Revolutionary AI Tool Promises Better Diagnosis and Management of Type 1 Diabetes

Researchers from Western Sydney University have developed an innovative artificial intelligence (AI) based tool that aims to transform how type 1 diabetes (T1D) is diagnosed and treated worldwide. This groundbreaking risk assessment method leverages microRNAs—small RNA molecules found in blood samples—to accurately predict the risk of developing T1D and forecast individual responses to various therapies. The tool, known as the Dynamic Risk Score (DRS4C), was created by analyzing molecular data from nearly 6,000 samples collected globally, including participants from Australia, Canada, Denmark, Hong Kong, India, New Zealand, and the United States.

The AI-powered risk score not only helps detect early signs of T1D but also distinguishes T1D from type 2 diabetes (T2D), which is often misdiagnosed, especially in adults. It predicts, within an hour of therapy initiation, which patients are likely to remain insulin-free and which will respond to specific treatments such as islet transplantation or drug therapy like imatinib. This real-time, dynamic assessment enables clinicians to tailor treatments more effectively and intervene sooner, potentially delaying disease progression.

Professor Anand Hardikar, leading the project, emphasized that current T1D testing methods rely mainly on biomarkers and symptoms, which often appear late, making early diagnosis challenging. The new AI-based approach could offer a powerful tool for early intervention, especially crucial since early-onset T1D, particularly before age 10, tends to be more aggressive and can significantly reduce life expectancy.

Community concerns around genetic testing were also addressed, with researchers clarifying that most T1D cases occur without a family history, highlighting environmental factors in disease development. The dynamic risk score offers an adaptive, stigma-free alternative to traditional genetic assessments by reflecting current disease risk.

Further, the study's findings open avenues for applying this risk modeling approach to other diabetes types, particularly T2D. The research underscores the importance of early prediction and personalized treatment strategies, paving the way for improved outcomes in diabetes care.

Published in Nature Medicine, this research signifies a major advancement in diabetes research and presents promising potential for better disease management in the future.

For more detailed information, see the original study: Mugdha V. Joglekar et al, 'A microRNA-based dynamic risk score for type 1 diabetes,' Nature Medicine (2025).