The dawn of fitness wearables has allowed us to track exercise, count steps and generally look like we know what we’re doing in the gym. But arguably their full potential has yet to be fully harnessed.
A study has just been published showing that everyday consumer wearables, including the Apple Watch, Android Wear, and products from Garmin, are capable of detecting the most common abnormal heart rhythm with 97 percent accuracy.
The study was carried out by mobile health data startup Cardiogram in conjunction with the University of California, San Francisco (UCSF). Titled, ‘Passive Detection of Atrial Fibrillation Using a Commercially Available Smartwatch’, it’s available to read online at JAMA Cardiology.
Read more: Health IoT: KardiaBand sensor could replace invasive blood tests
Training a deep neural network
Cardiogram and UCSF have developed a deep neural network capable of detecting atrial fibrillation. The condition will affect 25 percent of us at some point in our lives, is responsible for 25 percent of all strokes, and is often left undiagnosed.
“By using software to transform ordinary wearables into personal health monitors, we can literally save lives,” said Cardiogram co-founder Brandon Ballinger.
The software, named DeepHeart, is the result of 9,750 Cardiogram users taking part in UC San Francisco’s Health eHeart Study. Together, they contributed 139 million heart rate and step count measurements, which were used by the DeepHeart neural network.
To validate the software, 51 cardioversion patients at UCSF were tested. DeepHeart was able to distinguish between normal heart rhythm and atrial fibrillation with an accuracy of 97 percent.
Read more: Health IoT: App helps sports stars predict and manage injuries
Consumer wearables for health
Cardiogram claims to have over 750,000 monthly users, with 78 percent of those using its mobile application every day. That engagement rate, which is higher than that of the most popular social media networks, has coincided with the company’s recent research on major health conditions: hypertension, sleep apnea, diabetes, and, this week, atrial fibrillation.
Read more: Fitbit and Apple Watch can help predict diabetes, says report
“Over the last year, we’ve presented research on four major health conditions. The link between these health conditions and heart rate comes from the autonomic nervous system,” explained Ballinger.
“As you develop hypertension, for example, your pattern of beat-to-beat heart rate variability shifts – so your heart is not just an important organ in its own right, but also a vantage point into the rest of your health.”
The Cardiogram application is compatible with the Apple Watch, and any Android Wear watch with a heart rate sensor, including models from Huawei, LG, New Balance, and Montblanc.
Internet of Business says
Personal healthtech devices and other wearables that can be optimised to monitor health and fitness have been one of the big stories this year, alongside smart/connected car partnerships (good news) IoT security (bad news), and AI ethics (calls for action).
The revelation that wearables’ key application isn’t to offer us a window into our social world, as many people had thought, but a window into ourselves and how our own bodies are performing, has been transformative, and the levels of user engagement prove this.
The IoT is saving lives with data, and encouraging us to look after ourselves better: as positive an application of networked computing and analytics as you could wish for.