102, performing elastic network regularization processing on the specific metabolite structured matrix to determine the coefficients of the logistic regression model.
[0030] For example, since the present application uses a logistic regression model to analyze 10 metabolites in the specific metabolite structured matrix, the weight configuration corresponding to each metabolite needs to be considered in the logistic regression calculation. In the prior art weight configuration, on the one hand, it cannot meet the analysis of more (more than 7) weight items; on the other hand, it cannot consider whether the metabolites will have mutual influence on gout symptoms. The present application performs elastic network regularization processing on the specific metabolite structured matrix, reduces the overfitting risk of the logistic regression model, and improves the feature analysis capability of the model under multiple feature metabolites.
[0031] Specifically, the elastic network regularization processing is performed on the specific metabolite structured matrix to determine the coefficients of the logistic regression model, which specifically includes: constructing a regularization loss function, and calculating the optimal model coefficients based on the regularization loss function; determining the logistic regression loss by logistic regression analysis according to the optimal model coefficients; performing K-fold cross-validation on the specific metabolite structured matrix to obtain an optimal regularization strength parameter; constructing an elastic network regularization penalty term based on the optimal regularization strength parameter; and determining the coefficients of the logistic regression model by feature metabolite coefficient selection according to the logistic regression loss and the elastic network regularization penalty term.
[0032] In one embodiment, each row of the specific metabolite structured matrix is a sample type, and each column is a specific metabolite concentration after Z-score standardization.
[0033] In the regularization processing, the hyperparameter regularization strength controls the weight of the penalty term in the overall loss function. After constructing the initial regularization loss function, a reasonable range of regularization strength values is defined, and the obtained specific metabolite structured matrix is subjected to K-fold cross-validation.
[0034] By selecting a regularization strength value, using K-1 folds of data as the training set to fit the regularized logistic regression model, and using the remaining 1 fold of data as the validation set to evaluate the performance of the model, the average performance score of this regularization strength value in K rounds of validation is calculated.
[0035] Then, the logistic regression loss part is adjusted, the sample number and linear prediction value are determined, and the corresponding prediction probability calculation formula is constructed, and the Sigmoid function can be selected.
[0036] Next, the regularization strength value with the optimal performance score is selected as the optimal regularization strength parameter, and based on the optimal regularization strength parameter, the mixing ratio hyperparameter (1 and 0) is combined, the L1 regularization penalty term when the mixing ratio hyperparameter is 1, and the L2 regularization penalty term when the mixing ratio hyperparameter is 0.
[0037] Finally, through feature metabolite coefficient selection, the coordinate descent method can be used to minimize the objective function to solve the optimal model coefficient to determine the logistic regression model coefficient.
[0038] It should be noted that the hyperparameter regularization strength can also be set by experts in general cases, therefore, the optimal regularization strength parameter also needs to be verified by experts before being put into actual use, and periodically verified and updated.
[0039] Step 103, based on the logistic regression model coefficient, an initial logistic regression model is constructed, and nonlinear feature analysis of pathological interaction terms is performed on the initial logistic regression model to obtain a gout auxiliary judgment model.
[0040] For example, in the occurrence process of gout, the relationship between metabolite concentration and disease risk is not entirely linear, and the standard linear logistic regression model cannot effectively capture such nonlinear effects. The present application optimizes the weight value configuration of the logistic regression model through nonlinear feature analysis of pathological interaction terms, and improves the prediction accuracy and model generalization ability of gout attack.
[0041] Specifically, nonlinear feature analysis of pathological interaction terms is performed on the initial logistic regression model to obtain a gout auxiliary judgment model, including: performing feature construction on the initial logistic regression model to determine the uric acid metabolite feature polynomial term; performing metabolic influence driven analysis on the uric acid metabolite feature polynomial term to obtain an interaction item data group of the influence relationship between uric acid metabolites; based on the interaction item data group, the metabolic influence compliance item is determined through cross-validation judgment; and updating the metabolic influence compliance item to the initial logistic regression model to obtain a gout auxiliary judgment model.
