This study aims to compare the efficacy of single-indicator models versus comprehensive models in predicting cardiac deterioration events in patients with acute heart failure (AHF), providing a more precise predictive tool for clinical practice. This retrospective cohort study included 484 patients with AHF treated at our hospital between June 2018 and January 2023. Patients were categorized into a deterioration group and a non-deterioration group based on the occurrence of cardiac deterioration events within 1 year, defined as cardiogenic shock, cardiac arrest, or the need for mechanical circulatory support. We collected clinical data, laboratory markers, and imaging indicators for analysis. Both single-indicator models and comprehensive models (clinical data + indicators) were constructed and evaluated using the area under the receiver operating characteristic (ROC) curve (AUC) to assess their predictive performance. Among the 484 AHF patients, 121 were in the deterioration group and 363 were in the non-deterioration group. Among the single indicators, WBC had the highest AUC of 0.683. The indicator model (WBC, NOMO, Cr, BUN, Troponin, NT-proBNP, D-Dimer, LVEF, and RVFAC) achieved an AUC of 0.886 in the training set and 0.876 in the validation set. The comprehensive model (age, time from onset to admission, heart failure type, WBC, NOMO, Cr, BUN, troponin, NT-proBNP, LA, D-dimer, fibrinogen, and RVFAC) had an AUC of 0.940 in the training set and 0.925 in the validation set. In the training set, the comprehensive model had a significantly higher AUC than the indicator model (P < .05), while no significant difference was observed between the 2 in the validation set (P > .05). Furthermore, decision curve analysis (DCA) and calibration curve analysis indicated that the comprehensive model provided greater clinical benefits and better predictive accuracy in clinical applications. The comprehensive model demonstrates superior predictive capability for cardiac deterioration events in AHF patients, significantly outperforming both single-indicator and indicator models. This suggests that a comprehensive assessment can more accurately identify high-risk patients, offering a more reliable basis for clinical decision-making.
Read full abstract