Singapore – 9 May, 2024 – New research led by the University of Dundee [1], United Kingdom, demonstrated the feasibility of AI to automatically identify and classify patients with heart failure from archived echocardiographic images with Us2.ai software.
Heart failure (HF) is a highly prevalent yet under-diagnosed condition with high mortality and morbidity [2]. Echocardiography is a foundational investigation to diagnose HF and differentiate the types of HF i.e. HF with reduced (HFrEF), mildly reduced (HFmrEF) and preserved (HFpEF) ejection fraction [3, 4].
Electronic health records (EHRs) are an increasingly high-quality data source that can be used for the creation of pragmatic cohort studies [5], disease surveillance, case selection for clinical trials (RCTs) [6], and quality improvement initiatives [7]. The quality and quantity of EHR data are expanding and increasingly include EHR-linked biobanks [8-9] and EHR-linked imaging data [10].
This study aimed to identify and classify patients with HF from routinely stored EHR data, linked to Scottish Health Research Register (SHARE) [11] bioresource and echocardiographic data collected from the Tayside and Fife region of Scotland using a deep learning-based approach.
AI-automated analyses of DICOM echocardiographic images using Us2.ai software was combined with analysis of biomarkers from routinely stored plasma samples using Roche Modular E170 (Roche Diagnostics, Mannheim, Germany). AI image analysis accurately quantified both systolic and diastolic left ventricular function, as well as structural characteristics of left- and right- atria and ventricles. The research team demonstrated the feasibility of identification and differentiation of HF particularly distinguishing between types of HF at large scale in a streamlined time and cost-efficient manner.
”Our approach has potential clinical implications, especially in the precision required in HFpEF clinical trials and the broader context of heart failure diagnosis and surveillance. The automation our DL algorithms not only makes the diagnosis process more efficient than traditional methods but also paves the way for identifying heart failure cohorts more pragmatically. When combined with biobank data, such as that from the SHARE project, our methods hold the promise of accelerating biomarker validation and fostering innovations in drug discovery for heart failure treatment” said Dr Chim Lang, from the University of Dundee’s School of Medicine.
These data are further supported by prior evidence that AI-automated echocardiographic image analyses with Us2.ai is interchangeable with human experts [12], produces measurements comparable to gold-standard invasive hemodynamic filling pressures [13], is generalizable in both real-world and research cohorts worldwide [14], and has potential for mobile screening applications [15].
About Us2.ai
Us2.ai uses machine learning to automate the fight against heart disease. The company’s software tools improve clinical decision-making and cardiovascular research for clinical trials using echocardiography, the safest and most common cardiac imaging modality. Us2.ai connects institutions and imaging labs around the world on a platform of ready-to-use automation tools for view classification, segmentation and reporting of findings according to International Guidelines and recommendations.
References:
Media ContactCompany Name: Us2.aiContact Person: Christine GouillardEmail: Send EmailCountry: SingaporeWebsite: https://us2.ai/