Identifying systemic lupus erythematosus from serum proteomic profiles using machine learning and genetic risk stratification - PubMed
3 hours ago
- #systemic lupus erythematosus
- #proteomics
- #machine learning
- Study developed machine learning models to identify systemic lupus erythematosus (SLE) from serum proteomic profiles.
- Analyzed 44,173 UK Biobank participants, including 383 lupus patients, and replicated findings in independent cohorts from Sweden and China.
- Lupus showed the highest number of dysregulated proteins among autoimmune diseases, clustering closely with rheumatoid arthritis.
- Machine learning models outperformed linear models in identifying preexisting lupus and predicting future cases, achieving ~90% sensitivity at 95% specificity.
- Key proteins influencing lupus identification included SCARB2, SOD2, CD302, Galectin-9, and GGT5, suggesting potential novel biomarkers.