Related Experiment Video
Updated: Aug 5, 2026
Intravascular Ultrasound Image-Based Finite Element Modeling Approach for Quantifying In Vivo Mechanical Properties of Human Coronary Artery
Published on: December 6, 2024
Digital phenotyping of aortic stenosis-related remodeling reveals complementary structural, electrical, and
Insights
This study introduces three AI-driven digital biomarkers to comprehensively assess aortic stenosis (AS) remodeling. These biomarkers predict AS progression and the need for aortic valve replacement, offering a new framework for cardiovascular disease phenotyping.
Area of Science:
- Cardiology
- Artificial Intelligence
- Biomarkers
Background:
- Aortic stenosis (AS) is a complex aging disease involving valvular calcification and multi-system remodeling.
- Current diagnostic measures incompletely capture the full spectrum of AS-related structural, electrical, and hemodynamic changes.
Purpose of the Study:
- To develop and validate AI-derived digital biomarkers for a multidimensional assessment of AS.
- To investigate the distinct genetic and transcriptomic architectures underlying these digital phenotypes.
Main Methods:
- Utilized AI to derive three digital biomarkers: cine-CMR DASSi (structural), AI-ECG (electrical), and phase-contrast CMR peak aortic velocity (hemodynamic).
- Analyzed data from 68,714 UK Biobank participants for associations with prevalent AS and prediction of aortic valve replacement.
- Performed genetic and transcriptomic analyses on the digital phenotypes.
Main Results:
- All three AI biomarkers were independently associated with prevalent AS and prospectively predicted aortic valve replacement.
- Peak aortic velocity showed strong alignment with established AS genetics.
- DASSi and AI-ECG identified a shared myocardial remodeling axis, largely independent of clinical AS susceptibility genes.
Conclusions:
- Aortic stenosis is a multidimensional remodeling syndrome, not fully characterized by single measures.
- The novel digital phenotyping framework provides complementary, biologically informative axes for dissecting complex cardiovascular diseases like AS.
- AI-derived biomarkers offer a promising approach for comprehensive disease assessment and risk stratification.
Abstract:
Aortic stenosis (AS) is a heterogeneous disease of aging characterized by valvular calcification and distinct structural, electrical, and hemodynamic remodeling that are incompletely captured by any single diagnostic measure. Here we show that three AI-derived digital biomarkers resolve AS-related remodeling into complementary structural (cine-CMR Digital AS Severity Index, DASSi), electrical (AI-ECG), and hemodynamic (phase-contrast CMR peak aortic velocity) axes. Among 68,714 UK Biobank participants, all three biomarkers were independently associated with prevalent AS and prospectively predicted aortic valve replacement. Genetic and transcriptomic analyses of the digital phenotypes revealed partially distinct, heritable architectures: peak aortic velocity aligned closely with clinical AS genetics, whereas DASSi and AI-ECG defined a shared myocardial-remodeling axis largely independent of clinical AS susceptibility. These findings support AS as a multidimensional remodeling syndrome and establish a novel digital phenotyping framework for dissecting complex cardiovascular disease into complementary, biologically informative axes.
