Razionale scientifico

Artificial intelligence is reshaping medicine at a pace that rheumatology cannot afford to ignore. The specialty is characterised by multisystem diseases, heterogeneous phenotypes, and composite outcome measures that have long outgrown the analytical capacity of conventional statistics. Machine learning, large language models, and digital biomarkers now offer tools capable of navigating this complexity — provided they are implemented with rigour and a clear clinical purpose. The clinical stakes are high across multiple disease areas. In rheumatoid arthritis, machine learning models trained on real-world cohorts are stratifying patients by likelihood of biologic response and predicting radiographic progression, while video-based tools such as motion capturing are objectifying joint assessment in ways that reduce examiner subjectivity and enable remote monitoring. In spondyloarthritis, diagnostic delay still averages seven to ten years, and AI-assisted screening of primary care records and imaging repositories holds genuine promise for earlier referral; telemonitoring platforms integrating patient-reported outcomes and wearable data — as illustrated by Prof. Antoni Chan's experience in axial SpA — are already enabling treat-to-target strategies without frequent in-person visits. In connective tissue diseases, pulmonary fibrosis remains one of the gravest complications, and deep learning applied to high-resolution CT is now achieving radiologist-level performance in detecting and quantifying fibrotic involvement, while multimodal prognostic models integrating imaging, pulmonary function, and remote sensor data are beginning to guide antifibrotic therapy decisions. Across all these contexts, well-designed LLM pipelines can structure and synthesise the scattered clinical narratives — radiology reports, MDT notes, outpatient letters — that currently make longitudinal management unnecessarily fragmented. Yet a fundamental gap persists between this promise and everyday clinical practice. Most rheumatologists have limited exposure to the methodological foundations that determine whether a model is trustworthy and reproducible. EULAR is revising its Points-to-Consider on AI, editorial standards for LLM-assisted writing are being debated, and regulatory frameworks are still catching up. RheumatologIA was conceived to bridge this gap — offering, in two intensive days, a trajectory from Python and machine learning foundations through to prompt engineering, digital phenotyping, and the ethics of AI in publishing. The faculty unites EULAR leadership, digital rheumatology pioneers, biomedical physicists, and AI developers around a single mission: to equip clinicians with the critical literacy and hands-on skills needed to evaluate, use, and shape AI responsibly in the care of their patients.

Luogo

Sala Convegni The Nicolaus Hotel
Via Cardinale Agostino Ciasca, 27, Bari (BA)

Programma

Responsabili del corso

• Florenzo Iannone

  Responsabile scientifico

  Specializzazione: Reumatologia

  Struttura: Università di Bari

  Ruolo: Prof. Ordinario Reumatologia

Relatori e docenti

• Orazio Angelini

  Relatore

  Specializzazione: Fisica

  Struttura: Amazon Research, King's College London

  Ruolo: Researcher

• Diego Benavent Nunez

  Relatore

  Specializzazione: Reumatologia

  Struttura: Hospital Universitario La Paz, Madrid

  Ruolo: Medical Doctor

• Emre Bilgin

  Moderatore

  Specializzazione: Reumatologia

  Struttura: Hacettepe University, Turchia

  Ruolo: Rheumatologist

• Marc Blanchard

  Tutor

  Specializzazione: Reumatologia

  Struttura: Lausanne University Hospital (CHUV)

  Ruolo: Ph.D. candidate in digital health solutions for chronic rheumatic conditions

• Antoni Chan

  Relatore

  Specializzazione: Reumatologia

  Struttura: Royal Berkshire NHS Foundation Trust, UK

  Ruolo: Consultant Rheumatologist

Con il contributo non vincolante di

Domande frequenti