Constructing a clinical prediction model for patients with coronary heart disease: a study based on carotid ultrasound and clinical related factors.

Background and objective: The incidence of coronary heart disease(CHD) is high and its onset is insidious. CHD often occurs due to the formation of atherosclerotic plaques in the coronary arteries, which leads to narrowing or even occlusion of the lumen, and is often accompanied by the formation of atherosclerotic plaques in the carotid arteries. This study aims to construct a prediction model for patients with CHD by evaluating the carotid ultrasound(CDU) and integrating clinical related factors. Methods: A total of 649 patients were enrolled and divided into the CHD and the non-CHD groups according to coronary angiography (CAG) results. CDU assessment was performed 24 h before CAG, and basic patient information and clinical related factors were recorded. Patients were randomly divided into training and validation sets in a 7:3 ratio. In the training set, univariate and multivariate logistic regression analyses were used to screen independent risk factors for patients with CHD, and a risk prediction model was constructed based on these factors. The model was validated using the receiver operating characteristic and calibration curves, decision curve analysis, and clinical impact curve. Results: Multivariate logistic regression analysis of the training set exhibited that male chest pain, diabetes, triglyceride, Apolipoprotein A, plaque score, and carotid stenosis were independent predictors of CHD (p < 0.05). The nomogram constructed using these indicators exhibited high predictive performance, calibration, and clinical value in training and validation sets. Conclusion: The CHD prediction model constructed by CDU combined with clinical related factors is simple, practical, and has stable prediction performance, which can stratify suspicious patients and further guide clinical diagnosis and treatment.