Establishment and Validation of a Noninvasive Diagnostic Model for Premature Coronary Artery Disease in Women

Background and objectiveWomen with premature coronary artery disease (PCAD) present atypical clinical symptoms and electrocardiographic features, and the detection rate of positive coronary artery disease (CAD) is relatively low. This study proposes the establishment and validation of a noninvasive diagnostic prediction model for PCAD in women, with the aim of improving the accuracy of diagnosis.Methods405 female patients with suspected CAD from two medical centers were randomly divided into two cohorts, with 300 as the training cohort and 105 as the validation cohort. The baseline clinical data, laboratory examination results and imaging results were collated.The screened risk factors were incorporated into multivariate logistic regression, LASSO regression, random forest, and support vector machine recursive feature elimination (SVM-RFE) to construct a diagnostic model. This model was then subjected to external validation in a validation cohort, and the optimal noninvasive diagnostic model was displayed via a nomogram.ResultsNine risk factors screened by univariate logistic regression, including age, menopause status, postprandial 2 h blood glucose (PBG), glycosylated hemoglobin(HbA1c), recombinant cardiac troponin I (cTnI), pro-brain natriuretic peptide (pro-BNP), lactate dehydrogenase (LDH), apolipoprotein AI (ApoAI), apolipoprotein AI/B (ApoAI/ApoB), triglycerides (TCs), and abnormal ECGs, were used to construct a diagnostic model. The area under the curve (AUC) values of the SVM model in the training and validation cohorts were 0.8284316 and 0.9630194, respectively, which were greater than those of the other three noninvasive diagnostic models. These results indicated that the risk factors extracted by the SVM model more accurately reflected the pathogenic features of PCAD in women. Consequently, the SVM model was the optimal noninvasive diagnostic model, and the calibration curve indicated that the model had high diagnostic efficiency.ConclusionThe noninvasive diagnostic model for PCAD in women, which was developed via the support vector machine (SVM) method with nine risk factors, has high discrimination ability and accuracy. This model has the potential to improve diagnostic precision and serve as a valuable reference for clinical diagnosis and treatment.