Machine learning-based unsupervised phenotypic clustering analysis of patients with IgA nephropathy: Distinct therapeutic responses of different groups

Chinese Medical Journal, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 7, 2025

Immunoglobulin A nephropathy (IgAN) has a heterogeneous clinical presentation. Comparison of different IgAN subgroups may facilitate the application more targeted therapies. This study was aimed to distinct disease phenotypes in and develop prognostic models for renal composite outcomes. Clinical pathological data were from 2000 patients with biopsy-proven primary four centers, including First Affiliated Hospital Sun Yat-sen University (SYSU), Fifth University, Huadu District People's Guangzhou, Jieyang SYSU China between January 2009 December 2018 (training cohort: 1203 patients, validation 797 patients). Components principal components analysis (PCA) used fit k-means clustering algorithm identify subgroups. subgroup-based prediction model developed assess prognosis therapeutic efficacy each subgroup. The PCA-k-means identified Subgroup 1 had significantly better long-term survival upon administration renin-angiotensin system blocker (adjusted hazard ratio [aHR]: 0.16, 95% confidence interval [CI]: 0.10-0.27, P <0.001). 2 significant improvement corticosteroid therapy (aHR: 0.19, CI: 0.06-0.61, = 0.005). Subgroups 3 4 milder changes relatively stable kidney function several years. (predominantly males) high incidence metabolic risk factors, necessitating intensive monitoring; subgroup females) recurrent macroscopic hematuria. These patterns similar cohort. demonstrated an area under curve 0.856 dataset. unsupervised method provided reliable classification into according features, prognoses, treatment responsiveness. Our utility assessment IgAN.

Language: Английский

---

Novel metabolic and inflammatory stratification of overweight/obesity to characterize risks of adverse outcomes: A large population‐based cohort study

Diabetes Obesity and Metabolism, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 19, 2025

Abstract Aims The growing epidemic of overweight and obesity elevates disease risks, with metabolic disorders inflammation critically involved in the pathogenic mechanisms. This study refines subtyping using inflammatory markers to enhance risk assessment personalized prevention. Materials Methods Based on UK Biobank, this retrospective included participants classified as or obese (BMI ≥25 kg/m 2 ). K‐means clustering was performed biomarkers. Multivariate Cox regression analysis assessed complications mortality over a follow‐up period 13.5 years. Genome‐Wide Association Studies (GWAS) Phenome‐Wide (PheWAS) explored cluster‐specific genetic traits. Results Among 126 145 (mean [IQR] age: 55.0 [14.0] years; 61 983 males [49.1%]), five clusters were identified: (1) Low Metabolic Risk‐related, (2) Hypertension‐Related, (3) Mixed Hyperlipidemia‐Related, (4) Elevated Lipoprotein(a)‐Related (5) High BMI Inflammation‐Related. Cluster 1 exhibited lower than other clusters. had highest incidence stroke, linked variants affecting blood circulation. 3 showed risks for ischaemic heart disease, characterized by enriched cholesterol metabolism pathways. 4 associated high cardiovascular risks. 5 diabetes, asthma, chronic obstructive pulmonary osteoarthritis mortality, obesity‐related variants. We also proposed method applying classification clinical settings. Conclusions provides insights into heterogeneity individuals obesity, aiding identification high‐risk patients who may benefit from targeted interventions.

Language: Английский

---

Machine learning-based disease risk stratification and prediction of metabolic dysfunction-associated fatty liver disease using vibration-controlled transient elastography: Result from NHANES 2021–2023

BMC Gastroenterology, Journal Year: 2025, Volume and Issue: 25(1)

Published: April 14, 2025

Language: Английский

---

Accurate and sensitive low-density lipoprotein (LDL) detection based on the proximity ligation assisted rolling circle amplification (RCA)

Analytical Methods, Journal Year: 2024, Volume and Issue: 16(13), P. 1894 - 1900

Published: Jan. 1, 2024

Accurate and sensitive low-density lipoprotein (LDL) detection.

