Artificial intelligence based characterization of multi-organ ultrasound congestion across the heart failure spectrum
L Del Punta, G Aru, A Sirbu, N De Biase, S Taddei, G Prencipe, S Masi, N R PuglieseAbstract
Background
clinical and biohumoral assessment provide an incomplete characterization of congestion in heart failure (HF)
Purpose
AI-driven analysis of the relationships between ultrasound (US) congestion and clinical and echocardiographic parameters.
Methods
1,588 subjects (651 Stage A–B, 376 HF with reduced left ventricular ejection fraction [HFrEF, <50%], 561 HF with preserved ejection fraction [HFpEF, ≥50%]) underwent clinical evaluation, laboratory testing, echocardiography, and US assessment of congestion, including inferior vena cava (IVC), lung ultrasound (LUS), renal (RVF), portal (PVF) and hepatic venous flow (HVF).
Results
856 patients had no US signs of congestion, 458 had one US sign, and 274 had ≥2 US signs (multi-organ congestion). AI-based predictive models were developed for each site of congestion, and for multi-organ congestion using a 3-item model (IVC, LUS, RVF). Congestion-related features clustered into four domains: medical history, biohumoral variables, left heart morphology and function, right heart and pulmonary circulation. The 3-item model identified mitral annular systolic velocity, systolic and diastolic pulmonary artery pressure, triglycerides, left atrial volume index, diabetes, treatment with furosemide or angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers as key predictors of multi-organ congestion (area under the curve = 0.79).
Conclusion
AI-assisted integration of multi-organ ultrasound characterizes congestion as a multidimensional phenotype.For image description, please refer to the figure legend and surrounding text.Determinants of multi-organ congestionFor image description, please refer to the figure legend and surrounding text.