[0042] In one embodiment, the threshold effect, saturation effect and synergistic effect of the interdependence between multiple parameters need to be considered. During the experiment, there may be a critical concentration of urate that induces acute inflammatory response. Below this threshold, the risk increases slowly; once it is exceeded, the risk of inducing acute inflammatory response increases exponentially, while for betaine, the inflammatory response gradually weakens as the concentration increases.
[0043] Further, there is a cross between different metabolic pathways, which may appear to be much greater than the single factor metabolism.
[0044] First, the initial logistic regression model is used to construct the features of uric acid metabolites, and the high-order terms of the core metabolites that are considered to have nonlinear effects, such as urate, are created to generate new features of urate. If the coefficient of urate in the final model is negative, then the high-order term (quadratic or cubic) is positive and negative, and if this data feature appears, it indicates that there is a saturation effect of urate.
[0045] Then, the features of different gout-related metabolic pathways are multiplied to generate interaction terms. The coefficient of the interaction term is used to represent the regulatory relationship between two metabolites. A positive interaction term coefficient indicates that when the concentrations of two metabolites are high, their contributions to risk will be superimposed. Similarly, a negative interaction term coefficient indicates that one metabolite may antagonize the effect of another, thereby determining the interaction term data set.
[0046] Finally, since the direct introduction of polynomials and interaction terms will sharply increase the feature dimension, the present application determines the metabolic impact compliance term by cross-validation, and updates the metabolic impact compliance term to the initial logistic regression model to obtain a gout auxiliary judgment model.
[0047] Step 104, input the specific metabolite structured matrix into the gout auxiliary judgment model to determine the probability parameter of gout attack.
[0048] For example, the specific metabolite structured matrix is input into the gout auxiliary judgment model, which combines the mass spectrometry detection results of 10 specific metabolites, and predicts the risk of gout attack by a logistic regression algorithm, thereby realizing computer-aided judgment of gout under multiple metabolites.
[0049] Specifically, the specific metabolite structured matrix is input into the gout auxiliary judgment model to determine a gout attack probability parameter, including: performing weight configuration of metabolite concentrations on the specific metabolite structured matrix, and performing weighted calculation on the metabolite concentrations to determine a linear combination score of uric acid metabolites; wherein the metabolite items of the weight configuration include: serine, piperidine acid, trigonelline, L-isoleucine, uracil, xanthine, glycocholic acid, urate, L-pyroglutamic acid, betaine; the linear combination score is calculated by a Sigmoid function to obtain the gout attack probability parameter.
[0050] In one embodiment, the weight of each item in the Sigmoid function in the logistic regression algorithm is obtained after elastic network regularization processing and nonlinear feature analysis.
[0051] For the logarithmic ratio of the patient's gout, a series of metabolite logarithmic transformation weighted sums are calculated; wherein the coefficient of each item represents the relationship between the metabolite and the risk of gout attack, which is explained by the following formula.
[0052] (1) Wherein, is the logarithmic ratio of the patient's gout; The numerical term of is the weight of each item in the Sigmoid function, is the intercept term; In turn, they are betaine, serine, piperidine acid, trigonelline, L-isoleucine, uracil, xanthine, urate, glycocholic acid, and L-pyroglutamic acid.
[0053] It should be noted that the weight value determination of the formula is obtained after elastic network regularization processing and nonlinear feature analysis, and the above-mentioned values are only applicable to the case where the above-mentioned 10 metabolites are used as characteristic items. When the metabolite species and quantity need to be updated, the values need to be modified accordingly.
[0054] Then, the calculation of the prediction score is obtained by the exponential transformation of the logarithmic ratio combined with the Sigmoid function to obtain the gout attack probability parameter, and the calculation of the prediction score is explained by the following formula.
[0055]
[0056] Wherein, is the gout attack probability parameter, is the exponential transformation form of the logarithmic ratio of the patient's gout.
[0057] The present application calculates the logarithmic ratio , the Sigmoid function is converted into a probability value between 0 and 1 to predict the risk of gout attack, indicating whether the patient is prone to gout attack.
[0058] Step 105, according to the gout attack probability parameter, through diagnostic consistency verification, the auxiliary diagnosis result of gout disease is obtained.
