Early Detection of Heart Failure with Autonomous AI-Based Model Using Chest Radiographs: A Multicenter Study

This study evaluated the ability of an AI-generated index from chest radiography (CXR) findings to predict heart failure (HF) development within one year. A total of 1,117 patients (mean age 67.6 years; 487 males, 630 females) were included, with 413 patients diagnosed with HF within a year and 704 without HF. AI (qXR-HF, Qure.AI) analyzed cardiac silhouette, pleural effusion, and an HF index, achieving an AUC of 0.798 (95% CI: 0.77–0.82), accuracy of 0.73, sensitivity of 0.81, and specificity of 0.68. The AUC varied by lead time to HF diagnosis (<3 months: 0.85; 4–6 months: 0.82; 7–9 months: 0.75; 10–12 months: 0.71), while accuracy (0.68–0.72) and specificity (0.68) remained stable. These findings suggest that AI-assisted opportunistic screening of CXR may help in early HF detection, supporting further investigation into its clinical utility.