ACR Meeting Abstracts

ACR Meeting Abstracts

Abstract Number: 0628

Α user-friendly machine-learning tool for early damage prediction in patients with Systemic Lupus Erythematosus

Panagiotis Garantziotis1, Dionysis Nikolopoulos2, Spyridon Katechis3, Alp Temiz4, Danae-Mona Nöthling1, Christina Adamichou5, Prodromos Sidiropoulos6, Georg Schett7, Fanouriakis Antonis3, Dimitrios Boumpas8 and George Bertsias9, 1Department of Medicine 3 - Rheumatology and Immunology, Friedrich-Alexander-Universität (FAU) Erlangen-Nürnberg and Uniklinikum Erlangen, Erlangen, Germany, 2Karolinska Institutet and Karolinska University Hospital, Division of Rheumatology, Department of Medicine Solna, Stockholm, Sweden, 3National and Kapodistrian University of Athens Medical School, Athens, Greece, Athens, Greece, 4Department of Medicine 3 - Rheumatology and Immunology, Friedrich-Alexander-Universität Erlangen-Nürnberg and University Hospital Erlangen, Erlangen, Germany, 5Medical School University of Crete, Heraklion, Greece, Heraklion, Germany, 6Department of Rheumatology, Clinical Immunology and Allergy, University Hospital of Heraklion, Heraklion, Greece, 7Uniklinikum Erlangen, Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany, Erlangen, Germany, 8Joint Rheumatology Program, University of Athens Medical School, Athens, Greece, 9Rheumatology and Clinical Immunology, University Hospital of Heraklion and University of Crete Medical School and Foundation for Research and Technology-Hellas (FORTH), Infections and Immunity, Institute of Molecular Biology and Biotechnology, Heraklion, Greece

Meeting: ACR Convergence 2025

Keywords: , ,

Session Information

Date: Sunday, October 26, 2025

Title: (0593–0640) Systemic Lupus Erythematosus – Diagnosis, Manifestations, & Outcomes Poster I

Session Type: Poster Session A

Session Time: 10:30AM-12:30PM

Background/Purpose: Organ damage drives morbidity and mortality in systemic lupus erythematosus (SLE), yet no validated tools exist to predict early damage. We aimed to develop and validate a machine learning model using standard clinical features to forecast organ damage within five years since diagnosis.

Methods: We analyzed a discovery cohort of 914 SLE patients and an independent validation cohort of 50 SLE patients to associate clinical and immunological features with increases in the SLICC/ACR Damage Index within five years of diagnosis (physician-based). The discovery cohort was split into training (70%) and test (30%) sets. Clinically curated panels of classification and non-criteria features were used to build LASSO logistic regression models, with hyper-parameter tuning via 10-fold cross-validation. Performance was assessed by AUC, sensitivity, specificity, and accuracy. Cutoffs were selected by maximizing the F1 score or Youden’s index.

Results: A LASSO logistic regression model, integrating both classification criteria and non-criteria clinical features, outperformedexisting classification criteria in predicting early organ damage, with an AUC of 0.86 (95% CI 95% CI 0.76 to 0.97) in the external validation cohort. Key damage predictors were neurologic involvement (SLICC-2012), class III/IV lupus nephritis (EULAR/ACR-2019), and myocarditis. The model achieved a specificity of 0.898 (95% CI: 0.78–0.95) as a binary classifier. When simplified into an 8 variable scoring system, a threshold of > 3 yielded an AUC of 0.86 (95% CI: 0.75–0.96).

Conclusion: We developed and validated an interpretable tool for predicting early organ damage in SLE. Pending further validation, this tool may facilitate individualized risk assessment and damage-preventative strategies including the timely use of biologic treatments.

Disclosures: P. Garantziotis: None; D. Nikolopoulos: None; S. Katechis: None; A. Temiz: None; D. Nöthling: None; C. Adamichou: None; P. Sidiropoulos: None; G. Schett: Cabaletta, 6, Eli Lilly, 6, Janssen, 6, Kyverna, 6, Novartis, 6, UCB, 6; F. Antonis: None; D. Boumpas: None; G. Bertsias: AbbVie, 6, AstraZeneca, 2, 5, 6, Eli Lilly, 2, 6, GSK, 2, 5, 6, MSD, 5, Novartis, 2, 6, Otsuka, 6, Pfizer, 6.

To cite this abstract in AMA style:

Garantziotis P, Nikolopoulos D, Katechis S, Temiz A, Nöthling D, Adamichou C, Sidiropoulos P, Schett G, Antonis F, Boumpas D, Bertsias G. Α user-friendly machine-learning tool for early damage prediction in patients with Systemic Lupus Erythematosus [abstract]. Arthritis Rheumatol. 2025; 77 (suppl 9). https://acrabstracts.org/abstract/%ce%b1-user-friendly-machine-learning-tool-for-early-damage-prediction-in-patients-with-systemic-lupus-erythematosus/. Accessed .

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/%ce%b1-user-friendly-machine-learning-tool-for-early-damage-prediction-in-patients-with-systemic-lupus-erythematosus/