The abstract is read at the publisher; the summary is JClub's.
Machine learning classifiers using UK Biobank serum proteomic profiles identified pre-existing systemic lupus erythematosus with ~90% sensitivity at 95% specificity in immunomodulated patients.
- Why it matters: Proteome-wide risk models for lupus are underexplored, limiting objective identification and early prediction beyond genetic risk stratification.
- What they did: In UK Biobank, 44,173 participants with proteomic data (including 383 lupus cases) were used to build proteomic linear and machine learning models, then compared with polygenic risk scores and replicated in independent Sweden (HPA) and China cohorts.
- The result: Lupus had the most dysregulated proteins among autoimmune diseases and ~70% of proteomic associations replicated with moderate effect-size correlation; the machine learning model generalized to future lupus prediction and highlighted SCARB2, SOD2, CD302, Galectin-9, and GGT5.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.