Breakthrough AI Model Predicts Patient-Specific Mortality Risks for Diabetes Patients
Moscow, 15 Dec 2025 (ONA) --- A team of scientists at the Institute of Biology and Biomedicine, Lobachevsky University in Russia has developed a novel artificial intelligence (AI) model designed to predict the risk of all-cause mortality in diabetic patients. The advanced model is also capable of explaining the rationale behind its predictions to medical specialists.
According to a university statement, the AI was trained on a comprehensive dataset comprising the physiological indicators of over 550 diabetic patients monitored for a period of 17 years.
Mikhail Ivanenko, the lead researcher and Director of the Institute of Biology and Biomedicine, noted that the study's innovation lies in creating a precise and interpretable predictive tool. "Our method for interpreting AI predictions allows us to identify the complex interrelationships between dozens of patient parameters," he stated. "For instance, the analysis reveals that a patient's age, the duration of their disease, and the number of diabetic complications are the most significant risk factors for mortality. This system can generate a personalized risk profile for each individual, such as illustrating that a 68% increased mortality risk is primarily driven by elevated creatinine levels, age, and the presence of four specific complications."
The model's development has also shed light on the significance of several lesser-known biomarkers. These include NT-proBNP, which indicates underlying myocardial stress; creatinine, a marker of kidney function; and specific N-glycan structures in blood serum, which serve as biomarkers for immune regulation and aging processes.
This breakthrough effectively transforms explainable AI from an abstract computational concept into a practical clinical tool that augments a physician's diagnostic decision-making. It represents a significant advancement with the potential to extend and enhance the lives of the millions of people worldwide living with diabetes.
--- Ends/Khalid
