New AI Technology Shows Promise in Early Detection of Diabetes Using X-Rays and Medical Records

A groundbreaking study by researchers at the University of Texas Medical Branch published in Nature, reveals exciting progress in using artificial intelligence to predict type 2 diabetes early on. The study shows that an advanced deep learning system, trained with large sets of X-rays and medical records, effectively detected type 2 diabetes with high accuracy.

“Early detection and interventions are crucial in managing type 2 diabetes and preventing complications,” said Dr. Jorge Rodriguez Fernandez, co-author, and professor in the Department of Neurology at UTMB. “The results offer new opportunities for AI to harness disease discovery and health initiatives based on existing data.”

Diabetes, especially type 2 diabetes, has been increasing in the U.S population, leading to higher healthcare costs and health problems. Current methods for screening rely on blood glucose which can have limitations, often leading to late detection.

The research team leveraged the vast amount of X-ray images and medical records available to teach the AI system how to identify type 2 diabetes. The learning model was trained on more than 270,000 X-rays from over 160,000 patients and then tested it on 9,943 X-rays. Based on the results, the AI model correctly assigned a higher risk for type 2 diabetes 84% of the time, based solely on the frontal X-ray.

Moreover, the study highlights its far-reaching potential, particularly for underserved populations with limited healthcare access. Since chest x-rays are one of the most common radiologic exams, the AI model’s capabilities can provide screenings in patients who might lack primary care providers.

This research showcases how AI technology can revolutionize disease detection and improve public health efforts. By using existing X-ray and medical data, healthcare providers can use this AI system to find individuals at higher risk for type 2 diabetes and provide targeted care early on. Early detection can lead to better health outcomes and reduced healthcare expenses.

“We plan to continue improving and testing the AI system to make sure it works fairly and effectively in real-world medical settings,” says Dr. Ayis Pyrros, lead author of the study and Section Head of Neuroradiology at Duly Health and Care. “Our goal is to implement this system on a larger scale to benefit more people in the future.”