Answers are grounded in this paper and cite their sources.
Cutaneous lupus erythematosus (CLE) poses challenges in differential diagnosis, subtype classification, and longitudinal management because of its heterogeneous phenotypes and variable association with systemic lupus erythematosus (SLE). This narrative review evaluates artificial intelligence (AI) evidence by clinical task, data modality, disease relevance, and validation maturity. Direct CLE studies provide early evidence for image-based lesion and subtype classification, including a multicentre multimodal system that improved dermatologists’ diagnostic accuracy. A separate proof-of-concept study used natural language processing (NLP) to estimate Cutaneous Lupus Erythematosus Disease Area and Severity Index (CLASI) activity and damage scores from physical-examination notes. This text-based result should not be interpreted as automated severity scoring from clinical photographs: image-based lesion segmentation and CLASI estimation remain proposed applications. CLE transcriptomic, single-cell, and spatial studies reveal subtype-related B-cell signatures, interferon-associated keratinocyte states, specialized fibroblasts, and altered cell-cell communication; these findings are biological foundations for future models rather than validated patient-level prediction systems. Across applications, the level of CLE relevance is distinct from the maturity of validation. Current limitations include small or incompletely characterized disease-specific subsets, inconsistent image acquisition, uncertain independence of test cohorts, and limited prospective assessment of clinical utility. Progress requires transparent reporting of CLE-specific denominators, patient-level data separation, independent external validation, calibration, pigmentation-aware subgroup assessment, and clinician-supervised prospective evaluation. AI should therefore be regarded as an investigational decision-support approach for CLE, with clinical benefit established through disease-specific testing.
No takes yet. Share an insight, caveat, or question.
Xue et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: