Uncovering the immune microenvironment and molecular subtypes of hepatitis B-related liver cirrhosis and developing stable a diagnostic differential model by machine learning and artificial neural networks

Frontiers in Molecular Biosciences, Journal Year: 2023, Volume and Issue: 10

Published: Sept. 22, 2023

Background: Hepatitis B-related liver cirrhosis (HBV-LC) is a common clinical disease that evolves from chronic hepatitis B (CHB). The development of can be suppressed by pharmacological treatment. When CHB progresses to HBV-LC, the patient's quality life decreases dramatically and drug therapy ineffective. Liver transplantation most effective treatment, but lack donor required for transplantation, high cost procedure post-transplant rejection make this method unsuitable patients. Methods: aim study was find potential diagnostic biomarkers associated with HBV-LC bioinformatics analysis classify into specific subtypes consensus clustering. This will provide new perspective early diagnosis, treatment prevention HCC in Two study-relevant datasets, GSE114783 GSE84044, were retrieved GEO database. We screened feature genes using differential analysis, weighted gene co-expression network (WGCNA), three machine learning algorithms including least absolute shrinkage selection operator (LASSO), support vector recursive elimination (SVM-RFE), random forest (RF) total five methods. After that, we constructed an artificial neural (ANN) model. A cohort consisting GSE123932, GSE121248 GSE119322 used external validation. To better predict risk development, also built nomogram And multiple enrichment analyses samples performed understand biological processes which they significantly enriched. different analyzed Immune infiltration approach. Results: Using data downloaded GEO, developed ANN model based on six genes. clustering classified them two subtypes, C1 C2, it hypothesized patients subtype C2 might have milder symptoms immune analysis. Conclusion: column line graphs showed excellent predictive power, providing diagnosis possible HBV-LC. delineation facilitate future

Language: Английский

---

Hybrid Explainable Artificial Intelligence Models for Targeted Metabolomics Analysis of Diabetic Retinopathy

Diagnostics, Journal Year: 2024, Volume and Issue: 14(13), P. 1364 - 1364

Published: June 27, 2024

Diabetic retinopathy (DR) is a prevalent microvascular complication of diabetes mellitus, and early detection crucial for effective management. Metabolomics profiling has emerged as promising approach identifying potential biomarkers associated with DR progression. This study aimed to develop hybrid explainable artificial intelligence (XAI) model targeted metabolomics analysis patients DR, utilizing focused identify specific metabolites exhibiting varying concentrations among individuals without (NDR), those non-proliferative (NPDR), proliferative (PDR) who have type 2 mellitus (T2DM). A total 317 T2DM patients, including 143 NDR, 123 NPDR, 51 PDR cases, were included in the study. Serum samples underwent using liquid chromatography mass spectrometry. Several machine learning models, Support Vector Machines (SVC), Random Forest (RF), Decision Tree (DT), Logistic Regression (LR), Multilayer Perceptrons (MLP), implemented solo models two-stage ensemble approach. The trained validated 10-fold cross-validation. SHapley Additive exPlanations (SHAP) employed interpret contributions each feature predictions. Statistical analyses conducted Shapiro-Wilk test normality, Kruskal-Wallis H group differences, Mann-Whitney U Bonferroni correction post-hoc comparisons. SVC + MLP achieved highest performance, an accuracy 89.58%, precision 87.18%, F1-score 88.20%, F-beta score 87.55%. SHAP revealed that glucose, glycine, age consistently important features across all classes, while creatinine various phosphatidylcholines exhibited higher importance class, suggesting their severe DR. XAI particularly ensemble, demonstrated superior performance predicting progression compared models. application facilitates interpretation importance, providing valuable insights into metabolic physiological markers different stages These findings highlight combined techniques detection, interventions, personalized treatment strategies

Language: Английский

---

Unveiling efferocytosis-related signatures through the integration of single-cell analysis and machine learning: a predictive framework for prognosis and immunotherapy response in hepatocellular carcinoma

Frontiers in Immunology, Journal Year: 2023, Volume and Issue: 14

Published: July 27, 2023

Hepatocellular carcinoma (HCC) represents a prominent gastrointestinal malignancy with grim clinical outlook. In this regard, the discovery of novel early biomarkers holds substantial promise for ameliorating HCC-associated mortality. Efferocytosis, vital immunological process, assumes central position in elimination apoptotic cells. However, comprehensive investigations exploring role efferocytosis-related genes (EFRGs) HCC are sparse, and their regulatory influence on immunotherapy targeted drug interventions remain poorly understood.RNA sequencing data characteristics patients were acquired from TCGA database. To identify prognostically significant HCC, we performed limma package conducted univariate Cox regression analysis. Subsequently, machine learning algorithms employed to hub genes. assess landscape different subtypes, CIBERSORT algorithm. Furthermore, single-cell RNA (scRNA-seq) was utilized investigate expression levels ERFGs immune cells explore intercellular communication within tissues. The migratory capacity evaluated using CCK-8 assays, while sensitivity prediction reliability determined through wound-healing assays.We have successfully identified set nine genes, termed EFRGs, that hold potential establishment hepatocellular carcinoma-specific prognostic model. leveraging individual risk scores derived model, able stratify into two distinct groups, unveiling notable disparities terms infiltration patterns response immunotherapy. Notably, model's accurately predict responses substantiated experimental investigations, encompassing assay, CCK8 experiments HepG2 Huh7 cell lines.We constructed an EFRGs model serves as valuable tools assessment decision-making support context chemotherapy.

