# Artificial Intelligence in Rheumatology: From Algorithms to Clinical Impact in Osteoporosis and Chronic Inflammatory Rheumatic Diseases

## TL;DR

This review explores how AI is being used in rheumatology, particularly for osteoporosis and inflammatory diseases, highlighting its potential and current limitations in clinical practice.

## Contribution

The paper provides a comprehensive review of AI applications in rheumatology, emphasizing methodological robustness and clinical applicability.

## Key findings

- AI models show strong performance in predicting bone mineral density and fractures in osteoporosis.
- AI improves imaging interpretation for sacroiliitis in chronic inflammatory rheumatic diseases.
- Hybrid AI models combining imaging, clinical, and biological data are promising but require more validation.

## Abstract

Background: Artificial intelligence (AI) is transforming medicine by supporting data-driven diagnosis, prognosis, and personalized care. In rheumatology, AI applications are rapidly expanding in imaging, disease monitoring, and therapeutic decision support. This review aimed to summarize current evidence on AI in osteoporosis and chronic inflammatory rheumatic diseases, with a focus on methodological robustness and clinical applicability. Methods: A narrative review was conducted following SANRA criteria. PubMed and the Cochrane Library were systematically searched for studies published between January 2015 and July 2025 using MeSH terms and free-text keywords related to AI, osteoporosis, and inflammatory rheumatic diseases. A total of 323 articles were included. Results: Machine learning and deep learning models show strong performance in osteoporosis for predicting bone mineral density (BMD), bone loss, and fractures. In chronic inflammatory rheumatic diseases, AI improves imaging interpretation, particularly for sacroiliitis. AI tools also demonstrate potential for predicting disease risk and activity, diagnostic support and treatment response. Hybrid models combining imaging, clinical, and biological data appear particularly promising. However, most studies rely on retrospective single-center datasets, with limited external validation, suboptimal explainability, and scarce evidence of real-world implementation. Conclusions: AI holds significant promise for advancing diagnosis and personalized management in osteoporosis and rheumatic diseases. However, major challenges persist, including heterogeneous data quality, inconsistent methodological reporting, limited clinical validation, and barriers to integration into routine practice. Bridging the gap between algorithmic performance and clinical impact will require prospective studies, robust validation frameworks, and strategies to build trust among clinicians and patients.