Diagnosis of Heart Failure With Preserved Ejection Fraction: Machine Learning of Spatiotemporal Variations in Left Ventricular Deformation

Abstract

Machine learning of spatiotemporal variations of LV strain rate during rest and exercise could be used to identify patients with HFpEF and to provide an objective basis for diagnostic classification.

How to diagnose heart failure with preserved ejection fraction: the HFA–PEFF diagnostic algorithm: a consensus recommendation from the Heart Failure Association (HFA) of the European Society of Cardiology (ESC)

“…Additional diagnostic criteria for HFpEF have been published, including one scoring system, but they differ in echocardiographic cut‐off values, the role of comorbidities, the inclusion of biomarkers, the role of invasive haemodynamic assessment, and the role of exercise stress testing , 4 , 6–8 . Understanding of the pathophysiology of HFpEF has advanced, 9–13 diagnostic options have evolved, 14–17 and this novel information needs to be integrated into a new comprehensive diagnostic algorithm for suspected HFpEF.…”

“…4 Additional diagnostic criteria for HFpEF have been published, including one scoring system, 5 but they differ in echocardiographic cut-off values, the role of comorbidities, the inclusion of biomarkers, the role of invasive haemodynamic assessment, and the role of exercise stress testing. 3,4,[6][7][8] Understanding of the pathophysiology of HFpEF has advanced, [9][10][11][12][13] diagnostic options have evolved, [14][15][16][17] and this novel information needs to be integrated into a new comprehensive diagnostic algorithm for suspected HFpEF. A writing committee initiated by the HFA of the ESC has therefore produced an updated consensus recommendation-the HFA-PEFF diagnostic algorithm (Figure 1). Its key elements are (i) the concept that identification of HFpEF involves all levels of care, including general practitioners, internists, general cardiologists, HF specialists, and invasive cardiologists; (ii) a stepwise diagnostic approach from initial clinical assessment to more specialized tests will therefore be useful; (iii) the diagnosis is not always straightforward, so the integration of distinct parameters from complementary diagnostic domains into a new diagnostic score is recommended; (iv) for the subset of patients with an inconclusive score, definitive diagnosis (or exclusion) will require invasive haemodynamics and/or non-invasive or invasive exercise stress tests; and (v) underlying pathophysiological alterations [such as chronotropic…”

Unsupervised clustering for phenotypic stratification of clinical, demographic, and stress attributes of cardiac risk in patients with nonischemic exercise stress echocardiography

“…On the other hand, machine learning methods, such as unbiased cluster analysis, can operationalize phenotyping approaches: These methods have been recently proposed for meaningful categorization of patients with heart failure . The application of cluster analysis and a multiparametric approach for natural clustering of conventional variables are helpful in isolating hidden prognostic phenotypes not visualized by expert guideline‐based approaches . Cluster‐based approaches can be a useful aid in diagnostic approaches, risk stratification, and phenotypic characterization of several cardiac conditions such as diastolic dysfunction, valvular heart disease, myocardial ischemia, and coronary artery disease.…”

Machine Learning Approaches for Myocardial Motion and Deformation Analysis

“…However, these analyses consider each spatial location and temporal instant independently from the others. The statistical analysis can also consider the motion patterns over the entire cardiac cycle as high-dimensional objects, as simply demonstrated through a PCA on temporal strain traces concatenated over the heart segments (36,37). This approach reminds earlier work on Active Appearance Motion Models (38), which statistically analyzed both displacement and image intensity information over the entire cardiac cycle.…”