Although onset of diabetes most commonly appears in adults younger than 65, many older adults remain at risk to either contract the disease or experience worsening symptoms, which often can present major care challenges.
One way to preempt some of the challenges is through artificial intelligence. AI can serve as a novel early detection tool when models quickly can access and analyze large data sets, provided the information is accurate.
One such recent AI model can accurately predict warning signs for diabetes using X-ray images taken during medical visits of many types, researchers say.
The X-ray images themselves stem from medical issues ranging from chest pain and breathing difficulty to pre- or post-surgery reviews. The AI model, created by researchers at Emory University, was trained on 270,000 X-rays from approximately 160,000 patients.
X-rays are not commonly used to diagnose diabetes, but the AI was able to make connections between the location of fatty tissues and disease risk, the study authors stated.
The researchers now are looking into ways to further test the model and incorporate it into electronic health record systems to alert physicians and other caregivers.
The Centers for Disease Control and Prevention estimates that approximately 300,000 older adults get a first-time diagnosis for diabetes.
The prevalence of diabetes has more than doubled over the past 35 years, and roughly 33% of seniors have the disease, according to the Endocrine Society.
Effectively treating diabetes, without incurring other side effects, is difficult in long-term care settings. Some treatments leave residents and patients with a high rate of hypoglycemia — roughly 35%, in fact, the McKnight’s Clinical Daily recently reported.
In addition, many older adults who are overtreated for diabetes do not have their medication appropriately adjusted. New treatments, such as SGLT2Is, soon could be adopted within long-term care. Another promising solution to lowering blood glucose levels is kombucha tea, researchers reported last week.
