Novel Approach to Acute Heart Failure Risk Stratification Using AI-Derived Congestion Index from Chest Radiographs and Clinical Parameter.

This study developed a machine learning-based Clinical Decision Support System (CDSS) by integrating an AI-derived Congestion Index (CIx) from chest X-rays with clinical and laboratory data in patients with acute decompensated heart failure (ADHF). Among 9,286 patients, the model incorporating imaging data showed the highest predictive accuracy (AUROC 0.750). CIx demonstrated prognostic performance comparable to NT-proBNP and showed a significant increase in event rates across quartiles. The final CDSS-HF index was also strongly associated with outcomes. These findings support the use of AI-based imaging biomarkers for enhanced risk stratification in ADHF.