Dennis Anselmo was glad he used an Apple Watch — it saved his life when it spotted a dangerous heart condition. He’s not unique. A range of studies already exist that prove Apple Watch can identify heart problems, but when its data is coupled with deep learning systems it can achieve much more.
Having already built the world’s most accurate wearable heart monitor, Apple’s sensor development teams must now be exploring new sensor technologies. There have been claims these include a non-invasive diabetes sensor. Why not? We know the company’s commitment to the sector runs deep.
The need to achieve regulatory approval for some of these tools before putting them on sale to the public acts as a brake to public launch.
Regulatory delay is a good thing, as it’s important we know the data is accurate and based on good science. Apple is also willing to work with third-party developers who can provide accessories to extend the capabilities of its devices.
Data analysis/machine intelligence/deep learning of the data itself has a big part to play in the future of medical care.
There are several examples of this, and many ways in which such information could yield actionable insights. This is because analytics can make sense of lots of different kinds of data.
Did you know that during the incubation period of the flu virus, many of us become more sociable as the illness seeks out new hosts? Wearable devices may be able to assess data like this, providing warning and helping curtail the spread of disease.
Straight from the heart
A team from UCSF Cardiology and Apple Watch developers, Cardiogram, have spent the last year developing and testing a deep learning A.I. they hope will accurately detect the warning signs of stroke.
The Apple Watch tech gathers your heart rate data and works to identify anomalies. (There is a good article explaining some of the complexities of this here.)
“Using a technique called semi-supervised deep learning, we’re building an algorithm that can distinguish between the pattern of heart rate variability found in normal heart rhythm and that in atrial fibrillation,” the developers say.
The technology builds on previous research that proves mobile device data can accurately assess cardiovascular health. If you use an Apple Watch you can contribute to this Research Kit study here.
“Whereas the average medical study struggles to recruit 100 patients, building our study on Apple Watch has let us recruit 10,000 participants in under a year. And unlike the average electronic medical record system, Apple provides clean APIs that let us easily access heart rate, step count, workouts, and more. All in all, we now have about 30 billion data points,” Cardiogram co-founder, Brandon Ballinger, told me.
Beyond the heart
It isn’t just about heart. The science suggests that heart rate data can act as a fairly accurate measure for other conditions, such as sleep apnea, high cholesterol, and others.
“It may seem surprising that Apple Watch’s heart rate sensor could detect a condition like diabetes,” Ballinger told me in an email. “But it’s already known in the medical literature that as people develop diabetes, their resting heart rate goes up and their beat-to-beat heart rate variability goes down, so the correlation is there. The surprising thing is just how accurate a deep-learning-based algorithm can get — we’ve found 80+ percent accuracy for many conditions.”
Ballinger recently told the HyperWellbeing Conference: “…the neural network algorithm beat hand-engineered biomarkers on all three of these tasks,” he said. He says Apple Watch data can detect abnormal heart rhythms with “95+ percent accuracy.”
The big data picture
Data captured by the Apple Watch is already being used to provide warning of poor heart health, but in conjunction with A.I. it already seems capable of identifying other problems early.
This isn’t the only scientific research that proves this. Johns Hopkins researchers are developing Apple Watch technology for epilepsy sufferers that identifies if they are about to have a fit before the fit takes place. Humana and Aetna are testing these devices. Ochsner Health has been using the watch to help people with chronic conditions track their health at home.
Researchers supported by the National Institutes of Health have revealed the ability of wearable biosensors to detect physiological changes that may indicate illness, even before symptoms appear.
“For those individuals who can’t easily come into a clinic, they’re still under a sort of health surveillance mechanism,” Jessilyn Dunn, Ph.D., a postdoctoral research fellow at Stanford University, told Science Daily.
This is the potential for Apple Watch and Apple’s solutions in digital health — to develop an ecosystem that can monitor for the warning signs of health problems, access medical help and suggest logical interventions before conditions become serious.
“There’s a really exciting potential for this type of technology to really revolutionize the health care model,” said Dunn. Assuming everyone can afford the tech, I suppose.