Artificial intelligence first detects type 1 diabetes in children

Predictive tool that uses artificial intelligence offers hope for early diagnosis of type 1 diabetes in children across the UK, which reduces the risk of malignant diabetic ketoacidosis (DKA), according to early research presented at the Diabetes UK Professional Conference 2022.

Type 1 diabetes is a serious autoimmune condition that cannot be prevented and the gradual destruction of insulin-producing beta cells can begin months or even years before diagnosis. Symptoms usually appear very close to the diagnosis. It is important to diagnose the signs and symptoms of diabetes early and to make sure that both children and adults who develop it do not become seriously ill.

A quarter of children and adolescents (25%)1) They will not be diagnosed with type 1 diabetes as long as they are in DKA2A life-threatening condition that can even lead to coma or death.

With funding from Diabetes UK, Dr. Julia Townson, a senior research fellow at Cardiff University, and her team used artificial intelligence and electronic health records to develop an assessment tool. The tool was trained on data from 1 million children to identify patterns in children’s GP records that indicate they have undiagnosed type 1 diabetes.

To find out what can best diagnose type 1 diabetes, the team tested a combination of factors different from GP records, such as urinary tract infections or bedwetting, family history of type 1 diabetes, and prescribing antibiotics.

To see if the Predictive Tool could accurately diagnose children with type 1 diabetes already, researchers estimated how well it worked using GP and hospital records from another 1.5 million children.

When the researchers set the level to 10% (i.e. 1 in 10 children under the age of 15 will be alerted by consulting a GP), the tool successfully detects 75% of children who develop type 1 over the next 90 days. On average, it allowed children to be diagnosed – and started life-saving insulin therapy 11 days earlier than they were, which could be the difference between life and death for anyone.

More work is now needed to test whether the new tool can be widely used with GPs and in patients in primary care. Following the presentation at the Diabetes UK Professional Conference 2022, the Cardiff team will also present their research for peer review and publication in the journal as a routine process for scientific research.

Dr. Lucy Chambers, Head of Research Communications at Diabetes UK, says:

“Type 1 diabetes is not currently preventable and is a very serious condition. If left undiagnosed, it can progress very quickly and lead to dangerous complications, so finding a way to catch the condition and treat it early can help prevent emergency hospital treatment and save lives.

“This research, funded by Diabetes UK, will be used by the NHS to warn GPs about the possibility of developing type 1 diabetes in children in the future, helping more children to make an accurate and rapid diagnosis and to start living as best as possible with type 1 diabetes.”

Dr. Julia Townson, a senior research fellow at the Cardiff University Center for Trials Research, said:

“We are now looking at how this tool works with primary care computer systems. If possible, this will enable us to conduct a feasibility study with GPs to assess the viability of the tool. The use of the tool may still be a few years away – but we hope it will provide a crucial new way to early diagnosis of life-saving diabetes.

Beth Baldwin of Cardiff loses his son Peter in the DKA. Peter was just 13 years old when he died in January 2015 of undiagnosed type 1 diabetes. She says:

“Life without Peter is difficult, but the ability to accelerate the diagnosis of type 1 diabetes in others in the future and his memory will save lives, guide us and move us forward.

“If our work stops even a family with the experience of losing a child with type 1 diabetes, it will be worth it.”

Beth has been campaigning tirelessly to raise awareness about type 1 diabetes and its symptoms in memory of her son. She added:

“It is important for parents to be aware of the 4 Ts – the toilet, the thirst, the fatigue, the thinness – so that they can detect the symptoms of type 1 diabetes in their children as quickly as possible. If detected, it is important to take immediate action and seek medical help.

Currently, the best way to diagnose type 1 diabetes early is to learn about its symptoms. The four most common symptoms of type 1 diabetes are called ‘The 4Ts’:

  • Toilet – going to the toilet too much, getting a baby bed wet in the past or heavy nappies in babies
  • Thirst – really thirsty and unable to quench thirst
  • Fatigue – Feels tired more than usual
  • Skinny – Weight loss will be followed by fatigue and constant tiredness

Leave a Comment