New typing system for metabolic dysfunction-related fatty liver disease based on clustering analysis

Abstract

The invention discloses a new typing system for metabolic dysfunction related fatty liver diseases based on clustering analysis, and belongs to the technical field of bioinformatics analysis. Comprising a clustering module, an evaluation module and a typing module, the clustering module is used for collecting database data of metabolic dysfunction related fatty liver disease patients and randomly distributing the database data of the patients into a training set, a verification set and a test set according to the proportion of 6: 2: 2; a clustering model is constructed, clustering features of the training set are screened through the clustering model, the clustering features are clustered, and a clustering result is obtained; the evaluation module is used for predicting the complication risk of the clustering result, calculating a risk ratio and a credible interval and evaluating the clustering effect; and the typing module is used for carrying out complication typing on the metabolic dysfunction related fatty liver disease patient by adopting the clustering model to obtain a typing result.

Description

Technical Field

[0001] The present invention belongs to the technical field of bioinformatics analysis, and in particular relates to a new classification system for metabolic dysfunction-related fatty liver disease based on cluster analysis. Background Art

[0002] Metabolic dysfunction-associated steatotic liver disease (MASLD) is a new consensus to replace nonalcoholic fatty liver disease (NAFLD) determined in 2023 after four rounds of Delphi surveys. MASLD is a chronic liver disease with a sharply increased prevalence in recent years, affecting 20%-30% of the world's population. It is considered a liver manifestation of metabolic syndrome and is associated with an increased risk of end-stage liver diseases such as cirrhosis, hepatocellular carcinoma, and liver failure. In addition, MASLD increases the risk of extrahepatic consequences, including type 2 diabetes (T2DM), cardiovascular and cerebrovascular diseases, and kidney diseases. In addition, about half of the deaths in MASLD patients are related to complications. MASLD has caused a heavy social and economic burden and is one of the major public health issues.

[0003] MASLD is defined as patients with hepatic steatosis and at least one of the five metabolic factors. Compared with NAFLD, it emphasizes metabolic factors more and can better identify patients with metabolic disorders, but it still cannot identify the high heterogeneity and metabolic characteristics of MASLD patients, which makes the management and treatment of MASLD patients difficult. At present, due to the unobvious clinical manifestations of MASLD, high-risk groups do not pay enough attention to MASLD, and they often have serious complications when they come to the hospital, so it is necessary to identify MASLD patients at risk of complications early. At present, the severity stratification of MASLD is a simple dichotomy (i.e., non-steatohepatitis and fatty liver hepatitis), without further distinguishing different types of MASLD, which are highly heterogeneous between individuals. Existing technologies cannot identify the heterogeneity and complication risk of MASLD patients, and there is an urgent need to stratify the complication risk of MASLD patients. Summary of the invention

[0004] The present invention aims to address the deficiencies of the prior art, proposes a new classification system for metabolic dysfunction-associated fatty liver disease based on cluster analysis, establishes a classification system applicable to all MASLD patients, identifies the heterogeneity of MASLD patients, and forms a new MASLD classification system that can predict the risk of complications, stratifies the risk of complications in MASLD patients, and overcomes the difficulties in treatment and management caused by the high heterogeneity of MASLD patients.

[0005] To achieve the above-mentioned object, the present invention provides the following scheme: a new classification system for metabolic dysfunction-related fatty liver disease based on cluster analysis, comprising: a clustering module, an evaluation module and a classification module;

[0006] The clustering module is used to collect database data of patients with metabolic dysfunction-related fatty liver disease, and randomly allocate the database data of the patients into a training set, a validation set, and a test set in a ratio of 6:2:2; and to construct a clustering model, use the clustering model to screen clustering features for the training set, and cluster the clustering features to obtain a clustering result;

[0007] The evaluation module is used to predict the complication risk of the clustering results, calculate the risk ratio and the confidence interval, and evaluate the clustering effect;

[0008] The classification module is used to use the clustering model to classify complications of patients with metabolic dysfunction-related fatty liver disease to obtain classification results.

