{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,17]],"date-time":"2026-03-17T16:15:32Z","timestamp":1773764132645,"version":"3.50.1"},"reference-count":26,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,3,17]],"date-time":"2026-03-17T00:00:00Z","timestamp":1773705600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Immunol."],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>Rheumatoid arthritis (RA) is a heterogeneous autoimmune disease where cytokine-driven dysregulation of gene networks poses a significant challenge for accurate diagnosis and targeted therapy. While transcriptomic data hold immense promise, most machine learning models lack the interpretability to decipher the underlying biological mechanisms, particularly the specific roles of key cytokines.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>We developed GeneCytNet, a novel deep learning framework that integrates a Variational Autoencoder (VAE) for nonlinear feature compression with a Graph Attention Network (GAT) to model gene-gene interactions. The model was developed on a synthetic cohort of 240 RA and 120 healthy control samples, with an independent holdout cohort of 100 RA and 50 controls, each with 15,000 gene features, designed as a robust proof-of-concept. Performance was benchmarked against classical models, and generalizability was assessed via cross-validation and the independent holdout. Crucially, we introduced in silico cytokine perturbation experiments to simulate the effect of modulating IL-6, TNF-\u03b1, and IL-1\u03b2 responsive gene modules on RA risk prediction.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>GeneCytNet achieved superior classification performance, with a test AUC of 0.962 \u00b1 0.005, accuracy of 0.914 \u00b1 0.007, and an F1-score of 0.915 \u00b1 0.006, outperforming all baseline models. Cross-validation confirmed robustness (mean AUC = 0.957 \u00b1 0.006). The perturbation experiments provided mechanistically interpretable insights, revealing that the IL-6\u2013responsive module had the most significant effect on RA probability (+0.12 \u00b1 0.03), followed by TNF-\u03b1 (+0.08 \u00b1 0.02) and IL-1\u03b2 (+0.06 \u00b1 0.02). This hierarchy of cytokine effect sizes aligns with established clinical evidence.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>\n                      GeneCytNet demonstrates that advanced, interpretable deep learning can simultaneously achieve high diagnostic accuracy and generate testable biological hypotheses. By functioning as a\n                      <jats:italic>virtual patient simulator<\/jats:italic>\n                      , our framework bridges the gap between prediction and mechanism, offering a powerful tool for precision diagnostics, biomarker discovery, and the design of cytokine-targeted therapies in RA and other complex diseases.\n                    <\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fimmu.2026.1738625","type":"journal-article","created":{"date-parts":[[2026,3,17]],"date-time":"2026-03-17T14:19:34Z","timestamp":1773757174000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["GeneCytNet: an interpretable deep learning framework for rheumatoid arthritis classification and in silico cytokine perturbation modeling"],"prefix":"10.3389","volume":"17","author":[{"given":"Chen","family":"Chen","sequence":"first","affiliation":[{"name":"Department of Orthopedics, the First Affiliated Hospital of Xiamen University, School of Medicine, Xiamen University","place":["Xiamen, Fujian, China"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Dagang","family":"Li","sequence":"additional","affiliation":[{"name":"Department of Orthopedics, the First Affiliated Hospital of Xiamen University, School of Medicine, Xiamen University","place":["Xiamen, Fujian, China"]}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Lujia","family":"Xu","sequence":"additional","affiliation":[{"name":"Xiamen University Tan Kah Kee College","place":["Xiamen, Zhangzhou, China"]}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1965","published-online":{"date-parts":[[2026,3,17]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","DOI":"10.3390\/app15126896","article-title":"Articular