A deep learning model that uses a single chest X-ray to predict the 10-year risk of death from heart attack or stroke, resulting from atherosclerotic cardiovascular disease, has been developed by researchers. The results of the study were presented today (November 29) at the annual meeting of the Radiological Society of North America (RSNA).
Deep learning is an advanced type of artificial intelligence (AI) that can be trained to search X-ray images to find patterns associated with disease.
“Our deep learning model offers a potential solution for opportunistic population-based cardiovascular disease risk screening using existing chest x-ray images,” said study lead author Jakob Weiss. , MD, radiologist affiliated with the Cardiovascular Imaging Research Center in Massachusetts. General Hospital and the AI in Medicine program at Brigham and Women’s Hospital in Boston. “This type of screening could be used to identify people who would benefit from statins but who are currently untreated.”
Current guidelines recommend estimating the 10-year risk of major adverse cardiovascular events to determine who should receive a statin for primary prevention.
“Based on a single existing chest x-ray image, our deep learning model predicts future major adverse cardiovascular events with similar performance and additional value to the established clinical standard.” — Jakob Weiss, MD
This risk is calculated using the Atherosclerotic Cardiovascular Disease Risk Score (ASCVD), a statistical model that takes into account a multitude of variables, including age, sex, race, systolic blood pressure, treatment high blood pressure, smoking, type 2 diabetes and blood tests. Statins are recommended for patients with a 10-year risk of 7.5% or greater.
“The variables needed to calculate the risk of ASCVD are often not available, making population-based screening approaches desirable,” Dr. Weiss said. “Because chest X-rays are commonly available, our approach can help identify those at high risk.”
Dr. Weiss and a team of researchers trained a deep learning model using a single chest x-ray (CXR) input. They developed the model, known as CXR-CVD risk, to predict the risk of death from cardiovascular disease using 147,497 chest x-rays from 40,643 participants in the prostate cancer screening trial, Lung, Colorectal and Ovarian Cancer, a multicenter randomized controlled trial designed and sponsored by the National Cancer Institute.
“We’ve known for a long time that X-rays capture information beyond traditional diagnostic results, but we didn’t use that data because we didn’t have robust and reliable methods,” Dr. Weiss said. “Advances in AI make this possible now.”
The researchers tested the model using a second independent cohort of 11,430 outpatients (mean age 60.1 years; 42.9% male) who underwent routine outpatient chest x-rays at Mass General Brigham and were potentially eligible for statin therapy.
Of 11,430 patients, 1096, or 9.6%, experienced a major adverse cardiac event during the median follow-up of 10.3 years. There was a significant association between the risk predicted by the CXR-CVD deep learning risk model and the observed major cardiac events.
The researchers also compared the model’s prognostic value to the established clinical standard for deciding statin eligibility. This could only be calculated for 2401 patients (21%) due to missing data (eg blood pressure, cholesterol) in the electronic record. For this subset of patients, the CXR-CVD risk model performed similarly to the established clinical standard and even provided additional value.
“The beauty of this approach is that all you need is an X-ray, which is acquired millions of times a day around the world,” Dr. Weiss said. “Based on a single existing chest x-ray image, our deep learning model predicts future major adverse cardiovascular events with similar performance and additional value to the established clinical standard.”
Dr Weiss said further research, including a randomized controlled trial, is needed to validate the deep learning model, which could ultimately serve as a decision support tool for treating physicians.
“What we’ve shown is that a chest X-ray is more than a chest X-ray,” Dr. Weiss said. “With an approach like this, we get a quantitative measure, which allows us to provide both diagnostic and prognostic information that helps the clinician and the patient.”
Co-authors are Vineet Raghu, Ph.D., Kaavya Paruchuri, MD, Pradeep Natarajan, MD, MMSC, Hugo Aerts, Ph.D., and Michael T. Lu, MD, MPH The researchers were supported in part by funding from the National Academy of Medicine and the American Heart Association.
Meeting: 108th Scientific Assembly and Annual Meeting of the Radiological Society of North America