[0009] Further preferably, the method for obtaining the clustering result by the clustering model includes:

[0010] A supervised learning algorithm is used to generate continuous output values ​​for the training set; the correlation coefficient method is used to calculate the correlation coefficient between each feature and the output value of the training set; and features with significant correlation coefficients are selected;

[0011] The rank correlation coefficient between the selected feature and the non-invasive serological index Liver Fibrosis Score-4 score is calculated, and the clustering feature is obtained based on the rank correlation coefficient.

[0012] Further preferably, the clustering features include: age, aspartate aminotransferase, waist-to-hip ratio, monocyte / lymphocyte ratio, low-density lipoprotein cholesterol, cholesterol and body mass index.

[0013] Further preferably, the unsupervised clustering method K-means is used to cluster the clustering features, and patients with metabolic dysfunction-related fatty liver disease are divided into 5 subtypes; the subtypes are: Class A, Class B, Class C, Class D, and Class E.

[0014] Further preferably, the prognosis of different subtypes is compared by Kaplan-Meier curve; the difference of outcome events between different subtypes is compared by logarithmic rank test; the hazard ratio and 95% confidence interval of different outcome events are calculated by Cox proportional hazard regression model; the outcome events include: intrahepatic complications and extrahepatic complications in patients with fatty liver disease associated with metabolic dysfunction;

[0015] When the result of the training set is consistent with the results of the validation set and the test set, the clustering model is reliable and stable.

[0016] Further preferably, the class E is a high-risk group for severe intrahepatic complications, and classes A, B, C, and D are combined into a low-risk group for complications.

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] The present invention clusters inflammatory and metabolic indicators related to the occurrence and development of MASLD, stratifies MASLD patients through non-invasive detection methods, and conducts in-depth research on their risk association with complications, so as to help identify high-risk groups of complications in MASLD patients at an early stage. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0020] Figure 1 A schematic diagram of a typing mode performed by a typing system according to an embodiment of the present invention;

[0021] Figure 2 A schematic diagram of correlation coefficients of various features of an embodiment of the present invention;

[0022] Figure 3 It is a schematic diagram of the cumulative curve of outcome events occurring in different subtypes according to the embodiments of the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0025] Embodiment 1:

[0026] This embodiment provides a new classification system for metabolic dysfunction-related fatty liver disease based on cluster analysis, including: a clustering module, an evaluation module and a classification module. Figure 1 Schematic diagram of the pattern of typing using the analysis system of this embodiment.

[0027] The clustering module is used to collect database data of patients with fatty liver disease related to metabolic dysfunction, and randomly allocate the patients' database data into training set, validation set and test set in a ratio of 6:2:2; and build a clustering model, use the clustering model to screen clustering features of the training set, and cluster the clustering features to obtain clustering results.

[0028] In this embodiment, data of 125,197 patients with MASLD were collected from the UK Biobank from 2006 to 2010, including 78,791 males (average age 58 years old) and 46,406 females (average age 59 years old), 92% of whom were Caucasians. Among them, the patient's data include: age, gender, population information, and metabolic, inflammatory and statistical indicators related to the occurrence and development of MASLD. Specifically, including: age, gender, population information, body mass index (BMI), alanine aminotransferase (ALT), aspartate aminotransferase (AST), albumin (ALB), total cholesterol (CHOL), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), glycosylated hemoglobin (HbA1c) and waist-hip ratio (Waist-hip rate, WHR), and inflammatory indicators-monocyte / lymphocyte ratio (Monocyte-lymphocyte ratio, MLR). They were randomly divided into three cohorts, namely training set, validation set and test set, in a ratio of 6:2:2 for subsequent analysis.