Language: Английский

---

An immune-related gene prognostic index for predicting prognosis in patients with colorectal cancer

Frontiers in Immunology, Journal Year: 2023, Volume and Issue: 14

Published: July 6, 2023

Background Colorectal cancer (CRC) is one of the most common solid malignant burdens worldwide. Cancer immunology and immunotherapy have become fundamental areas in CRC research treatment. Currently, method generating Immune-Related Gene Prognostic Indices (IRGPIs) has been found to predict patient prognosis as an immune-related prognostic biomarker a variety tumors. However, their role patients with remains mostly unknown. Therefore, we aimed establish IRGPI for evaluation CRC. Methods RNA-sequencing data clinical information were retrieved from The Genome Atlas (TCGA) Expression Omnibus (GEO) databases training validation sets, respectively. Immune-related gene was obtained ImmPort InnateDB databases. weighted co-expression network analysis (WGCNA) used identify hub genes. An then constructed using Cox regression methods. Based on median risk score IRGPI, could be divided into high-risk low-risk groups. To further investigate immunologic differences, set variation (GSVA) studies conducted. In addition, immune cell infiltration related functional differential subsets pathways. Results We identified 49 genes associated CRC, 17 which selected IRGPI. model significantly differentiates survival rates different independent factor correlates clinico-pathological factors such age tumor stage. Furthermore, developed nomogram improve utility score. Immuno-correlation groups revealed distinct (CD4 + T cells resting memory) pathways (macrophages, Type I IFNs responses, iDCs.), providing new insights microenvironment. At last, drug sensitivity that group sensitive 11 resistant 15 drugs. Conclusion Our study established promising predicting patients. This help better understand correlation between immunity perspective personalized treatment

Language: Английский

---

Experimentally validated oxidative stress -associated prognostic signatures describe the immune landscape and predict the drug response and prognosis of SKCM

Frontiers in Immunology, Journal Year: 2024, Volume and Issue: 15

Published: April 10, 2024

Skin Cutaneous Melanoma (SKCM) incidence is continually increasing, with chemotherapy and immunotherapy being among the most common cancer treatment modalities. This study aims to identify novel biomarkers for response in SKCM explore their association oxidative stress.

Language: Английский

---

Unveiling the role of regulatory T cells in the tumor microenvironment of pancreatic cancer through single-cell transcriptomics and in vitro experiments

Frontiers in Immunology, Journal Year: 2023, Volume and Issue: 14

Published: Sept. 11, 2023

In order to investigate the impact of Treg cell infiltration on immune response against pancreatic cancer within tumor microenvironment (TME), and identify crucial mRNA markers associated with cells in cancer, our study aims delve into role anti-tumor cancer.The ordinary transcriptome data for this was sourced from GEO TCGA databases. It analyzed using single-cell sequencing analysis machine learning. To assess level tissues, we employed CIBERSORT method. The identification genes most closely accomplished through implementation weighted gene co-expression network (WGCNA). Our involved various quality control methods, followed by annotation advanced analyses such as trajectory communication elucidate microenvironment. Additionally, categorized two subsets: Treg1 favorable prognosis, Treg2 poor based enrichment scores key genes. Employing hdWGCNA method, these subsets critical signaling pathways governing their mutual transformation. Finally, conducted PCR immunofluorescence staining vitro validate identified genes.Based results analysis, observed significant Subsequently, utilizing WGCNA learning algorithms, ultimately four cell-related (TRGs), among which exhibited correlations occurrence progression cancer. Among them, CASP4, TOB1, CLEC2B were poorer prognosis patients, while FYN showed a correlation better prognosis. Notably, differences found HIF-1 pathway between These conclusions further validated experiments.Treg played microenvironment, presence held dual significance. Recognizing characteristic vital understanding limitations cell-targeted therapies. FYN, close associations infiltrating suggesting involvement functions. Further investigation warranted uncover mechanisms underlying associations. emerged contributing duality cells. Targeting could potentially revolutionize existing treatment approaches

Language: Английский

---

Application progress of artificial intelligence in tumor diagnosis and treatment

Frontiers in Artificial Intelligence, Journal Year: 2025, Volume and Issue: 7

Published: Jan. 7, 2025

The rapid advancement of artificial intelligence (AI) has introduced transformative opportunities in oncology, enhancing the precision and efficiency tumor diagnosis treatment. This review examines recent advancements AI applications across imaging diagnostics, pathological analysis, treatment optimization, with a particular focus on breast cancer, lung liver cancer. By synthesizing findings from peer-reviewed studies published over past decade, this paper analyzes role diagnostic accuracy, streamlining therapeutic decision-making, personalizing strategies. Additionally, addresses challenges related to integration into clinical workflows regulatory compliance. As continues evolve, its oncology promise further improvements patient outcomes, though additional research is needed address limitations ensure ethical effective deployment.

