Researchers have developed a deep learning model that uses a single chest x-ray to predict the 10-year risk of dying from heart attack or stroke resulting from atherosclerotic cardiovascular disease. The results of this 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 for patterns associated with disease.
“Our deep learning model provides a potential solution for population-based opportunistic cardiovascular disease risk screening using existing chest radiographs,” said the study’s lead author, said Jakob Weiss, M.D., Ph.D., a radiologist at the Cardiovascular Imaging Research Center in Massachusetts. General Hospital and his AI in Medicine program at Boston’s Brigham and Women’s Hospital. “This type of screening can be used to identify individuals who would benefit from statin treatment but are not currently on treatment.”
Current guidelines recommend estimating 10-year risks of major adverse cardiovascular disease events to establish who should receive statins for primary prevention.
“Based on an existing single chest radiograph, our deep learning model predicts future major adverse cardiovascular events with performance and increments similar to established clinical standards.” — Jacob Weiss, M.D.
This risk was determined using the Atherosclerotic and Cardiovascular Disease (ASCVD) Risk Score, a statistical model that considers many variables including age, gender, race, systolic blood pressure, hypertension treatment, smoking, type 2 diabetes, and blood tests. calculated as Statins are recommended for patients with a 10-year risk of 7.5% or greater.
“An approach to population-based screening is preferable because the variables needed to calculate ASCVD risk are often not available,” said Dr. Weiss. “Since chest X-rays are commonly available, our approach may help identify at-risk individuals.”
Dr. Weiss and a team of researchers trained a deep learning model using a single chest X-ray (CXR) input. They developed a model known as CXR-CVD risk and used his 147,497 chest x-rays from his 40,643 participants in prostate, lung, colorectal, and ovarian cancer screening trials to It predicted the risk of death from cardiovascular disease. A controlled trial designed and sponsored by the National Cancer Institute.
“It has long been recognized that X-rays capture information beyond traditional diagnostic findings, but we have not used this data because we lacked a robust and reliable method,” said Dr. Weiss. “With advances in AI, it is now possible.”
The investigators conducted a second independent study of 11,430 outpatients (mean age 60.1 years, 42.9% male) who had routine outpatient chest radiographs at Mass General Brigham and who were potential candidates for statin therapy. I tested the model using a cohort.
Of the 11,430 patients, 1,096, or 9.6%, had a major adverse cardiac event during a median follow-up of 10.3 years. There was a significant association between risk predicted by the CXR-CVD risk deep learning model and observed major cardiac events.
The researchers also compared the prognostic value of the model to established clinical criteria for determining statin eligibility. This could only be calculated for 2,401 patients (21%) due to missing electronic record data (blood pressure, cholesterol, etc.). For this subset of patients, the CXR-CVD risk model performed similarly to established clinical criteria and even provided incremental value.
“The beauty of this approach is that it only requires X-rays, which are acquired millions of times a day around the world,” said Dr. Weiss. “Based on an existing single chest radiograph, our deep learning model predicts future major adverse cardiovascular events with performance and increments similar to established clinical criteria.”
Dr. Weiss said additional research, including controlled randomized trials, is needed to validate deep learning models, which could ultimately serve as a decision-support tool for treating physicians. I said yes.
“What we’ve shown is that chest x-rays are more than chest x-rays,” Dr. Weiss said. “Such an approach provides a quantitative measure that can provide both useful diagnostic and prognostic information to clinicians and patients.”
Co-authors are Vineet Raghu, Ph.D., Kaavya Paruchuri, MD, Pradeep Natarajan, MD, MMSC, Hugo Aerts, Ph.D., and Michael T. Lu, MD, MPH. National Academy of Medicine and American Heart Association.
Conference: 108th Scientific and Annual Meeting of the Radiological Society of North America