[0029] A supervised learning algorithm is used to generate continuous output values ​​for the training set; the correlation coefficient method is used to calculate the correlation coefficient between each feature and the output value of the training set; Figure 2 As shown, features with significant correlation coefficients are selected. The correlation coefficient is a statistical indicator first designed by statistician Karl Pearson, and is the amount of linear correlation between research variables. Due to the different research objects, the correlation coefficient has a variety of definitions, and this embodiment uses the Pearson correlation coefficient. The rank correlation coefficient of the selected continuous features and the non-invasive serological index liver fibrosis score-4 (FIB-4) score is calculated, and the clustering features are obtained according to the standard highly correlated with the FIB-4 score. In this embodiment, the clustering features include: age, aspartate aminotransferase (AST), waist-to-hip ratio (WHR), monocyte / lymphocyte ratio (MLR), low-density lipoprotein cholesterol (LDL-C), total cholesterol (CHOL) and body mass index (BMI).

[0030] Furthermore, the unsupervised clustering method K-means was used to cluster the cluster features and divide the patients with metabolic dysfunction-related fatty liver disease into five subtypes; the subtypes were: A, B, C, D, and E. K-means clustering is a basic algorithm widely used in cluster analysis, which is used to identify clusters with acceptable stability in the training set. Subsequently, the cluster distribution of the validation set and the test set was studied, aiming to achieve a distribution similar to that of the training set. K-means is a commonly used clustering algorithm based on Euclidean distance, which believes that the closer the distance between two targets, the greater the similarity. Cluster analysis was performed after the data was standardized, and the results were verified in the validation set and the test set using the K-means clustering method, and a cluster distribution similar to that of the training set clustering was obtained. Finally, the analysis showed that MLR, age, AST, WHR, LDL-C, CHOL, and BMI had better cluster quality both statistically and clinically, and the MASLD population could be divided into five subtypes. In different clusters, the baseline characteristics of different subtypes were described and compared according to the different characteristics of the variables. The distribution of continuous variables was expressed as median and interquartile range, and categorical variables were expressed as frequency and percentage. The chi-square test and Kruskal-Wallis rank sum test were used to assess the balance between groups of different characteristics.

[0031] The evaluation module is used to predict the complication risk of clustering results, calculate the risk ratio and confidence interval, and evaluate the clustering effect.

[0032] Specifically, the prognosis of different subtypes was compared by Kaplan-Meier curve; the difference of outcome events between different subtypes was compared by logarithmic rank test; the hazard ratio and 95% confidence interval (CI) of different outcome events were calculated by Cox proportional hazard model; in this embodiment, the outcome events include: intrahepatic complications and extrahepatic complications in patients with fatty liver disease related to metabolic dysfunction; intrahepatic complications refer to severe liver disease (SLD) in MASLD individuals, and extrahepatic complications mainly refer to end-stage renal disease (ESRD), stroke, heart failure (HF), coronary heart disease (CAD) and diabetes mellitus in MASLD individuals after diagnosis of MASLD. Figure 3 As shown, Figure 3 a. Figure 3 b. Figure 3 c. Figure 3 d. Figure 3 e. Figure 3 f are schematic diagrams of the cumulative curves of SLD, ESRD, Stroke, HF, CAD, and Diabetes in different subtypes. When the results of the training set are consistent with those of the validation set and the test set, the clustering model is reliable and stable.

[0033] The Cox proportional hazard model is a classic data analysis and modeling method. It is a semi-parametric regression model proposed by British statistician DRCOX in 1972. It can simultaneously study the relationship between single or multiple risk factors and the occurrence and time of event outcomes, thus overcoming the shortcomings of single-factor restrictions in simple survival analysis (such as the KM method, i.e., the product limit method). All tests are two-sided, and factors with a P value ≤ 0.05 are considered statistically significant.

