Impact of Changes in Consensus Diagnostic Recommendations on the Echocardiographic Prevalence of Diastolic Dysfunction

• Published in: Journal of the American College of Cardiology 2017-06, Vol.69 (25), p.3119-3121
• Main Authors: Huttin, Olivier, MD, PhD, Fraser, Alan G., MD, Coiro, Stefano, MD, MSc, Bozec, Erwan, PhD, Selton-Suty, Christine, MD, Lamiral, Zohra, MA, Frikha, Zied, MD, Rossignol, Patrick, MD, PhD, Zannad, Faiez, MD, PhD, Girerd, Nicolas, MD, PhD
• Format: Article
• Language: English
• Subjects: Age; Cardiology; Cardiovascular; Data management; Decision trees; Diagnostic software; Diagnostic systems; Echocardiography; Heart diseases; Heart failure; Internal Medicine; Learning algorithms; Machine learning; Population; Ultrasonic imaging; Ventricle
• ISSN: 0735-1097; 1558-3597
• DOI: 10.1016/j.jacc.2017.04.039

Patients with no histories of heart failure and with left ventricular ejection fractions >50% were classified according to: 1) the 2016 ASE/EACVI scoring system (3); 2) the 2009 ASE/EACVI recommendations (considering diastolic function grades 2 and 3); 3) the echocardiographic algorithm of the 2007 consensus statement (2); and 4) the Appleton definition (4) (Figure 1). Similar hard outcomes do exist for DD, including the occurrence of heart failure with preserved ejection fraction (HFpEF), defined as the need for an unplanned hospitalization for heart failure with normal ejection fraction in a population of patients without heart failure at baseline. A big-data approach could be used to develop DD diagnostic software that incorporates clinical decision trees developed by machine learning, which can test multiple diagnostic approaches and select the combination of variables that best predicts clinical events.