Language: Английский

---

The Role of Simple and Specialized Non-Invasive Tools in Predicting of Metabolic dysfunction-associated Fatty Liver Disease Severity and Prognosis

Obesity Medicine, Journal Year: 2025, Volume and Issue: unknown, P. 100582 - 100582

Published: Jan. 1, 2025

Language: Английский

---

The Inter-Organ Crosstalk Reveals an Inevitable Link between MAFLD and Extrahepatic Diseases

Nutrients, Journal Year: 2023, Volume and Issue: 15(5), P. 1123 - 1123

Published: Feb. 23, 2023

Fatty liver is known to be associated with extra-hepatic diseases including atherosclerotic cardiovascular disease and cancers, which affect the prognosis quality of life patients. The inter-organ crosstalk mediated by metabolic abnormalities such as insulin resistance visceral adiposity. Recently, dysfunction-associated fatty (MAFLD) was proposed a new definition for liver. MAFLD characterized inclusion criteria abnormality. Therefore, expected identify patients at high risk complications. In this review, we focus on relationships between multi-organ diseases. We also describe pathogenic mechanisms crosstalk.

Language: Английский

---

Clustering NAFLD: phenotypes of nonalcoholic fatty liver disease and their differing trajectories

Hepatology Communications, Journal Year: 2023, Volume and Issue: 7(4)

Published: March 24, 2023

Globally, in the general adult population, prevalence of NAFLD is estimated at 25%.1 The higher (∼40%–60%) overweight and obese subjects, particularly presence impaired metabolic health,2 highest global (∼55%–70%) found patients with diabetes.3 main cause chronic liver disease HCC.4 Furthermore, strong epidemiological relationships type 2 diabetes cardiovascular diseases indicate very close pathophysiological between obesity-associated cardiometabolic diseases.5 However, there large heterogeneity among respect to their risk diseases.3 This may result from fact that different major pathways are involved pathogenesis NAFLD. Among them associated a stronger hepatic genetic component. For example, variants PNPLA3 (148Met allele) TM6SF2 (167Lys strongly associate steatosis progression NASH, cirrhosis, HCC, but also absence insulin resistance, low blood triglycerides, LDL cholesterol concentration, protection coronary artery disease. predominantly driven by de novo lipogenesis adipose tissue dysfunction exist, which both resistance thought differ Yi et al6 set out identify clinically important groups assess long-term outcomes subphenotypes and, most recently, published findings Hepatology Communications. this, they analyzed data US Third National Health Nutrition Examination Survey, where fatty was diagnosed individuals abdominal ultrasound used linked mortality through December 2019. As dimensionality reduction approach, authors performed 2-stage cluster analysis (a hierarchical using Ward method determine optimum number clusters, followed an allocation each patient into particular cluster). Using 21 baseline variables, body mass index (BMI), waist circumference, hemoglobin, glycohemoglobin, waist-to-hip ratio, uric acid, HDL cholesterol, homeostasis model assessment were identified as variables prediction clusters. Three distinct clusters identified. Cluster 1 comprised younger (mean age 40 y), lean BMI 24 kg/m2) females (76%) profile lower comorbidities. consisted mostly older 50 34 (75%) high (83%) (34%). 3 composed 49 overweight/obese 30 males (72%), moderately elevated (15%), hypertension, atherogenic dyslipidemia. During median follow-up period 312 months compared 1, had all-cause mortality, after adjustment for age, sex, BMI, race/ethnicity. No differences observed 3.6 concluded NAFLD, who allocated did not have or dyslipidemia, pathophysiology related Unfortunately, could study incident fibrosis it would be expected no incidence these advanced stages exist further hypothesized 2, being having severe diabetes, mainly dysfunction. measurements free acids, this allowed estimate resistance,7 been test hypothesis. Finally, 3, kidney damage, lipogenesis. determination lipogenesis, serum acid ratios low-density lipoprotein triglycerides,8 helped investigate Altogether, provided interesting novel clustering help pathomechanisms Because differed factors, such fat distribution regarding stratification outcomes, proposed approach seems superior established models. Recently, another has 5 metabolic-associated (MAFLD) Chinese cohort validated results UK Biobank database.9 That only relatively small (age, total cholesterol/HDL ratio (a) levels). Patients exhibited risks heart disease, all-causes mortality. referred "severe resistance–related MAFLD," significantly worst survival than those other "mild obesity dyslipidemia-related MAFLD" (cluster 1), "age-related 2), "high (a)-related 4), mixed hyperlipidemia-related 5). Ye al.9 al.6 highlight specific depends on parameters generate parameters, probably phenotype, drive (Figure 1).FIGURE 1: Hypothetical depiction Several can hypothetical A–C often depicted. font size indicates large, moderate, impact assignment subjects individual X Y represent additional case (1-7) In cases, linking unknown. unclear whether mediate liver–associated mortality.What approaches mechanisms promoting diseases? purpose, focus hepatokines adipokines. respect, we recently how hepatokine fetuin-A adipokine adiponectin, together precisely measured content visceral mass, whom dysfunctional diseases.10 summary, approaches, cluster, principal component, factor analyses, became popular clinical research. They field research important, considering NAFLD.3 analytical careful selection necessary advance

