AI Tool Predicts Heart Muscle Disease Years Before Conventional Diagnosis

A revolutionary AI tool, crafted by researchers at the CarDS — Cardiovascular Data Science — Lab at the school of Medicine, is demonstrating remarkable potential in detecting cardiomyopathies well in advance of traditional medical diagnoses. The AI model, detailed in a study published in The Lancet Digital Health, analyzes point-of-care ultrasound data to pinpoint subtle indicators of heart muscle disease.Founded in 2020, the CarDS Lab has dedicated the past five years to pioneering AI-based applications aimed at enhancing the medical diagnosis of heart muscle diseases.

This innovative tool is specifically designed for the early detection of conditions such as hypertrophic cardiomyopathy and amyloid cardiomyopathy,both of which compromise the heart’s ability to effectively pump blood. early detection is paramount, enabling timely intervention and potentially improving patient outcomes. The AI’s capacity to identify these conditions years before conventional diagnosis represents a significant leap forward in cardiac care.

The Genesis of the AI Tool

The development of this AI tool was a meticulous undertaking, commencing nearly two years ago with the grant proposal. The research team dedicated approximately 1 1/2 years to compiling data from various hospitals to effectively train the AI model. The team’s primary focus was on creating AI models capable of utilizing data from readily available diagnostic tools.

Dr. Evangelos Oikonomou, a key figure in the project, emphasized the importance of accessibility in their work.We want to design AI tools that we can use with tests that are easy to perform and that can be easily available in the community, Dr. Oikonomou said. We don’t necessarily want to build AI tools for technologies that might be very hard to come across or find and are only restricted to very specific, high resource settings.

Leveraging Point-of-Care ultrasound

The team strategically chose to focus on point-of-care ultrasound, a portable and readily accessible diagnostic tool. This method involves using an ultrasound rod plugged into a smartphone to capture detailed images of the heart.While point-of-care ultrasound is widely available, it is indeed frequently enough used for basic assessments, and subtle abnormalities can be easily overlooked.

According to Dr. Oikonomou, many abnormal heart conditions go undetected even with the use of point-of-care ultrasound. Recognizing the subtle patterns indicative of cardiomyopathies requires extensive training and expertise, making it challenging for medical professionals to thoroughly examine every ultrasound image.

We see that the model is actually able to pick up on patterns that are visible to the human eye,but probably not detectable by an untrained operator. But it also goes a bit beyond that. It even seems to detect those conditions way before clinicians actually suspect what’s going on.

Dr. Evangelos Oikonomou

Training and testing the AI Model

To train the AI model, the researchers utilized real-world data collected from over 30,000 patients within the Yale Health System over a span of 10 years. Once a functional training model was established, the team began testing it with videos from patients who had not been previously seen by the AI. These patients were screened using point-of-care ultrasound devices in the emergency rooms of the Yale Health System, the Yale New Haven Health System, and the mount Sinai Health System.

The AI model demonstrated the ability to detect the diseases an average of two years before they were diagnosed in real-world practice. Furthermore, the model identified patients who were never tested for these conditions but were flagged as high risk, and these patients later experienced worse health outcomes.

We found that the algorithm could pick up the disease at an average of two years before the disease was eventually diagnosed in real-world practice. We also found that there were a lot of patients that were never actually tested for any of those conditions but were flagged as high risk by our models,and these patients went on to have worse outcomes.

Dr. Evangelos oikonomou

Impressive Accuracy and Multi-Site Validation

The AI models accurately detected the two types of cardiomyopathy with impressive AUROC (Area Under the receiver Operating Characteristic curve) scores of 0.95 and 0.98. An AUROC of 1 is considered a “perfect model,” highlighting the AI’s exceptional performance. The testing at Mount Sinai hospital System in New York yielded positive results, further validating the AI’s capabilities.

The CarDS lab has made this AI submission accessible for research purposes, promoting further exploration and development in the field.

The Future of AI in Cardiology

Dr. Oikonomou emphasized the necessity of integrating AI into medical practices to enhance accessibility and scalability. We need to make use of AI. We need to leverage AI to make those technologies more accessible, more scalable, he stated.

The CarDS lab’s choice of point-of-care ultrasound is strategic, as it is indeed indeed easily applicable for community-based screenings of abnormal heart disorders, costs less than $2,000, and can be readily connected to a smartphone.

gregory Holste, a doctoral student in the CarDS Lab, highlighted the meaning of the collaboration with external partners at Mount Sinai. This multi-site validation confirms that the model will work reliably when exposed to new settings and patient distributions beyond what was seen during model development, Holste said.

Looking ahead, Dr. Oikonomou shared plans for a clinical trial were some providers will have access to the AI tools, while others will not. The results from this trial will provide valuable insights into the clinical value of using AI technologies to detect abnormal heart muscle disorders.

Conclusion

The development of this AI tool represents a significant leap forward in the early detection and management of cardiomyopathies. By leveraging accessible technology and advanced algorithms, the CarDS Lab at the School of Medicine is paving the way for more proactive and effective cardiac care. The potential impact on patient outcomes is considerable, offering hope for earlier intervention and improved quality of life for individuals at risk of heart muscle disease.