Artificial Intelligence in Action: A Comprehensive Review on Machine and Deep Learning Methods in Sjögren's Syndrome Diagnosis

Abstract

Sjögren's syndrome (SS) is a chronic autoimmune disorder characterized by heterogeneous manifestations and often delayed diagnosis due to its nonspecific symptoms and reliance on invasive tests. This review highlights the transformative role of artificial intelligence (AI), particularly machine learning (ML), and deep learning (DL), in enhancing SS diagnostic accuracy across diverse clinical settings. Artificial intelligence–driven models have demonstrated remarkable performance in histopathological analysis, automating the evaluation of salivary gland biopsies and minimizing observer variability. Imaging diagnostics, such as salivary gland ultrasonography and computed tomography (CT), have benefited from DL models that outperform inexperienced radiologists in detecting glandular abnormalities. When paired with ML classifiers, noninvasive approaches using Raman spectroscopy and tongue imaging offer promising alternatives to traditional diagnostics. Moreover, AI applications in genomic and metabolomic profiling have unveiled novel biomarkers and molecular signatures for SS. In primary-care, ML models trained on electronic health records (EHRs) show potential in early case identification and reducing referral delays. These innovations collectively illustrate AI's capability to unify disparate data sources from visual and molecular to clinical and support timely, personalized diagnostics. The inference drawn from this review is that AI integration across multiple diagnostic modalities can bridge existing gaps in SS detection, making diagnosis faster, more objective, and accessible even in resource-limited settings.