Diabetes, which impacts 37 million Americans, can lead to complications if not diagnosed in a timely fashion. Currently, screening is recommended for adults aged 35 to 70 who are overweight or obese. That may not always catch all cases, however, and a new study has found another way to clue doctors in on a patient’s risk.
Multi-institutional research recently published in Nature Communications shows how AI can be used to glean diabetes risk from x-rays. Using deep learning, the researchers were able to develop a model that successfully detected an elevated diabetes risk within x-ray images collected during routine medical care. The researchers say this could be helpful for patients who don’t meet the screening guidelines for higher diabetes risk, such as those from races and ethnicities for which BMI isn’t a reliable indicator. It may also help prevent complications by catching the disease early.
Dr. Judy Wawira Gichoya, lead researcher and assistant professor of radiology and imaging sciences at Emory University, explains, “Chest x-rays provide an ‘opportunistic’ alternative to universal diabetes testing. This is an exciting potential application of AI to pull out data from tests used for other reasons and positively impact patient care.”
To conduct their research, the team used more than 270,000 chest x-ray images from 160,000 patients to train the AI model on features that best predicted a later diagnosis of diabetes. Deep learning determined these high-risk features, including the location of fatty tissue, which reflects current research showing that visceral fat in the upper body and abdomen is linked with type 2 diabetes.
When later testing their model on nearly 10,000 patients, the team found that it was better at predicting risk than a simple model with only clinical data. The model was also found to be capable of predicting diabetes risk up to three years before a diagnosis.
Dr. Francisco Pasquel, associate professor at Emory’s Division of Endocrinology, Metabolism and Lipids, says, “Diabetes is a chronic disease where body fat distribution matters. The longer the duration of the disease and the worse the glycemic control, the higher the risk of complications. The opportunistic approach of using chest x-rays to identify those at the highest risk of diabetes, even before a spike or drop in blood sugar levels occurs, is a promising method that may help improve outcomes through early preventive measures or treatment.”
Going forward, the team hopes to further validate their model and get it in use for diabetes screening, and then to see if chest x-rays can be used to diagnose other conditions.Whizzco