Abstract:
Objective To investigate risk factors for pulmonary hypertension (PH) in patients diagnosed with dilated cardiomyopathy (DCM) and develop a predictive model to facilitate early identification of patients who are at high risk, thus enabling optimized clinical management. Methods A total of 122 patients diagnosed with DCM patients were admitted to the Affiliated Hospital of Hangzhou Normal University between October 2022 and October 2024, were divided into PH and non-PH groups (61 each). Binary logistic regression was used to identify risk factors for PH in DCM and construct a predictive model. The predictive model ' s performance was evaluated using ROC curves, with calibration assessment being conducted in both the training and validation sets. Results Statistically significant differences were identified (P < 0.05) between the two groups with regard to QRS wave duration on electrocardiogram (ECG), the sum of R-wave amplitude in lead V1 and S-wave amplitude in lead V5 (RV1+SV5) on ECG, N-terminal pro-brain natriuretic peptide (NT-proBNP), left anterior descending coronary artery (LAD), and pulmonary artery systolic pressure (PASP). Binary logistic regression analysis revealed that QRS wave duration, RV1+SV5, NT-proBNP, and PASP were all influential factors for DCM complicated by PH (P < 0.05). ROC analysis demonstrated that the AUCs for predicting PH in DCM using QRS wave duration, RV1+SV5, NT-proBNP, and PASP were 0.862, 0.362, 0.840, and 0.848, respectively, the sensitivity were 70.50%, 88.50%, 68.90% and 86.90%, and the specificity were 95.10%, 9.80%, 90.20% and 73.80%, respectively. The calibration curve was utilized to evaluate the calibration of the model, and the results indicated effective discrimination of the nomogram. The prediction model ' s calculation formula was utilized to obtain prediction model data, which was then employed as test variables, with group assignment designated as the state variable. An ROC curve was undertaken for the purpose of analyzing the discrimination of each model, and the results showed an AUC of 0.985 for the prediction model, with a sensitivity of 91.80% and a specificity of 98.40%. A calibration curve for the nomogram was plotted, and there was a high degree of consistency between the calibration curves for the training and validation sets and the ideal curve, indicating a minor discrepancy between the prediction model and the actual validation. Conclusion QRS duration, NT-proBNP, and PASP are key predictors of PH in DCM. The model offers excellent predictive performance, supporting early screening and personalized treatment strategies for improved precision medicine.