Can machine learning models support physicians in systemic lupus erythematosus diagnosis. results from a monocentric cohort

Background: Several cardiovascular (CV) risk algorithms are available to predict CV events in the general population. However, their performance in patients with rheumatoid arthritis (RA) might differ from the general population. This cross-sectional multicentre study aimed to estimate the 10-year CV risk using two different algorithms in a large RA cohort and in patients with osteoarthritis (OA). Methods: In a consecutive series of RA patients and matched OA controls without prior CV events, clinical and serologic data and traditional CV risk factors were recorded. The 10-year CV risk was assessed with the Systematic COronary Risk Evaluation (SCORE) and the "Progetto Cuore" algorithms. Results: 1,467 RA patients and 342 OA subjects were included. RA patients were more frequently diabetic (9.9% vs 6.4%; p=0.04) and smokers (20.4% vs 12.5%; p=0.002) but had lower prevalence of obesity (15% vs 21%; p=0.003). Dyslipidaemia was more prevalent in OA (32.5% vs 21.7%; p<0.0001). The 10-year estimated CV risk was 1.6% (95%CI 1.3-1.9) in RA and 1.4% (95%CI 1.3-1.6) in OA (p=0.002) according to SCORE and 6.5% (95%CI 6.1-6.9) in RA and 4.4% (95%CI 3.9-5.1) in OA (p<0.001) according to "Progetto Cuore". Regardless of the score used, RA patients had a 3- to-4-fold increased 10-year risk of CV events compared to OA subjects. Conclusion: RA patients have a significantly higher 10-year risk of CV events than OA subjects. In addition to effective disease control and joint damage prevention, specific protective measures targeting modifiable traditional CV risk factors should be implemented in RA.