Associations Between Echocardiographic Traits and AI–ECG Predictions of Heart Failure Journal: Lancet Digital Health

Elias Stenhede 1,2, Eivind Bjørkan Orstad 3,2, Torbjørn Omland 4,3, Henrik Schirmer 3,4, Arian Ranjbar 1 · 2026

Read Associations Between Echocardiographic Traits and AI–ECG Predictions of Heart Failure Journal: Lancet Digital Health by Elias Stenhede 1,2, Eivind Bjørkan Orstad 3,2, Torbjørn Omland 4,3, Henrik Schirmer 3,4, Arian Ranjbar 1 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-24576.txt.

Introduction

Elias Stenhede and four colleagues ask what heart traits an AI electrocardiography model for heart failure is picking up. They applied a validated AI-ECG model to 8,147 patients who had ECG and echocardiography within three days at one hospital. Global longitudinal strain showed the strongest link to predicted risk, followed by mitral annular plane systolic excursion and ejection fraction. The pattern held in patients with preserved ejection fraction, with diastolic measures weighing more. The team validated the findings on 36,286 external pairs.

Catalog record

CREATORElias Stenhede 1,2, Eivind Bjørkan Orstad 3,2, Torbjørn Omland 4,3, Henrik Schirmer 3,4, Arian Ranjbar 1
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE29364
LICENSEhttp://creativecommons.org/licenses/by-sa/4.0/
RIGHTSCC BY SA
SOURCEarXiv

