Researchers at the University of Maryland have developed a novel calculator that forecasts short-term risks of diabetes complications, leveraging routine clinical data and machine learning to guide personalised prevention strategies.
Researchers at the University of Maryland School of Medicine have developed a calculator that aims to forecast, month by month, which diabetes complications are most likely to affect a patient next. The Diabetes Complications Risk Calculator, or DCRC, draws on routine clinical information already present in medical records and was led by Rozalina G. McCoy, an associate professor of medicine at the school. The findings were published in Nature Communications on 25 August 2026.
The tool is designed to move beyond the broad, long-term estimates used by many existing calculators. According to the University of Maryland, it can estimate the short-term risk of nine acute and chronic complications, including cardiovascular disease, stroke, kidney disease and nerve damage, then update those estimates as new information is added during regular care visits.
To build the model, the researchers analysed health data from more than 400,000 adults newly diagnosed with diabetes across the United States. Nature Communications says the team used machine-learning methods to identify patterns linking age, blood pressure, kidney function, other illnesses, medicines and laboratory results with later complications, and then tested the models in independent patient groups, including one from Mayo Clinic.
The study suggests complications can appear early after diagnosis. The University of Maryland said about one-third of patients in the analysis had at least one complication within a year, and more than 40% had one within two years. Higher risk was associated with older age, hypertension, impaired kidney function, other existing conditions and a longer duration of diabetes.
The researchers say the calculator is meant to support, not replace, clinicians. Its value may lie in helping doctors and patients decide where to focus prevention efforts, especially when several risks are competing for attention at the same appointment. The team also noted important limits: the data came largely from insured patients, some complications were predicted more accurately than others, and the tool still needs testing in everyday practice to show whether it improves care and outcomes.
Disclaimer: This content is for informational purposes only and is not intended to be a substitute for professional medical judgment, advice, diagnosis, or treatment.
