AI-powered tool predicts type 1 diabetes risk and treatment outcomes with precision

Researchers at Western Sydney University have developed an AI-powered tool that uses dynamic risk scores based on blood microRNAs to predict type 1 diabetes progression and treatment response. The tool can also help distinguish between type 1 and type 2 diabetes.

New Delhi: Australian researchers have pioneered a new artificial intelligence (AI)-powered tool to assess the risk of developing type 1 diabetes (T1D).

The tool, developed by researchers at Western Sydney University, predicts treatment responses, potentially changing how the disease is diagnosed and managed.

---

The tool utilises an innovative risk score — Dynamic Risk Score (DRS4C) which can classify individuals as having or not having T1D.

It is based on microRNAs — small RNA molecules measured from blood — to help accurately capture the changing risk of T1D.

“T1D risk prediction is timely, with therapies that can delay T1D progression becoming recognised and available. Since early-onset T1D before the age of 10 years is particularly aggressive and linked to up to 16 years of reduced life expectancy, accurately predicting progression gives doctors a powerful tool to intervene sooner,” said Professor Anand Hardikar, lead investigator from the University’s School of Medicine and Translational Health Research Institute.

In their article published in the journal Nature Medicine, the research analysed molecular data in 5,983 study samples from participants across India, Australia, Canada, Denmark, Hong Kong, New Zealand, and the US, to develop DRS4C.

By leveraging AI, the researchers enhanced the risk score, which was validated in 662 other participants. Just an hour after therapy, the risk score predicted which individuals with T1D would remain insulin-free.

In addition to T1D risk and drug efficacy prediction, the risk score could potentially discriminate T1D from Type 2 diabetes.

Dr. Mugdha Joglekar, lead researcher, from the School of Medicine and Translational Health Research Institute at the University, explained the difference between genetic and dynamic risk markers, adding that genetic testing offered a static view of risk.

“Genetic markers identify lifelong risk, it’s like knowing you live in a flood zone, but dynamic risk scores offer a real-time check on the rising water levels; it reflects current risk rather than a lifelong sentence, allowing for timely and adaptive monitoring without stigma,” said Joglekar.