Cardiovascular Disease and Population Health
Christina Nelson, Communication Specialist
According to the Centers for Disease Control and Prevention (CDC), heart disease is the leading cause of death in the United States for both men and women. Heart disease is strongly influenced by patients’ lifestyle choices, including diet and exercise. In honor of American Heart Month, we are exploring how applying artificial intelligence (AI) in cardiovascular imaging can be part of a model to improve population health.
Improving population health through 3D visuals
AI could help patients manage their heart health by creating advanced visualizations of digitized diagnostic images. At CuraCloud, we are researching how deep learning algorithms could be applied to coronary computed tomography angiogram (CCTA) scans to automatically evaluate the coronary arteries and generate a 3D reconstruction of the coronary vascular tree. Engaging patients with interactive visuals that illustrate their heart health may support efforts to encourage behavioral changes, leading to the prevention of more severe diseases.
At the 2019 Radiological Society of North America Annual Meeting, we demonstrated a prototype to gain feedback from clinicians who came to our booth. Some of the most enthusiastic responses we heard came from attendees who suffer from cardiovascular illnesses themselves. This raised our hypothesis that using personalized 3D visualizations with patients to see their plaque and stenosis findings could help improve the management of heart health. Further research needs to be done to demonstrate the effectiveness of collaborative medical decision tools.
Predicting cardiovascular risk from medical images
Many organizations including large employers, commercial health insurers, and government health programs bear financial risk for the care of the populations entrusted to them. Value-based reimbursement models encourage healthcare providers to go beyond providing fee-based services. Predicting which patients are most likely to experience cardiac events can focus prevention efforts. In theory, value-based arrangements help organizations to do a better job of population health management.
As an example, the American Heart Association’s Institute for Precision Cardiovascular Medicine recognized Chun Yuan, Ph.D last month for his research to detect and predict blocked arteries and cardiovascular risk using knee MRI images. Dr. Yuan applied deep learning to MRI knee scans to delineate vessel wall contours, quantify vascular features, and identify arteries with disease, resulting in reduced turnaround times. Shortening the time it takes to review images can help streamline patient care and ensure that patients can receive treatment in a timely manner. Other studies support the primary use of machine learning in cardiovascular care.
Replacing invasive procedures
Another area with strong value-based care opportunities is the replacement of invasive procedures with non-invasive procedures. Research has demonstrated that machine learning can be applied to CCTA images. This procedure is relevant to patients with stable angina who might otherwise be referred to the coronary catheterization lab where a wire would be inserted to physically measure their Fraction Flow Reserve (FFR), a measure of the pressure differences of blood flowing through the coronary arteries to identify the functional significance of stenoses. The non-invasive measurement is performed by algorithms processing the digital file created during the CTA, in which the patient is reclined on the table for a few minutes in the CT scanner.
The potential of FFR-CT to improve specificity in identifying significant coronary artery disease in patients with acute chest pain was demonstrated in a recent study by researchers at the Medical University of South Carolina. This study demonstrated the application of a relatively new non-invasive procedure in the Emergency Department to perform a “triple rule-out” for evaluating the coronary arteries, aorta, pulmonary arteries, and adjacent structures.
Personalized patient care
The application of machine learning to medical image analysis offers new ways to predict risk, diagnose, and treat patients in a highly personalized way, while also improving care for the populations from which they come.