Abstract

Valvular heart disease (VHD) is a significant public health challenge, with over half of cases remaining undiagnosed due to late symptom onset and the limitations of traditional stethoscope-based screening. This talk will explore two promising approaches to improve VHD screening.

First, we will discuss the development and validation of an AI-enhanced auscultation algorithm that directly predicts clinically significant VHD from electronic stethoscope recordings. This algorithm, trained on a large, diverse dataset of NHS patients, achieves a sensitivity of 98% for severe aortic stenosis and 94% for severe mitral regurgitation, significantly outperforming general practitioners in both sensitivity and specificity.

Second, we will introduce a novel handheld, multi-sensor device designed for untrained users to record high-quality heart sounds without requiring patients to undress. This device, equipped with advanced signal quality algorithms, significantly improves the likelihood of successful recordings across diverse body types and has the potential to enable low-cost, accessible screening in resource-constrained settings.

Together, these advancements offer a practical, non-invasive solution for early VHD detection, supporting timely intervention and improved patient outcomes. This talk will highlight the integration of AI, wearable technology, and health data in enhancing cardiac screening and population health.