# A study on predicting the risk of coronary artery disease in OSAHS patients based on a four-variable screening tool potential predictive model and its correlation with the severity of coronary atherosclerosis

## TL;DR

This study develops a four-variable model to predict coronary artery disease risk in sleep apnea patients and finds it correlates with atherosclerosis severity.

## Contribution

A novel four-variable predictive model for CAD in OSAHS patients is proposed and validated with machine learning and clinical metrics.

## Key findings

- Age, hypertension, AHI, and the 4V model are independent predictors of CAD in OSAHS patients.
- The 4V model shows strong predictive capability and correlates with coronary atherosclerosis severity via Gensini scores.
- Random forest analysis confirms AHI as the most important predictor in the model.

## Abstract

This study aims to evaluate the potential association between the four-variable screening tool (the 4 V) potential predictive model in predicting coronary artery disease (CAD) risk in patients with obstructive sleep apnea-hypopnea syndrome (OSAHS) and its correlation with the severity of coronary atherosclerosis, as measured by the Gensini scoring system.

1197 OSAHS patients with suspected CAD who were hospitalized in the First Affiliated Hospital of Xinjiang Medical University between March 2020 and February 2024 were selected. The patients were submitted to coronary angiography or Coronary Computed Tomography Angiography (CCTA) examination to confirm the diagnosis. There were 423 cases in the OSAHS plus CAD group and 774 cases in the OSAHS group. LASSO regression analysis was carried out for screening potential influencing factors. Propensity score matching (PSM) was used to balance covariables between groups, and 293 cases were included per group in a 1:1 ratio. Univariable and multivariable logistic regression analyses were employed to evaluate parameters independently associated with CAD and construct a nomogram model.Receiver operating characteristic (ROC) curve analysis, Hosmer-Lemeshow test, calibration curve and decision curve (DCA) analyses were employed to assess its predictive value in CAD. A random forest machine learning algorithm was used to evaluate the importance of each risk factor. Pearson's or Spearman's correlation coefficients were employed to assess the strengths of associations among all variables and between predictors and Gensini scores, reflected in heat maps and chord diagrams, respectively.

LASSO-logistic regression analysis revealed age (OR = 1.07, 95% CI: 1.05–1.1, P < 0.001), hypertension (OR = 1.29, 95% CI: 1.16–1.44, P < 0.001), AHI (OR = 1.02, 95% CI: 1.01–1.03, P = 0.007), and the 4 V (OR = 1.84, 95% CI: 1.21–2.79, P = 0.004) were independently associated with OSAHS plus CAD. The analysis of the ROC curve revealed that the combined utilization of the aforementioned predictors significantly enhances the potential predictive capability for patients with OSAHS developing CAD. The Hosmer-Lemeshow test, calibration curve, and DCA results indicate that potential predictive model based on the 4 V possesses significant clinical applicability in predicting OSAHS in conjunction with CAD. A comprehensive analysis utilizing the random forest machine learning algorithm demonstrated that the AHI exhibits the highest predictive value. Furthermore, the model's performance, as evaluated through out-of-bag error assessment, suggests robust efficacy. The correlation analysis results showed that the scores of the four-variable screening tool were positively correlated with the Gensini scores.

Age, hypertension, AHI, and the four-variable screening tool are independent risk factors for CAD in patients with OSAHS. The potential predictive model based on the 4 V is closely related to the prediction of CAD and its correlation with the severity of coronary atherosclerosis.