[0034] This example compares the risk of intrahepatic and extrahepatic complications for each cluster. The relationship between the outcome events of different subtypes in the training set, validation set, and test set calculated by Cox regression is shown in Table 1, where the A-type subtype is used as the reference group to obtain the risk ratio. In the training set, the median follow-up was 4,989 days, and the highest risk of SLD in class E was more than 14 times that of class A. In the training set, the ESRD risk of classes D and E was almost twice that of class A. Class D has the highest risk of stroke, and class B has the lowest risk of stroke. Class D has the highest risk of HF, followed by class E, and class B has the lowest risk; class D has the highest risk of CAD, and class B has the lowest risk. Class E has the highest risk of type 2 diabetes, followed by class D. The results of the training set are consistent with those of the validation set and the test set. The reliability and stability of the clustering model are verified.

[0035] Table 1

[0036]

[0037] The classification module is used to classify complications of patients with fatty liver disease associated with metabolic dysfunction using a clustering model to obtain classification results.

[0038] After the stability and reliability of the clustering model were confirmed through the above verification, it was applied to the MASLD population. Furthermore, class E was defined as the high-risk group for severe intrahepatic complications, and classes A, B, C, and D were combined into the low-risk group for complications. This facilitates the subsequent development of personalized intervention measures for the high-risk group for complications and reduces the occurrence of complications in high-risk patients.

[0039] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A new classification system for metabolic dysfunction-related fatty liver disease based on cluster analysis, characterized by: include: Clustering module, evaluation module and typing module; The clustering module is used to collect database data of patients with metabolic dysfunction-related fatty liver disease, and randomly allocate the database data of the patients into a training set, a validation set, and a test set in a ratio of 6:2:2; A clustering model is constructed, the clustering model is used to screen clustering features of the training set, and the clustering features are clustered to obtain clustering results; The evaluation module is used to predict the complication risk of the clustering results, calculate the risk ratio and the confidence interval, and evaluate the clustering effect; The classification module is used to use the clustering model to classify complications of patients with metabolic dysfunction-related fatty liver disease to obtain classification results.

2. The new classification system for metabolic dysfunction-related fatty liver disease based on cluster analysis according to claim 1, characterized in that: The method for obtaining the clustering result by the clustering model includes: A supervised learning algorithm is used to generate continuous output values ​​for the training set; the correlation coefficient method is used to calculate the correlation coefficient between each feature and the output value of the training set; and features with significant correlation coefficients are selected; The rank correlation coefficient between the selected feature and the non-invasive serological index Liver Fibrosis Score-4 score is calculated, and the clustering feature is obtained based on the rank correlation coefficient.

3. The new classification system for metabolic dysfunction-related fatty liver disease based on cluster analysis according to claim 1, characterized in that: The clustering features included age, aspartate aminotransferase, waist-to-hip ratio, monocyte / lymphocyte ratio, low-density lipoprotein cholesterol, cholesterol, and body mass index.

4. The new classification system for metabolic dysfunction-related fatty liver disease based on cluster analysis according to claim 1, characterized in that: The clustering features were clustered using the unsupervised clustering method K-means, and patients with metabolic dysfunction-related fatty liver disease were divided into five subtypes; the subtypes were: A, B, C, D, and E.

5. The new classification system for metabolic dysfunction-related fatty liver disease based on cluster analysis according to claim 4, characterized in that: The prognosis of different subtypes was compared by Kaplan-Meier curves; the differences in outcome events between different subtypes were compared by logarithmic rank test; Cox proportional hazards regression model was used to calculate the hazard ratios and 95% confidence intervals of different outcome events; The outcome events include: intrahepatic and extrahepatic complications in patients with fatty liver disease associated with metabolic dysfunction; When the result of the training set is consistent with the results of the validation set and the test set, the clustering model is reliable and stable.

6. The new classification system for metabolic dysfunction-related fatty liver disease based on cluster analysis according to claim 4, characterized in that: The class E is a high-risk group for severe intrahepatic complications, and classes A, B, C, and D are combined into a low-risk group for complications.

Images