Apple Watch app 85 per cent accurate in diagnosing diabetes

A technological breakthrough means smart watches could pave the way for diagnosing diabetes people in the future.
A study, based on data from 14,000 users of DeepHeart, a popular Apple Watch app, has shown the wearable technology was able to identify people with diabetes with 85% accuracy. Examples of wearable technology include Apple Watch, Android Wear and Fitbit.
The technology comprises a built-in sensor which works alongside a “neural network”. The DeepHeart app uses an artificial intelligence algorithm that takes into account the wearer’s heart rate and step count.
The heart and pancreas are linked via the body’s nervous system, so when a person starts to develop diabetes their heart pattern changes.
The pioneering wearable kit also showed it could accurately detect high cholesterol, high blood pressure and sleep apnea to 74%, 81% and 83% accuracy respectively.
The research was a joint project between a health app development company health app Cardiogram and the University of California San Francisco (UCSF).
Cardiogram co-founder Johnson Hsieh said: “Researchers at Cardiogram and UCSF validated the accuracy of DeepHeart, a deep neural network, in distinguishing between people with and without diabetes, achieving 85 per cent accuracy on a large data set which included 200 million heart rate and step count measurements.”
Early detection of type 2 diabetes could help people seek treatment much earlier, which in the long term means they could avoid further related health complications.
Brandon Ballinger, another Cardiogram co-founder, said: “While there have been many attempts to build special-purpose glucose-sensing hardware to detect diabetes, this is the first large-scale study showing that ordinary heart rate sensors – when paired with an artificial intelligence-based algorithm – can identify early signs of diabetes.
“By detecting diabetes earlier, we can help people live longer and healthier lives.”
The findings were presented on Wednesday 7 February at the Thirty-Second AAAI Conference on Artificial Intelligence in New Orleans.