Predictive potential of electrocardiographic and echocardiographic parameters for obstructive coronary involvement in patients with stable coronary artery disease
Aim . To analyze the predictive potential of electrocardiographic (ECG) and echocardiographic parameters to develop predictive models for obstructive coronary atherosclerosis in patients with stable coronary artery disease (CAD). Material and methods . This single‑center retrospective study was conducted, analyzing data from 1005 patients with stable CAD. Based on invasive coronary angiography results, two following groups were identified: group 1‑774 patients (77%) with obstructive CAD, and group 2‑231 patients (23%) with non‑obstructive CAD. Mathematical statistics and multivariate logistic regression were used to process and analyze the data, which were used to develop predictive models for obstructive CAD. Results . Multistage analysis showed that ECG parameters did not have significant intergroup differences and had no predictive value for predicting obstructive CAD. In models with continuous predictors, echocardiographic parameters did not provide an acceptable level of prediction, which was achieved only after incorporating hematological factors into the model structure. SHAP‑based categorization of continuous predictors allowed us to identify following threshold values associated with an increased risk of obstructive CAD: right ventricular basal diameter ≤2,6 cm, right atrial anteroposterior diameter (≤5,2 cm) and left atrial anteroposterior diameter (≤5 cm), fibrinogen ≥3,6 g/L, and lymphocytes ≤29%. The algorithm with categorical factors outperformed the continuous predictor model in terms of predictive quality. Conclusion . ECG parameters had no value for predicting obstructive CAD. The obstructive CAD prognostic model based on categorical predictors demonstrated higher accuracy and reliability than the model with continuous predictors.