[0059] For example, according to the literature "China's hyperuricemia and gout diagnosis and treatment guidelines (2019)", the positive consistency rate and negative consistency rate of the low seniority physician combined with the software evaluation result and the high seniority physician evaluation result are calculated; wherein the positive consistency rate is the proportion of the low seniority physician combined with the software evaluation positive result in the high seniority physician evaluation positive result * 100%; the negative consistency rate is the proportion of the low seniority physician combined with the software evaluation negative result in the high seniority physician evaluation negative result * 100%.
[0060] Specifically, according to the gout attack probability parameter, through diagnostic consistency verification, the auxiliary diagnosis result of gout disease is obtained, including: obtaining the physician gout diagnosis data, and performing Kappa test on the gout attack probability parameter and the physician diagnosis data to determine the diagnostic consistency parameter; wherein the diagnostic consistency parameter includes: positive consistency rate, negative consistency rate; the consistency rate threshold of the diagnostic consistency parameter is determined to obtain the auxiliary diagnosis result of gout disease.
[0061] In one embodiment, in the clinical research experiment of gout attack analysis, the subject completes the clinical research corresponding to the technical scheme of the present application, does not violate the inclusion criteria and / or meets the exclusion criteria, and has no missing primary evaluation index.
[0062] Firstly, the baseline analysis of the subject is carried out, and the baseline information includes the demographic information of the subject, the vital signs in the screening period, the medical history and allergy history, and the laboratory examination in the screening period (blood routine examination, liver function examination, kidney function examination, coagulation function examination, urine routine examination).
[0063] In the clinical research experiment corresponding to the embodiments of the present application, the statistical results of the 50 subjects' current medical history are 100% present and 0% nmiss (missing), the statistical results of the past medical history are 54% present past medical history and 46% no past medical history, and the vital signs, blood routine, laboratory examination (urine test, blood biochemistry, C-reactive protein, CA724, etc.) of the subjects still need to be counted.
[0064] The initial diagnosis complaints (symptoms of gout attack) of some subjects are as follows: Subject 1, recurrent joint pain for 4 months. Subject had left ankle pain for 4 months, which was stabbing pain without obvious inducement. It was accompanied by local swelling, affecting activity and sleep. After treatment, joint pain recurred, and the highest blood uric acid was about 618 umol / L. The corresponding specialist condition was left ankle swelling and tenderness.
[0065] Subject 2, blood uric acid was found to be elevated during physical examination 5 years ago, with a blood uric acid value of 440 umol / L. The blood uric acid level was unstable and the highest acid level was about 480 umol / L. The corresponding specialist condition was gouty tophi.
[0066] Subject 3, joint pain for 3 years. The patient had right ankle pain after drinking 3 years ago, with a VAS score of 5. It was accompanied by local swelling and increased skin temperature, affecting activity. The highest blood uric acid was about 800 umol / L, and the typical joint pain occurred twice in the past year. Five months before this visit, the patient had left knee pain without obvious inducement, accompanied by swelling, affecting activity, with a VAS score of 5. The patient took etoricoxib regularly, 20 mg, qd, for 1 year. The corresponding specialist condition was gouty tophi.
[0067] Further, the evaluation period for high seniority physicians in the experiment was set to 30±5 days, and the evaluation period for low seniority physicians was set to 16±5 days. For high seniority physicians, the diagnosis results of hyperuricemia and gout (+5 days and -5 days) were 15 cases and 7 cases, respectively, for a total of 22 cases, accounting for 44%, 0 cases and 28 cases, for a total of 28 cases, accounting for 56%; for low seniority physicians, the diagnosis results of hyperuricemia and gout (+5 days and -5 days) were 15 cases and 0 cases, respectively, for a total of 15 cases, accounting for 30%, and 28 cases, for a total of 35 cases, accounting for 70%.
[0068] Therefore, the positive consistency rate of low seniority physicians combined with the software evaluation results is 80%, the negative consistency rate is 100%, and the Kappa value of Kappa test is calculated to be 0.7059; wherein, the p value of Kappa test in the present application (used to test a null hypothesis) is set to 0.0082.
