Diabetes cuts a wide swath across the U.S. population and is an inviting opportunity for health tech development. Unlike many cancers and neurological diseases, the biological mechanisms involved in both Type 1 and Type 2 diabetes are relatively well-understood, which give developers specific data factors to work with. Artificial intelligence is increasingly employed in health and medical screening, diagnosis, and prediction, and even in designing prevention, treatment, and care management plans. We’ve seen some new technologies using AI solutions in new tech developed for diabetes prevention or care. Medtronic, for example, partnered with IBM’s Watson’s cognitive computing platform to help prevent, identify, and manage diabetes. Also, the FDA recently cleared IDx’s IDx-DR to predict diabetic retinopathy, which can result in blindness.

AI application developer SweetchHealth Ltd. and WellSpan Health, an integrated health care system in parts of Maryland and Pennsylvania, joined forces to help identify patients at greatest risk for diabetes. Sweetch is a digital therapeutic application that uses an AI platform designed to fight chronic diseases including diabetes, hypertension, hyperlipidemia, and obesity. The application has three stages. Sweetch starts by collecting data from sensors in smartphones and compatible wearables. Sensor data is combined and analyzed in real time in conjunction with personal information about habits and preferences, geolocation, maps, weather, and more. The result of the first two steps is what Sweetch refers to as The Feedback Loop: “personalized and contextual real-time recommendations which continuously adapt to the user’s behavior.” The program sends the recommendations to the user’s mobile device.

WellSpan plans to use Sweetch for proactive prediabetes screening and providing personalized programs for people at the highest risk levels. WellSpan will use the app first with its 15,000 employees. If AI can generate individualized care and prevention plans based on volunteered data and real-time monitoring, the logical assumption is a positive end result. While some users may have concerns about the data collected by the system, those could be assuaged by better health outcomes and lower costs for all.