cartilage: structure, biomechanics, and the potential of conventional and advanced diagnostics","volume":"15","author":"Karpi\u0144ski","year":"2025","journal-title":"Appl Sci"},{"key":"B2","doi-asserted-by":"publisher","first-page":"e0317364","DOI":"10.1371\/journal.pone.0317364","article-title":"Cellular senescence-associated genes in rheumatoid arthritis: Identification and functional analysis","volume":"20","author":"Ao","year":"2025","journal-title":"PloS One"},{"key":"B3","doi-asserted-by":"publisher","first-page":"e0326876","DOI":"10.1371\/journal.pone.0326876","article-title":"Signatures of tumor microenvironment-related genes and long noncoding RNAs predict poor prognosis in osteosarcoma","volume":"20","author":"Lu","year":"2025","journal-title":"PloS One"},{"key":"B4","doi-asserted-by":"publisher","DOI":"10.3390\/biomedicines13092171","article-title":"Integrating machine learning and multi-omics to explore neutrophil heterogeneity","volume":"13","author":"Lin","year":"2025","journal-title":"Biomedicines"},{"key":"B5","doi-asserted-by":"publisher","first-page":"e18254","DOI":"10.1111\/jcmm.18254","article-title":"Bayesian-optimized deep learning for identifying essential genes of mitophagy and fostering therapies to combat drug resistance in human cancers","volume":"29","author":"Jin","year":"2025","journal-title":"J Cell Mol Med"},{"key":"B6","doi-asserted-by":"publisher","DOI":"10.1155\/mi\/6149780","article-title":"Identification of ifitm1 as a pivotal gene in mouse spinal cord injury using comprehensive machine learning algorithms","volume":"2025","author":"Yao","year":"2025","journal-title":"Mediators Inflamm"},{"key":"B7","doi-asserted-by":"publisher","DOI":"10.3390\/ijms26167789","article-title":"Exploring the heterogeneity of cancer-associated fibroblasts via development of patient-derived cell culture of breast cancer","volume":"26","author":"Ilyina","year":"2025","journal-title":"Int J Mol Sci"},{"key":"B8","doi-asserted-by":"publisher","first-page":"420","DOI":"10.1007\/s12672-024-01314-4","article-title":"Peng W","volume":"15","author":"Wu","year":"2024","journal-title":"A cancer-assoc fibroblasts relat Risk score (CAFscore) helps to guide prognosis Pers Treat Glioblastoma. Discov Oncol"},{"key":"B9","doi-asserted-by":"publisher","DOI":"10.3390\/jpm14050537","article-title":"Interactions and trends of interleukins, PAI-1, CRP, and TNF-\u03b1 in inflammatory responses during the perioperative period of joint arthroplasty: implications for pain management\u2014a narrative review","volume":"14","author":"Cocea","year":"2024","journal-title":"J Pers Med"},{"key":"B10","doi-asserted-by":"publisher","DOI":"10.2147\/JMDH.S537507","article-title":"Integrated bioinformatics and experimental validation reveal macrophage polarization-related biomarkers for osteoarthritis diagnosis","volume":"2025","author":"He","year":"2025","journal-title":"J Multidiscip Healthc"},{"key":"B11","doi-asserted-by":"publisher","first-page":"126","DOI":"10.1038\/s41540-024-00454-1","article-title":"Multiscale, mechanistic model of Rheumatoid Arthritis to enable decision making in late stage drug development","volume":"10","author":"Bedathuru","year":"2024","journal-title":"NPJ Syst Biol Appl"},{"key":"B12","doi-asserted-by":"publisher","DOI":"10.3390\/jpm15010017","article-title":"Patient-specific variability in interleukin-6 and myeloperoxidase responses in osteoarthritis: insights from synthetic data and clustering analysis","volume":"15","author":"Coleman","year":"2025","journal-title":"J Pers Med"},{"key":"B13","doi-asserted-by":"publisher","first-page":"25158","DOI":"10.1038\/s41598-025-10290-w","article-title":"Multomic analysis reveals the potential of LAG3 as a prognostic and immune biomarker and its validation in osteosarcoma","volume":"15","author":"Cao","year":"2025","journal-title":"Sci Rep"},{"key":"B14","doi-asserted-by":"publisher","first-page":"4549","DOI":"10.1038\/s41598-024-55187-2","article-title":"Bioinformatics analysis for the identification of Sprouty-related EVH1 domain-containing protein 3 expression and its clinical significance in thyroid carcinoma","volume":"14","author":"Zhang","year":"2024","journal-title":"Sci