[0069] According to the experimental results, the consistency rate of low seniority physicians independently diagnosing with the gold standard (i.e. the diagnosis results of high seniority physicians at 30±5 days) is 82.00%; the consistency rate of low seniority physicians combined with the software evaluation results with the gold standard (the diagnosis results of high seniority physicians at 30±5 days) is 86.00%, verifying that the diagnostic ability of low seniority physicians has improved after using the gout auxiliary judgment method of the present application.
[0070] Similarly, the consistency rate of the diagnosis of the senior physician independently with the gold standard (i.e., the diagnosis result of the senior physician on the 30th±5th day) is 86.00%; the consistency rate of the diagnosis of the senior physician in combination with the software evaluation result with the gold standard (i.e., the diagnosis result of the senior physician on the 30th±5th day) is 92.00%, and the diagnosis ability of the senior physician is improved after using the gout auxiliary judgment method of the application.
[0071] Finally, the satisfaction degrees of the operator and the subject are both 100%, proving that the method has good man-machine adaptability.
[0072] Further, after obtaining the gout symptom auxiliary diagnosis result through the diagnosis consistency verification according to the gout attack probability parameter, the method further includes: performing time series analysis on the gout symptom auxiliary diagnosis result to determine the dynamic change trend of the gout attack of the patient; performing acute attack state analysis on the dynamic change trend of the gout attack of the patient to obtain the metabolite interaction item of the gout acute attack; wherein the metabolite interaction item of the gout acute attack includes: the interaction item of uric acid and short-chain fatty acid, and the interaction item of xanthine and oxidative stress; based on the metabolite interaction item of the gout acute attack, the gout acute attack early warning based on the baseline change is performed to determine the gout attack early warning period of the patient.
[0073] Further, the metabolite interaction item of the gout acute attack is obtained by performing acute attack state analysis on the dynamic change trend of the gout attack of the patient, specifically including: constructing a feature interaction item based on the dynamic change trend, and determining the feature interaction coefficient by judging the interaction state of the constructed feature interaction item; and obtaining the metabolite interaction item of the gout acute attack by performing baseline level analysis on the gout metabolism baseline level of the patient according to the feature interaction coefficient; wherein the baseline period of the baseline level analysis is one week.
[0074] In one embodiment, the change rate or trend of a specific metabolite at different time points is calculated for each patient, and the interaction item of the original feature is added explicitly in the feature matrix as the analysis basis of the chronic state and acute attack.
[0075] The metabolic data (uric acid concentration, inflammation baseline level, etc.) in the baseline period needs to be evaluated for the risk of chronic state, and the steady-state markers that best define the chronic inflammation and metabolic state of the patient are screened.
[0076] The gout attack time expectation of the patient needs to be considered for the acute attack early warning, and the gout attack probability of the patient in one week is determined by analyzing the change amount of the recent metabolite of the patient relative to the personal baseline of the patient.
[0077] The above is the method embodiment of the application. Based on the same inventive concept, the application also provides a device, the structure of which is shown in Figure 2 .
[0078] Figure 2 An internal structure diagram of an apparatus is provided in an embodiment of the present application. As shown in the figure, Figure 2 The apparatus comprises: at least one processor 201; and a memory 202 in communication with the at least one processor; The memory 202 stores instructions executable by the at least one processor 201, and the instructions are executed by the at least one processor 201 to enable the at least one processor 201 to: acquire a uric acid metabolism body fluid sample, and perform tandem mass spectrometry on the uric acid metabolism body fluid sample to obtain a specific metabolite structured matrix; perform elastic network regularization processing on the specific metabolite structured matrix to determine a logistic regression model coefficient; based on the logistic regression model coefficient, construct an initial logistic regression model, and perform nonlinear feature analysis of pathological interaction items on the initial logistic regression model to obtain a gout auxiliary judgment model; input the specific metabolite structured matrix into the gout auxiliary judgment model to determine a gout attack probability parameter; and according to the gout attack probability parameter, obtain a gout disease auxiliary diagnosis result through diagnostic consistency verification.
