Comparisons of Anti-dsDNA Antibody Detection Methods by Chemiluminescent Immunoassay and Enzyme-Linked Immunosorbent Assay in Systemic Lupus Erythematosus.

This study aimed to compare the test results of anti-double-stranded DNA (anti-dsDNA) antibodies obtained using chemiluminescent immunoassay (CIA) and enzyme-linked immunosorbent assay (ELISA), and investigate predictors of inconsistent results. This retrospective study included 502 patients who underwent CIA and ELISA to determine their anti-dsDNA antibody values within a year. We compared the diagnostic power for SLE, disease activity, and predictive power for lupus nephritis (LN). A multivariate analysis was performed to determine the predictors of inconsistencies. CIA and ELISA were moderately correlated in terms of their consistency (Cronbach's α = 0.571), and yielded comparably favorable results in terms of SLE diagnostic power and SLE disease activity. However, if the patient had LN, CIA displayed higher predictive power than ELISA (0.620 vs. 0.555, p = 0.026). Compared with the CIA/ELISA double-positive group, the inconsistent group had lower anti-C1q circulating immune complexes (CIC) antibody values (OR: 0.42, 95% CI: 0.18-0.94, p = 0.036), and lower SLEDAI scores (≥4) (OR: 0.33, 95% CI: 0.14-0.79, p = 0.013). Anti-dsDNA antibody detection with CIA exhibited higher predictability for diagnosing LN than did ELISA. In the event of inconsistencies between anti-dsDNA methods, SLE disease activity and CIC test values should be considered simultaneously.

Application of Supervised Machine Learning to Recognize Competent Level and Mixed Antinuclear Antibody Patterns Based on ICAP International Consensus.

BACKGROUND: Antinuclear antibody pattern recognition is vital for autoimmune disease diagnosis but labor-intensive for manual interpretation. To develop an automated pattern recognition system, we established machine learning models based on the International Consensus on Antinuclear Antibody Patterns (ICAP) at a competent level, mixed patterns recognition, and evaluated their consistency with human reading. METHODS: 51,694 human epithelial cells (HEp-2) cell images with patterns assigned by experienced medical technologists collected in a medical center were used to train six machine learning algorithms and were compared by their performance. Next, we choose the best performing model to test the consistency with five experienced readers and two beginners. RESULTS: The mean F1 score in each classification of the best performing model was 0.86 evaluated by Testing Data 1. For the inter-observer agreement test on Testing Data 2, the average agreement was 0.849 (κ) among five experienced readers, 0.844 between the best performing model and experienced readers, 0.528 between experienced readers and beginners. The results indicate that the proposed model outperformed beginners and achieved an excellent agreement with experienced readers. CONCLUSIONS: This study demonstrated that the developed model could reach an excellent agreement with experienced human readers using machine learning methods.