Deep learning, in this case, means an advanced form of artificial intelligence (AI) trained to scan X-ray images to find patterns associated with diseases. And that worked.
“Our deep learning model offers a potential solution for population-based cardiovascular disease risk screening using existing chest X-rays,” says Dr. Jakob Weiss, a Boston-based radiologist and lead author of a new study. “This type of screening could be used to identify individuals who would benefit from statin medication but are not currently receiving treatment.” Statins are drugs that are used to lower cholesterol or lipids. Researchers from Jena have further developed the active ingredientthat it only needs to be sprayed twice a year.
Current US guidelines recommend assessing the 10-year risk of major cardiovascular disease to determine who should be on a statin for primary prevention. This risk is calculated using a statistical model that accounts for a variety of variables, including age, gender, blood pressure, smoking, type 2 diabetes, and blood tests. Statins are then recommended for patients with a 10-year risk of 7.5 percent or more.
Just one picture instead of many variables
However, the required variables are not always available. It would be all the nicer to be able to calculate the risk just as precisely from an easily available source.
“As chest X-rays are widely available, our approach can help identify individuals at high risk,” says Dr. White.
With a team of researchers, he trained a deep learning model using 147,497 chest X-rays from 40,643 subjects. “We have long recognized that X-ray images capture information beyond traditional diagnostic findings, but we did not use this data because we did not have robust, reliable methods,” says Dr. White. “Advances in AI now make it possible.”
test with control group
The research group tested the model on a second independent cohort of 11,430 outpatients who had undergone routine X-rays and were eligible for statin therapy. Of this control group, 1,096 (9.6%) experienced major cardiac disease during the median follow-up of 10.3 years. There was a significant correlation between the risk predicted by the X-ray model and the observed diseases.
The researchers also compared the model’s prognostic value to the established clinical standard for deciding whether to use a statin. This was calculated in 2,401 patients (21%) who had missing data (e.g. blood pressure, cholesterol) in the electronic records. In this subset of patients, the risk model performed similarly to the established clinical standard and even provided additional value.
Validation of the results
“The beauty of this approach is that you only need one x-ray, which is taken millions of times a day around the world,” said Dr. White. “Based on a single existing chest X-ray, our deep learning model predicts future major cardiovascular events with similar performance and value to the established clinical standard.”
However, according to Dr. Weiss needs more research, including a controlled, randomized trial, to validate the deep learning model, which could ultimately serve as a decision-making tool for treating physicians. “We have shown that a chest X-ray is more than just a chest X-ray,” said Dr. White. “With an approach like this, we get a quantitative measure that allows us to provide both diagnostic and prognostic information that helps the doctor and the patient.”