Development and validation of diagnostic and prognostic models for heart failure in unstable angina patients

OBJECTIVE: To develop diagnostic models to predict initial heart failure (HF) hospitalisation in patients with unstable angina (UA) and prognostic models to predict rehospitalisation for recurrent or new-onset HF after discharge, aiming to enhance early identification and patient care strategies. METHODS: We retrospectively analysed data from 12 857 acute coronary syndrome patients (January 2015 to March 2023). After screening, 7092 UA patients were included and randomly divided into a training cohort (4964) and a validation cohort (2128). Logistic regression and least absolute shrinkage and selection operator (LASSO) regression identified risk factors. Diagnostic models were developed and evaluated using receiver operating characteristic curves. Using the same predictors, prognostic models predicted rehospitalisation at 1, 2 and 6 months postdischarge. RESULTS: The lambda.1se criterion in LASSO regression identified three key risk factors: left ventricular diameter, N-terminal pro-B-type natriuretic peptide and ischaemic cardiomyopathy. The diagnostic model's area under the curve (AUC) was 0.938 (95% CI 0.909 to 0.955) in the training cohort and 0.931 (95% CI 0.901 to 0.956) in the validation cohort, showing consistent and reliable performance. Prognostic models for 1, 2 and 6 months postdischarge had AUC values of 0.770, 0.794 and 0.751, respectively, supporting their utility in prognostic evaluations. CONCLUSION: The UA diagnostic and prognostic HF model was developed and validated. These models can quickly identify high-risk patients, enabling prompt and tailored interventions.