A new study conducted by Cardiogram from the data collected by Apple Watch users indicate that it is possible to detect early signs of diabetes using the smartwatch. The data collected was 85 percent accurate in distinguishing between diabetic and non-diabetic users.
The study was conducted by a team of researchers at Cardiogram and University of California who used the Cardiogram DeepHeart neural network to analyze the collected data. The data pool comprised of 14,011 users, with a separate set of data being used for comparison purposes to determine the accuracy of the neural network.
The study included not only the Apple Watch but also Android Wear running smartwatches and wearables from Fitbit and Garmin. This is not the first time that Cardiogram has conducted such a study. It had previously trained its DeepHeart neural network to detect hypertension, sleep apnea, and atrial fibrillation.
Diabetes affects millions of people across the world, with many patients suffering from pre-diabetes without even knowing about it. So far, the only way to detect pre-diabetes is by regularly checking the glucose level in the blood. However, as the study from Cardiogram shows, it is actually possible to use an Apple Watch or any other wearable containing a regular optical heart rate scanner to detect early signs of diabetes.
Cardiogram’s co-founder Brandon Ballinger says that DeepHeart AI is not designed to remove doctors and medical experts from the process of determining whether patients have pre-diabetes/diabetes or not. Instead, it is aimed at ensuring that people at risk are able to get the medical help they require on time. He confirms that the Cardiogram app will gain some kind of pre-diabetic screening using DeepHeart, though fails to mention when.
Ballinger also says that Cardiogram could be used to offer a full diagnosis of diabetes if Apple ends up including the necessary blood pressure sensor or glucose sensor on a future Apple Watch variant.