Rep"},{"key":"B15","doi-asserted-by":"publisher","DOI":"10.1016\/j.bioactmat.2025.05.004","article-title":"Advancement in smart bone implants: the latest multifunctional strategies and synergistic mechanisms for tissue repair and regeneration","volume":"51","author":"Li","year":"2025","journal-title":"Bioact Mater"},{"key":"B16","doi-asserted-by":"publisher","first-page":"e70043","DOI":"10.1111\/cpr.70043","article-title":"Intelligent manufacturing for osteoarthritis organoids","volume":"2025","author":"Lyu","year":"2025","journal-title":"Cell Prolif"},{"key":"B17","doi-asserted-by":"publisher","DOI":"10.3389\/fimmu.2024.1381272","article-title":"An overview of multi-omics technologies in rheumatoid arthritis: applications in biomarker and pathway discovery","volume":"15","author":"Gong","year":"2024","journal-title":"Front Immunol"},{"key":"B18","doi-asserted-by":"publisher","DOI":"10.1093\/rheumatology\/keab521","article-title":"Early prediction of clinical response to anti-TNF treatment using multi-omics and machine learning in rheumatoid arthritis","volume":"61","author":"Yoosuf","year":"2022","journal-title":"Rheumatology"},{"key":"B19","doi-asserted-by":"publisher","DOI":"10.3389\/fgene.2023.1210722","article-title":"Single-cell sequencing and transcriptome analyses in the construction of a liquid\u2013liquid phase separation-associated gene model for rheumatoid arthritis","volume":"14","author":"Tai","year":"2023","journal-title":"Front Genet"},{"key":"B20","doi-asserted-by":"publisher","DOI":"10.3389\/fgene.2021.604714","article-title":"Coupling of co-expression network analysis and machine learning validation unearthed potential key genes involved in rheumatoid arthritis","volume":"12","author":"Xiao","year":"2021","journal-title":"Front Genet"},{"key":"B21","doi-asserted-by":"publisher","DOI":"10.3390\/bioengineering10020173","article-title":"Machine learning methods for cancer classification using gene expression data: A review","volume":"10","author":"Alharbi","year":"2023","journal-title":"Bioengineering"},{"key":"B22","doi-asserted-by":"publisher","DOI":"10.3390\/ijms26062757","article-title":"The identification of novel therapeutic biomarkers in rheumatoid arthritis: a combined bioinformatics and integrated multi-omics approach","volume":"26","author":"Tariq","year":"2025","journal-title":"Int J Mol Sci"},{"key":"B23","doi-asserted-by":"publisher","DOI":"10.3389\/fdata.2024.1406365","article-title":"Machine learning-based remission prediction in rheumatoid arthritis patients treated with biologic disease-modifying anti-rheumatic drugs: findings from the Kuwait rheumatic disease registry","volume":"7","author":"Alsaber","year":"2024","journal-title":"Front Big Data"},{"key":"B24","doi-asserted-by":"publisher","first-page":"p.e60635","DOI":"10.1371\/journal.pone.0099812","article-title":"Development of a multi-biomarker disease activity test for rheumatoid arthritis","volume":"8","author":"Centola","year":"2013","journal-title":"PloS One"},{"key":"B25","doi-asserted-by":"publisher","DOI":"10.34133\/research.0776","article-title":"Machine learning-assisted development of injectable, mechanically robust, and energy metabolism-modulating brushite cements","volume":"8","author":"Liu","year":"2025","journal-title":"Research"},{"key":"B26","doi-asserted-by":"publisher","DOI":"10.3389\/fimmu.2025.1574783","article-title":"Integrated multi-omics for potential biomarkers and molecular mechanism of persistent inflammatory refractory rheumatoid arthritis","volume":"16","author":"Zhang","year":"2025","journal-title":"Front Immunol"}],"container-title":["Frontiers in Immunology"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fimmu.2026.1738625\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,17]],"date-time":"2026-03-17T14:19:34Z","timestamp":1773757174000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fimmu.2026.1738625\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,17]]},"references-count":26,"alternative-id":["10.3389\/fimmu.2026.1738625"],"URL":"https:\/\/doi.org\/10.3389\/fimmu.2026.1738625","relation":{},"ISSN":["1664-3224"],"issn-type":[{"value":"1664-3224","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,3,17]]},"article-number":"1738625"}}