Deep Learning Model of Diastolic Dysfunction Risk Stratifies the Progression of Early-Stage Aortic Stenosis.

Area of Science:

• Cardiology
• Artificial Intelligence in Medicine
• Medical Imaging Analysis

Background:

• Aortic stenosis (AS) development and progression from aortic valve (AV) sclerosis are highly variable and difficult to predict.
• Early identification of individuals at risk for AS progression is crucial for timely intervention.

Purpose of the Study:

• To investigate if a deep learning (DL) model, previously validated for assessing diastolic dysfunction (DD) via echocardiography, can identify latent risks for AS development and progression.
• To evaluate the predictive capability of DL-assessed DD for AS onset and adverse cardiovascular outcomes.

Main Methods:

• 898 participants with AV sclerosis from the ARIC cohort were analyzed.
• DL-predicted probability of DD was associated with new AS diagnosis and a composite of mortality or AV interventions.
• Validation was performed in two additional cohorts using cardiac MRI and PET/CT imaging for valvular inflammation and calcification.

Main Results:

• In the ARIC cohort, higher DL-predicted DD probability correlated with increased AS development (aHR: 3.482) and mortality/AV interventions (aHR: 7.033).
• A multivariable Cox model incorporating DL-DD probability effectively predicted AS progression in a CMR cohort (C-index: 0.798).
• Model predictions positively correlated with valvular 18F-NaF uptake in the PET/CT cohort (r=0.62).

Conclusions:

• Echocardiography-based deep learning assessment of diastolic dysfunction can effectively stratify latent risk in early-stage aortic stenosis.
• This AI-driven approach offers a promising tool for identifying patients at high risk of AS progression.