Improving musculoskeletal care with AI enhanced triage through data driven screening of referral letters

npj Digital Medicine, Journal Year: 2025, Volume and Issue: 8(1)

Published: Feb. 14, 2025

Abstract Musculoskeletal complaints account for 30% of GP consultations, with many referred to rheumatology clinics via letters. This study developed a Machine Learning (ML) pipeline prioritize referrals by identifying rheumatoid arthritis (RA), osteoarthritis, fibromyalgia, and patients requiring long-term care. Using 8044 referral letters from 5728 across 12 clinics, we trained validated ML models in two large centers tested their generalizability the remaining ten. The were robust, RA achieving an AUC-ROC 0.78 (CI: 0.74–0.83), osteoarthritis 0.71 0.67–0.74), fibromyalgia 0.81 0.77–0.85), chronic follow-up 0.63 0.61–0.66). RA-classifier outperformed manual systems, as it prioritised over non-RA cases ( P < 0.001 ), while system could not differentiate between two. other classifiers showed similar prioritisation improvements, highlighting potential enhance care efficiency, reduce clinician workload, facilitate earlier specialized Future work will focus on building clinical decision-support tools.

Language: Английский

---

Review of 2024 publications on the applications of artificial intelligence in rheumatology

Clinical Rheumatology, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 27, 2025

Language: Английский

---

Applications of Artificial Intelligence in Vasculitides: A Systematic Review

ACR Open Rheumatology, Journal Year: 2025, Volume and Issue: 7(3)

Published: March 1, 2025

Objective Vasculitides are rare inflammatory disorders that sometimes can be difficult to diagnose due their diverse presentations. This review examines the use of artificial intelligence (AI) improve diagnosis and outcome prediction in vasculitis. Methods A systematic search PubMed, Embase, Web Science, Institute Electrical Electronics Engineers Xplore, Scopus identified relevant studies from 2000 2024. AI applications were categorized by data type (clinical, imaging, textual) task (diagnosis or prediction). Studies assessed for risk bias using Prediction Model Risk Bias Assessment Tool Quality Diagnostic Accuracy Studies–2. Results total 46 included. models achieved high diagnostic performance Kawasaki disease, with sensitivities up 92.5% specificities 97.3%. Predictive complications, such as intravenous Ig resistance showed areas under curves between 0.716 0.834. Other vasculitis types, especially those imaging data, less studied often limited small datasets. Conclusion The current literature shows algorithms enhance prediction, deep‐ machine‐learning showing promise disease. However, broader datasets, more external validation, integration newer like large language needed advance clinical applicability across different types.

Language: Английский

---

Evaluating the performance of large language models in health education for patients with ankylosing spondylitis/spondyloarthritis: a cross-sectional, single-blind study in China

BMJ Open, Journal Year: 2025, Volume and Issue: 15(3), P. e097528 - e097528

Published: March 1, 2025

Objectives To evaluate the potential of large language models (LLMs) in health education for patients with ankylosing spondylitis (AS)/spondyloarthritis (SpA), focusing on accuracy information transmission, patient acceptance and performance differences between different models. Design Cross-sectional, single-blind study. Setting Multiple centres China. Participants 182 volunteers, including 4 rheumatologists 178 AS/SpA. Primary secondary outcome measures Scientificity, precision accessibility content answers provided by LLMs; answers. Results LLMs performed well terms scientificity, accessibility, ChatGPT-4o Kimi outperforming traditional guidelines. Most AS/SpA showed a higher level understanding responses from LLMs. Conclusions have significant medical knowledge transmission education, making them promising tools future practice.

Language: Английский

---