Language: Английский

---

Developing and validating a machine learning model to predict multidrug-resistant Klebsiella pneumoniae-related septic shock

Frontiers in Immunology, Journal Year: 2025, Volume and Issue: 15

Published: Jan. 10, 2025

Multidrug-resistant Klebsiella pneumoniae (MDR-KP) infections pose a significant global healthcare challenge, particularly due to the high mortality risk associated with septic shock. This study aimed develop and validate machine learning-based model predict of MDR-KP-associated shock, enabling early stratification targeted interventions. A retrospective analysis was conducted on 1,385 patients MDR-KP admitted between January 2019 June 2024. The cohort randomly divided into training set (n = 969) validation 416). Feature selection performed using LASSO regression Boruta algorithm. Seven learning algorithms were evaluated, logistic chosen for its optimal balance performance robustness against overfitting. overall incidence shock 16.32% (226/1,385). predictive identified seven key factors: procalcitonin (PCT), sepsis, acute kidney injury, intra-abdominal infection, use vasoactive medications, ventilator weaning failure, mechanical ventilation. demonstrated excellent performance, an area under receiver operating characteristic curve (AUC) 0.906 in 0.865 set. Calibration robust, Hosmer-Lemeshow test results P 0.065 (training) 0.069 (validation). Decision indicated substantial clinical net benefit. presents validated, high-performing offering valuable tool decision-making. Prospective, multi-center studies are recommended further evaluate applicability effectiveness diverse settings.

Language: Английский

---

Development and validation of a nomogram model of lung metastasis in breast cancer based on machine learning algorithm and cytokines

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

Published: April 14, 2025

The relationship between cytokines and lung metastasis (LM) in breast cancer (BC) remains unclear current clinical methods for identifying (BCLM) lack precision, thus underscoring the need an accurate risk prediction model. This study aimed to apply machine learning algorithms key factors BCLM before developing a reliable model centered on cytokines. population-based retrospective included 326 BC patients admitted Second Affiliated Hospital of Xuzhou Medical University September 2018 2023. After randomly assigning training cohort (70%; n = 228) or validation (30%; 98) were identified using Least Absolute Shrinkage Selection Operator (LASSO), Extreme Gradient Boosting (XGBoost) Random Forest (RF) models. Significant visualized with Venn diagram incorporated into nomogram model, performance which was then evaluated according three criteria, namely discrimination, calibration utility plots, receiver operating characteristic (ROC) curves decision curve analysis (DCA). Among cohort, 70 developed LM. A predict 5-year 10-year by incorporating five variables, endocrine therapy, hsCRP, IL6, IFN-ɑ TNF-ɑ. For cohorts had AUC values 0.786 (95% CI: 0.691-0.881) 0.627 0.441-0.813), respectively, while corresponding 0.687 0.528-0.847) 0.797 0.605-0.988), respectively. ROC further confirmed model's strong discriminative ability, plots indicated that predicted observed outcomes good agreement both cohorts. Finally, DCA demonstrated effectiveness practice. Using algorithms, this aa could effectively identify who at higher LM, providing valuable tool decision-making settings.

Language: Английский

---

Bayesian‐optimized deep learning for identifying essential genes of mitophagy and fostering therapies to combat drug resistance in human cancers

Journal of Cellular and Molecular Medicine, Journal Year: 2025, Volume and Issue: 29(2)

Published: Jan. 1, 2025

Abstract Dysregulated mitophagy is essential for mitochondrial quality control within human cancers. However, identifying hub genes regulating and developing mitophagy‐based treatments to combat drug resistance remains challenging. Herein, BayeDEM (Bayesian‐optimized Deep learning Essential of Mitophagy) was proposed such a task. After Bayesian optimization, demonstrated its excellent performance in critical osteosarcoma (area under curve [AUC] ROC: 98.96%; AUC PR curve: 100%). CERS1 identified as the most gene (mean (|SHAP value|): 4.14). Inhibition sensitized cisplatin‐resistant cells cisplatin, restricting their growth, proliferation, invasion, migration colony formation inducing apoptosis. Mechanistically, inhibition restricted destroy cells, including membrane potential loss unfavourable dynamics, rendering them susceptible cisplatin‐induced More importantly, facilitated immunosuppressive microenvironment by significantly modulating T‐cell differentiation, adhesion antigen presentation, mainly affects malignant osteoblasts early‐mid developmental stage. Immunologically, potentially modulated MIF signalling transmission between B DCs, CD8+ T NK monocytes through MIF‐(CD74 + CXCR4) receptor–ligand interaction, thereby biological functions these immune cells. Collectively, emerged promising tool oncologists identify pivotal governing mitophagy, enabling mitophagy‐centric therapeutic strategies counteract resistance.

Language: Английский

---