Cardiovascular Risk Prediction in Ankylosing Spondylitis: From Traditional Scores to Machine Learning Assessment

Abstract

Introduction: The performance of seven cardiovascular (CV) risk algorithms is evaluated in a multicentric cohort of ankylosing spondylitis (AS) patients. Performance and calibration of traditional CV predictors have been compared with the novel paradigm of machine learning (ML).

Methods: A retrospective analysis of prospectively collected data from an AS cohort has been performed. The primary outcome was the first CV event. The discriminatory ability of the algorithms was evaluated using the area under the receiver operating characteristic (ROC) curve (AUC), which is like the concordance-statistic (c-statistic). Three ML techniques were considered to calculate the CV risk: support vector machine (SVM), random forest (RF), and k-nearest neighbor (KNN).

Results: Of 133 AS patients enrolled, 18 had a CV event. c-statistic scores of 0.71, 0.61, 0.66, 0.68, 0.66, 0.72, and 0.67 were found, respectively, for SCORE, CUORE, FRS, QRISK2, QRISK3, RRS, and ASSIGN. AUC values for the ML algorithms were: 0.70 for SVM, 0.73 for RF, and 0.64 for KNN. Feature analysis showed that C-reactive protein (CRP) has the highest importance, while SBP and hypertension treatment have lower importance.

Conclusions: All of the evaluated CV risk algorithms exhibit a poor discriminative ability, except for RRS and SCORE, which showed a fair performance. For the first time, we demonstrated that AS patients do not show the traditional ones used by CV scores and that the most important variable is CRP. The present study contributes to a deeper understanding of CV risk in AS, allowing the development of innovative CV risk patient-specific models.

Keywords: Ankylosing spondylitis; C-reactive protein; Cardiovascular risk; Machine learning.

Fig. 1

ROC curves of traditional cardiovascular risk algorithms. c-statistics scores: 0.71 (95% CI 0.52–0.87), 0.61 (95% CI 0.41–0.81), 0.66 (95% CI 0.51–0.81), 0.68 (95% CI 0.50–0.86), 0.66 (95% CI 0.48–0.84), 0.72 (95% CI 0.55–0.89), 0.67 (95% CI 0.48–0.86), 0.71 (95% CI 0.52–0.87), 0.63 (95% CI 0.44–0.83), 0.66 (95% CI 0.51–0.81), 0.68 (95% CI 0.49–0.86), 0.66 (95% CI 0.48–0.83), 0.72 (95% CI 0.55–0.89) and 0.65 (95% CI 0.46–0.85) for SCORE (a), CUORE (b), FRS (c), QRISK2 (d), QRISK3 (e), RRS (f), ASSIGN (g), SCORE*1.5 (h), CUORE*1.5 (i), FRS*1.5 (l), QRISK2-RA (m), QRISK3-RA (n), RRS*1.5 (o), and ASSIGN-RA (p)

Fig. 2

Calibration plots comparing observed vs. predicted risk for SCORE (a), CUORE (b), FRS (c), QRISK2 (d), QRISK3 (e), RRS (f), ASSIGN (g), SCORE*1.5 (h), CUORE*1.5 (i), FRS*1.5 (l), QRISK2-RA (m), QRISK3-RA (n), RRS*1.5 (o), and ASSIGN-RA (p)

Fig. 3

ROC curves of machine learning-based cardiovascular risk algorithms. c-Statistics scores: 0.70 (95% CI 0.55–0.85) for SVM (a), 0.73 (95% CI 0.61–0.85) for RF (b), and 0.64 (95% CI 0.50–0.77) for KNN (c). Calibration plots comparing observed vs. predicted risk for KNN (d), RF (e), and SVM (f). g Random forest’s importance