[0079] Some embodiments of the present application provide a non-volatile computer storage medium corresponding to Figure 1 , which stores computer executable instructions, and the computer executable instructions are set to: acquire a uric acid metabolism body fluid sample, and perform tandem mass spectrometry on the uric acid metabolism body fluid sample to obtain a specific metabolite structured matrix; perform elastic network regularization processing on the specific metabolite structured matrix to determine a logistic regression model coefficient; based on the logistic regression model coefficient, construct an initial logistic regression model, and perform nonlinear feature analysis of pathological interaction items on the initial logistic regression model to obtain a gout auxiliary judgment model; input the specific metabolite structured matrix into the gout auxiliary judgment model to determine a gout attack probability parameter; and according to the gout attack probability parameter, obtain a gout disease auxiliary diagnosis result through diagnostic consistency verification.
[0080] Each of the embodiments in the present application is described in a progressive manner, and the same or similar parts of each of the embodiments can be referred to each other. Each of the embodiments mainly describes the difference from other embodiments. In particular, for the Internet of Things apparatus and medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiments.
[0081] The system and medium provided by the embodiments of the present application are one-to-one corresponding, and therefore the system and medium also have similar beneficial technical effects to the method corresponding thereto. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the system and medium will not be described here again.
[0082] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. In addition, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.
[0083] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system), and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus generate a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that carries out the functions specified in one or more flows and / or blocks.
[0084] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction apparatus, which implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that carries out the functions specified in one or more flows and / or blocks.
[0085] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable data processing apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable data processing apparatus provide a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that carries out the functions specified in one or more flows and / or blocks.
[0086] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memories.
[0087] Memory can include non-persistent memory, Random Access Memory (RAM), and / or non-volatile memory, such as Read Only Memory (ROM) or flash memory, in computer readable media. Memory is an example of computer readable media.
[0088] Computer readable media includes permanent and non-permanent, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital versatile disc (DVD), or other optical storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible to a computing device. According to the definition herein, computer readable media does not include transitory media, such as modulated data signals and carrier waves.
[0089] It should also be noted that the terms "comprising", "containing", or any other variant thereof are intended to cover a non-exclusive inclusion, such that a process, method, article or apparatus that comprises a list of elements does not include only those elements in the list, but can also include other elements not expressly listed or inherent to such process, method, article or apparatus. Without further limitation, an element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or apparatus that includes the element.
[0090] The above merely provides an example of the present application and is not intended to limit the present application. The present application can have various modifications and changes for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the scope of claims of the present application.
Claims
1. A method for computer-aided determination of gout based on logistic regression, characterized in that, The method comprises: acquiring a uric acid metabolism body fluid sample, and performing tandem mass spectrometry on the uric acid metabolism body fluid sample to obtain a specific metabolite structured matrix; performing elastic network regularization processing on the specific metabolite structured matrix to determine a logistic regression model coefficient; based on the logistic regression model coefficient, constructing an initial logistic regression model, and performing nonlinear feature analysis of pathological interaction items on the initial logistic regression model to obtain a gout auxiliary judgment model; inputting the specific metabolite structured matrix into the gout auxiliary judgment model to determine a gout attack probability parameter; according to the gout attack probability parameter, obtaining a gout disease auxiliary diagnosis result through diagnostic consistency verification.
2. The computer-aided gout diagnosis method based on logistic regression according to claim 1, characterized in that, The specific metabolite structured matrix is obtained by performing tandem mass spectrometry on the uric acid metabolism body fluid sample, specifically comprising: performing liquid chromatography separation on the uric acid metabolism body fluid sample to obtain separated specific metabolites; configuring parameters for multiple reaction monitoring scanning of the separated specific metabolites to determine key source parameters; based on the key source parameters, performing characteristic optimization on the separated specific metabolites to obtain characteristic sub-ions; wherein the characteristic optimization includes cluster voltage optimization and collision energy optimization; according to the characteristic sub-ions, determining the specific metabolite structured matrix through chromatographic peak integration operation.
