08 February 2018
A new study by Cardiogram, a company that offers an app which can break down heart rate data collected by the Apple Watch, stated that the Apple Watch can be used to detect the signs of diabetes.
Cardiogram researchers teamed up with the University of California, San Francisco, for the study. Cardiogram’s DeepHeart neural network was able to establish that the Apple Watch’s heart rate data was approximately 85% accurate at differentiating between people with diabetes and people without diabetes.
Cardiogram collected 200 million sensor measurements from 14,011 participants, using an Apple Watch or Android Wear device. The Cardiogram app was able to aggregate the data that included heart rate, step count, and other activity.
The company stated that Prediabetes is a condition that often goes unnoticed and undiagnosed because traditional methods of detection require glucose-sensing hardware. Apple Watch and an AI-based algorithm like Cardiogram's DeepHeart has the potential to alert users if there is an indication of prediabetes, after which the users can get the same verified from medical professionals.
Cardiogram stated that its study is the first large-scale study which demonstrates how an ordinary heart rate sensor, like the one in the Apple Watch and other devices like the Fitbit, can detect early signs of diabetes. This is because the pancreases are connected to the heart through the nervous system. If the person is diabetic the heart rate variability changes when a person starts experiencing symptoms of diabetes.
Early detection of diabetes could be key in cutting down on diabetes-related diseases. Diabetes is a growing problem in the US as well as across the globe. In the US alone 100 million adults suffer from pre-diabetes and nearly 1 out of 4 goes undiagnosed. Ballinger and his colleagues stated that they plan to look at a number of diseases which can be detected via heart sensors.