How’s this for ingenuity? It’s not just the very latest and priciest wearable gadgets that can offer cutting-edge health tracking.
Researchers at digital health startup Cardiogram and UCSF (University of California San Francisco) have found that DeepHeart, a deep neural network, is managing to distinguish between people with diabetes and those without with a strikingly high accuracy rate – 85%.
And this is across a big data set of 200 million heart rate and step count measurements.
What happened was that 14,011 users of the Cardiogram app on Apple Watch and Android Wear devices with heart sensors were recruited into the study. Then, substantial amounts of health sensor data were used to train the neural network.
How, you might ask, can step count data reveal diabetes?
Johnson Hsieh, one of the founders of Cardiogram, explained: ‘Your heart is connected with your pancreas via the autonomic nervous system. As people develop the early stages of diabetes, their pattern of heart rate variability shifts. In 2015, the Framingham Heart Study showed that high resting heart rate and low heart rate variability predicts who will develop diabetes over a 12-year period. In 2005, the ARIC study showed that heart rate variability declines faster in diabetics than non-diabetics over a 9-year period.’
In other words, the data has been studied to spot correlations or anomalies, so that if something’s spotted, you might make a trip to the doctor to check things out.
Cardiogram says that more than 250,000 people use the app, around three quarters of them on a daily basis.
DeepHeart has had its accuracy validated before. So why does this matter?
Well, because, as Hsieh points out, ‘These health conditions are important since 19% of people with hypertension, 80% of people with sleep apnea, and roughly 20% of people with atrial fibrillation remain undiagnosed.’
The most interesting thing is that Cardiogram works with current wearables including, as mentioned, the first Apple Watch. Now that a neural network such as DeepHeart can analyze the data from apps like this in such a complex way, the possibilities are exciting and wide-ranging.
The capability to predict diabetes is not in the Cardiogram app in explicit terms, but gathering data is the first, intriguing step, so that those suspected to be at risk can be encouraged to visit a health professional. Expect more developments in due course.