Source: https://arxiv.org/abs/2605.24576

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Associations Between Echocardiographic Traits and AI–ECG Predictions of Heart Failure Journal: Lancet Digital Health Elias Stenhede 1,2, Eivind Bjørkan Orstad 3,2, Torbjørn Omland 4,3, Henrik Schirmer 3,4, Arian Ranjbar 1 Abstract \purpleparagraph Background Artificial intelligence-enabled electrocardiography (AI–ECG) can detect heart failure (HF), including disease not captured by left ventricular ejection fraction (LVEF), but the cardiac phenotypes underlying model predictions remain unclear. We therefore investigated whether AI–ECG-predicted HF risk aligns with established echocardiographic measures of myocardial dysfunction, remodelling, and filling pressures. \purpleparagraph Methods We retrospectively analysed ECG and echocardiography data from 8147 8147 patients who underwent both examinations within three days at Akershus University Hospital between 1 January 2023 and 1 June 2025. A previously validated AI–ECG model for HF detection was applied to all ECGs. Spearman’s rank correlation 𝝆 \boldsymbol{\rho} quantified associations between echocardiographic parameters and AI–ECG risk. Subgroup analyses were performed by sex and left ventricular ejection fraction (LVEF). External validation included 36 286 36\,286 ECG-echocardiography pairs from Columbia University Irving Medical Center. \purpleparagraph Results Global longitudinal strain (GLS) showed the strongest correlation ( 𝝆 = \boldsymbol{\rho=} 0.57 0.57 ), followed by mitral annular plane systolic excursion (MAPSE) ( 𝝆 = \boldsymbol{\rho=} − 0.49 -0.49 ) and LVEF ( 𝝆 = \boldsymbol{\rho=} − 0.45 -0.45 ). In patients with LVEF>50%, correlations remained substantial for GLS, MAPSE, and diastolic-related parameters. Volumetric left ventricular indices correlated less strongly in women, whereas diastolic indices showed stronger correlations in women than in men. \purpleparagraph Conclusion Physiological validation showed that AI–ECG HF risk predictions align primarily with measures of systolic function, particularly global longitudinal strain, while also capturing diastolic-related abnormalities in patients with preserved LVEF. This approach may improve clinical interpretability and identify opportunities for model refinement. Introduction Cardiovascular disease is the leading cause of death globally [ 26 ] , and heart failure (HF) is a common end-stage of most cardiovascular disease. HF is severely underdiagnosed due to the relatively high diagnostic complexity and low specificity of symptoms such as fatigue and dyspnea. Current diagnostic algorithms include risk factors, signs, and levels of N-terminal pro-B-type natriuretic peptide (NT-proBNP) to qualify for echocardiographic examination to rule out HF or determine its phenotype. In the last few years, AI-interpreted ECGs (AI–ECG) have emerged as a promising candidate to improve diagnostics [ 18 ] . AI–ECG models can detect a range of cardiovascular diseases, including HF [ 23 ] and its associated phenotypes [ 22 , 12 , 2 ] , but are often perceived as black-box models by clinicians. Several post-hoc explainability techniques have been proposed and used on the ECG, such as Gradient-weighted Class Activation Mapping, Integrated Gradients, and Layer-wise Relevance Propagation [ 23 , 30 , 28 , 3 , 4 ] . While these methods may highlight salient features in the ECG, their reliability is limited [ 1 ] , and they provide little mechanistic insight into cardiac function. To address this limitation, we investigate the use of established echocardiographic parameters as predictors for AI–ECG model outputs for HF detection. This approach enables the identification of HF phenotypes for which the model performs particularly well or poorly, bridging the gap between deep learning predictions and measurable cardiac physiology. Echocardiography Echocardiography is a cornerstone in the evaluation of patients with suspected HF, given its widespread availability and ability to provide a real-time assessment of cardiac structure and function [ 11 , 19 ] . Several echocardiographic parameters provide insight into systolic performance, intracardiac pressures, and volumes, as well as overall cardiac mechanics. The indices included in the present study are summarised in Table 1 and described below. Table 1 : Echocardiographic parameters considered in this study and their diagnostic roles. Abbreviation Full name Diagnostic role GLS Global longitudinal strain Left ventricular systolic function MAPSE Mitral annular plane systolic excursion Left ventricular systolic function LVEF Left ventricular ejection fraction Left ventricular systolic function TAPSE Tricuspid annular plane systolic excursion Right ventricular systolic function CI Cardiac index Global pump performance E/e’ E/e’ ratio Estimate of LV filling pressures e’ Average e’ velocity Diastolic function E/A Mitral E/A ratio Diastolic function A Mitral A-wave velocity Diastolic function LAVi Left atrial volume index Chronic elevation of LV filling pressures / diastolic burden IVS Interventricular septal thickness Left ventricular hypertrophy PWT Posterior wall thickness Left ventricular hypertrophy LVMi Left ventricular mass index Hypertrophy and remodeling assessment LVEDVi LV end-diastolic volume index LV dilation / chamber size assessment PASP Pulmonary artery systolic pressure Pulmonary hypertension 1. Diagnostic roles follow established echocardiographic interpretation guidelines [ 11 ] . Left ventricular ejection fraction (LVEF) is calculated as stroke volume divided by end diastolic volume and remains, despite its well-known limitations, the lingua franca for phenotyping HF across populations [ 13 , 15 ] . HF is conventionally classified into three groups based on LVEF: reduced (<40%), mildly reduced (41-49%), and preserved ( \geq 50%). This LVEF-based classification accurately identifies advanced systolic HF, but its ability to distinguish cases declines when LVEF is mildly reduced or within the normal range [ 24 ] . Global longitudinal strain (GLS) has, over the past two decades, emerged as a more sensitive and reproducible parameter, capable of detecting myocardial dysfunction even when LVEF is within the normal range, as seen in conditions such as diastolic HF, hypertrophic cardiomyopathy, and amyloidosis [ 25 , 19 ] . Complementary indices of longitudinal performance are the mitral and tricuspid annular plane systolic excursion (MAPSE/TAPSE), which are obtainable in almost all patients and reflect basal systolic shortening in the left and right ventricles, respectively. MAPSE correlates closely with both LVEF and GLS, but its added value becomes limited in the context of more global and advanced indices of LV systolic function [ 29 , 9 ] . For the right ventricle, TAPSE remains the main index for systolic function, with established prognostic value [ 11 , 16 ] . To complement volumetric and deformation indices, Doppler-derived cardiac output (CO, stroke volume multiplied by heart rate) and cardiac index (CI, CO divided by body surface area) provide measures of ventricular performance and can help identify reduced systolic function even with a normal LVEF [ 15 ] . Diastolic function is assessed using a complementary set of echocardiographic indices reflecting left ventricular relaxation, compliance, and filling pressures. These include transmitral inflow parameters (E and A velocities and the E/A ratio), early diastolic mitral annular velocities (e’), and the E/e’ ratio, which together characterise ventricular relaxation and estimate left ventricular filling pressure. Pulmonary artery systolic pressure (PASP) reflects right ventricular afterload and is closely associated with left atrial pressure in the absence of pulmonary disease and right-sided valvular stenosis. Indexed left atrial volume (LAVi) serves as an integrated marker of chronically elevated filling pressures and long-standing diastolic dysfunction. Collectively, these indices form the echocardiographic basis for evaluating left ventricular filling pressures and grading diastolic dysfunction, which are pivotal to the diagnosis and phenotyping of heart failure with preserved ejection fraction (HFpEF) [ 24 , 19 ] . Finally, structural remodelling of the left ventricle (LV) reflects progressive myocardial adaptation to genetic or acquired influences and the cumulative effect of altered loading conditions over time. Septal and posterior wall thickness (IVS/PWT), LV mass, and indexed end-diastolic volume (LVEDi) are among the key echocardiographic

Read the complete text