3. The computer-aided gout diagnosis method based on logistic regression according to claim 1, characterized in that, The specific metabolite structured matrix is obtained by performing tandem mass spectrometry on the uric acid metabolism body fluid sample, specifically comprising: constructing a regularization loss function, and calculating optimal model coefficients based on the regularization loss function; according to the optimal model coefficients, determining a logistic regression loss through logistic regression analysis; performing K-fold cross-validation on the specific metabolite structured matrix to obtain an optimal regularization strength parameter; based on the optimal regularization strength parameter, constructing an elastic network regularization penalty term; according to the logistic regression loss and the elastic network regularization penalty term, determining the logistic regression model coefficient through characteristic metabolite coefficient selection.
4. The computer-aided gout diagnosis method based on logistic regression according to claim 1, characterized in that, The specific metabolite structured matrix is obtained by performing tandem mass spectrometry on the uric acid metabolism body fluid sample, specifically comprising: performing characteristic construction of uric acid metabolites on the initial logistic regression model to determine uric acid metabolite characteristic polynomial terms; performing metabolic influence driving analysis on the uric acid metabolite characteristic polynomial terms to obtain interaction item data sets of influence relationships between uric acid metabolites; based on the interaction item data sets, determining metabolic influence compliance items through cross-validation judgment; updating the metabolic influence compliance items to the initial logistic regression model to obtain the gout auxiliary judgment model.
5. The computer-aided gout diagnosis method based on logistic regression according to claim 1, characterized in that, The specific metabolite structured matrix is obtained by performing tandem mass spectrometry on the uric acid metabolism body fluid sample, specifically comprising: weighting the metabolite concentration of the specific metabolite structured matrix, and performing a weighted calculation on the metabolite concentration to determine a linear combination score of uric acid metabolites; wherein the metabolite items of the weighting include: serine, piperidine acid, trigonelline, L-isoleucine, uracil, xanthine, glycocholic acid, urate, L-pyroglutamic acid, betaine; performing a Sigmoid function calculation on the linear combination score to obtain the gout attack probability parameter.
6. The computer-aided gout diagnosis method based on logistic regression according to claim 1, characterized in that, According to the gout attack probability parameter, through diagnostic consistency verification, a gout syndrome auxiliary diagnosis result is obtained, specifically including: Obtaining physician gout diagnosis data, and performing Kappa test on the gout attack probability parameter and the physician diagnosis data to determine a diagnostic consistency parameter; wherein the diagnostic consistency parameter includes: positive consistency rate, negative consistency rate; Performing a consistency rate threshold determination on the diagnostic consistency parameter to obtain a gout syndrome auxiliary diagnosis result.
7. The computer-aided gout diagnosis method based on logistic regression according to claim 1, characterized in that, After obtaining the gout syndrome auxiliary diagnosis result according to the gout attack probability parameter through diagnostic consistency verification, the method further includes: Performing time series analysis on the gout syndrome auxiliary diagnosis result to determine the dynamic change trend of the patient's gout attack; Performing acute attack state analysis on the dynamic change trend of the patient's gout attack to obtain metabolite interaction items of gout acute attack; wherein the metabolite interaction items of gout acute attack include: interaction items of uric acid and short-chain fatty acids, interaction items of xanthine and oxidative stress; Based on the metabolite interaction items of gout acute attack, through baseline change gout acute attack early warning, the patient's gout attack early warning period is determined.
8. The computer-aided gout diagnosis method based on logistic regression according to claim 7, characterized in that, Performing acute attack state analysis on the dynamic change trend of the patient's gout attack to obtain metabolite interaction items of gout acute attack, specifically including: Performing feature interaction item construction on the dynamic change trend, and performing interaction state determination on the constructed feature interaction items to determine a feature interaction coefficient; According to the feature interaction coefficient, through patient gout metabolism baseline level analysis, the metabolite interaction items of gout acute attack are obtained; wherein the baseline period of the baseline level analysis is one week.
9. A logic regression based computer aided gout diagnosis device, characterized in that, The device includes: at least one processor; and a memory connected in communication with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform a gout computer-aided judgment method based on logistic regression according to any one of claims 1-8.
10. A computer storage medium storing computer-executable instructions, which, when executed by a processor, cause the processor to perform acts comprising: The computer executable instructions, when executed, implement a gout computer-aided judgment method based on logistic regression according to any one of claims 1-8.