Language: Английский

---

Efficient Hybrid CNN Method to Classify the Liver Diseases

Journal of Wireless Mobile Networks Ubiquitous Computing and Dependable Applications, Journal Year: 2023, Volume and Issue: 14(3), P. 36 - 47

Published: Sept. 30, 2023

This study focuses on classifying liver diseases using dynamic CT scan images and deep learning techniques. The primary objective is to develop accurate efficient models for distinguishing between different disease categories. Three models, ResNet50, ResNet18, AlexNet, are employed three-class classification, including Hepatitis/cirrhosis, Hepatitis/Fatty liver, Hepatitis/Wilson's Disease. dataset comprises of the each manually segmented identify lesions. To enhance model performance, data pre-processed by resizing, normalization, augmentation. split into training, validation, test sets evaluation. performance assessed confusion matrices, accuracy, sensitivity, specificity. Results show varying accuracies classes, indicating strengths limitations models. overcome limits classifiers, a framework Efficient Hybrid CNN method classify Liver (EHCNNLD) proposed, combining predictions from three with weighted probabilities. Proposed EHCNNLD demonstrates improved accuracy classification power, enhancing overall classification. highlights potential techniques in medical image analysis clinical diagnosis. findings provide valuable insights developing robust paving way research patient care advancements.

Language: Английский

---

Identification of Mitophagy-Associated Genes for the Prediction of Metabolic Dysfunction-Associated Steatohepatitis Based on Interpretable Machine Learning Models

Journal of Inflammation Research, Journal Year: 2024, Volume and Issue: Volume 17, P. 2711 - 2730

Published: May 1, 2024

Background: This study aims to elucidate the role of mitochondrial autophagy in metabolic dysfunction-associated steatohepatitis (MASH) by identifying and validating key mitophagy-related genes diagnostic models with potential. Methods: The gene expression profiles clinical information MASH patients healthy controls were obtained from Gene Expression Omnibus database (GEO). Limma functional enrichment analysis used identify differentially expressed (mito-DEGs) patients. Machine learning select mito-DEGs evaluate their efficacy early diagnosis MASH. levels validated using datasets cell models. A nomogram was constructed assess risk progression based on mito-DEGs. molecular subtypes evaluated. Results: Four mito-DEGs, namely MRAS, RAB7B, RETREG1, TIGAR identified. Among machine employed, Support Vector demonstrated highest AUC value 0.935, while Light Gradient Boosting model exhibited accuracy (0.9189), kappa (0.7204), F1-score (0.9508) values. Based these models, RETREG1 selected for further analysis. logistic regression could accurately predict diagnosis. DEGs excellent prediction performance. three independent results found be consistent findings through bioinformatics Furthermore, our revealed significant differences patterns, immune characteristics, biological functions, pathways between Subtype-specific small-molecule drugs identified CMap database. Conclusion: Our research provides novel insights into mitophagy uncovers targets predictive personalized treatments. Keywords: steatohepatitis, mitophagy, biomarkers, model,

Language: Английский

---

Clinical application of cluster analysis in patients with newly diagnosed type 2 diabetes

HORMONES, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 4, 2024

Language: Английский

---