AI Revolutionizes Cardiac Diagnosis with Ultrasound

Multiple research initiatives and pilot programs are demonstrating the potential of artificial intelligence (AI) to enhance the detection and diagnosis of various cardiac conditions using ultrasound technology. Studies have explored AI models for identifying significant valvular disease, ventricular dysfunction, and advanced heart failure, with one platform also being implemented in a clinical pilot to improve point-of-care diagnostics. These advancements aim to streamline diagnostic processes, increase accessibility, and potentially reduce hospital stays by enabling earlier detection of heart conditions.

AI models are showing significant potential in identifying valvular disease, ventricular dysfunction, and advanced heart failure, aiming to streamline diagnostics and increase accessibility.

Clinical Studies Advance AI in Cardiac Imaging

Recent research has highlighted the capabilities of AI in interpreting cardiac ultrasound data.

AISAP's Single-View Ultrasound Model

A clinical study published in Frontiers in Digital Health provided evidence for AISAP's deep learning model in accurately detecting significant valvular disease and ventricular dysfunction. The study reported that this detection is achievable using a single, focused ultrasound view, even when images are acquired by non-cardiologists using handheld devices.

The model was trained on over 120,000 echocardiographic studies and subsequently validated against a prospective patient cohort. Results indicated strong diagnostic performance, with the AI achieving an Area Under the Curve (AUC) of up to 0.97 for detecting reduced ejection fraction and 0.95 for right ventricular dysfunction during real-world prospective testing.

Dr. Lior Fisher, lead author from the Leviev Cardiovascular Institute at Sheba Medical Center, stated that these findings may reduce technical barriers to cardiac imaging.

Adiel Am-Shalom, CEO and Co-Founder of AISAP, noted that this validation supports the POCAD™ platform's potential for timely, bedside decision-making.

AISAP's FDA-cleared POCAD™ platform is currently used clinically, with the single-view research informing future innovations.

AI for Advanced Heart Failure Detection

Separately, a study published in npj Digital Medicine by investigators from Weill Cornell Medicine and NewYork-Presbyterian suggests that AI techniques applied to cardiac ultrasound data could improve the identification of patients with advanced heart failure. Currently, advanced heart failure is typically diagnosed using cardiopulmonary exercise testing (CPET), which requires specialized equipment and trained staff, limiting its availability.

The novel AI-powered method developed for this study aims to predict peak oxygen consumption (peak VO2), a key CPET measure, by utilizing ultrasound images of the heart combined with patient electronic health records.

Dr. Fei Wang, associate dean for AI and data science at Weill Cornell Medicine and a senior author, indicated this approach offers a pathway for assessing advanced heart failure patients using existing medical data.

The model, which processes moving ultrasound images, waveform imagery, and electronic health record data, was trained on deidentified data from 1,000 heart failure patients and tested on 127 new patients. The results showed an approximate 85% accuracy in distinguishing high-risk patients. The research team has initiated plans for clinical studies to pursue U.S. Food and Drug Administration approval. This work is the first publication from the Cardiovascular AI Initiative, a collaboration between Cornell, Columbia, and NewYork-Presbyterian.

Implementation of AI-Assisted Ultrasound in Clinical Practice

Northern Health in Australia has launched a research study to integrate AISAP, an AI-assisted Point-of-Care Ultrasound (POCUS) platform, into its health service. The initiative aims to improve the diagnosis and management of cardiac conditions.

Following training sessions in February for General Medicine and Respiratory clinicians, staff received hands-on training in AI-assisted cardiac POCUS. This included supervised bedside scanning and real-time AI-assisted interpretation.

Dr. Vinita Rane, Head of Medicine Unit 5, commented that this brings cardiac ultrasound closer to the bedside.

Dr. Peter Cheng, Emergency Department Physician, described the study as an "Australian first" multidisciplinary effort.

The AISAP platform is a secure, cloud-based system that uses AI to support clinicians in acquiring and interpreting cardiac ultrasound images at the bedside, offering real-time guidance and automated measurements.

The implementation study will evaluate the platform's impact on diagnostic accuracy, timeliness of decision-making, and patient flow, with goals to improve clinical outcomes and system efficiency.

Dr. Katharine See, Chief Health Outcomes Officer and principal investigator, stated that the study focuses on redesigning care delivery.

Funded by a grant from the Hospitals Contribution Fund of Australia (HCF) Research Foundation, the evaluation will assess the platform's effect on:

• Diagnosis acceleration
• Correlation with formal echocardiography
• Length of stay
• Clinician experience