New study claims Apple Watch to accurately detect diabetes symptoms

Cardiogram, the notable mobile health data firm, has apparently revealed that Apple’s wearable watch integrated with a heart rate sensor can detect the symptoms of diabetes in patients. The organization’s patent machine learning tool, DeepHeart, already deploys Apple Watch’s heart rate sensor to determine whether the wearer has sleep apnea, hypertension, and atrial fibrillation. Now however, diabetes has also been added to the list.

In order to validate its claims, Cardiogram, in collaboration with the University of California, performed clinical tests on nearly 14011 patients. More than 200 million sensor measurements were taken from these patients using Apple Watch. The combined team of researchers from both Cardiogram and UC then used the DeepHeart to determine the accuracy of data collected during the trials. According to Macworld, Apple Watch’s measurements were 85% authentic in determining the difference between the people with diabetes and without diabetes. One of the cofounders of Cardiogram also stated that one in every four cases of diabetes goes undetected, however, this methodology may significantly aid in reducing the rate of non-detection of the diabetes.

Apparently, this prediabetic condition often goes undiagnosed, since conventional procedures of diabetes detection require glucose monitoring device. However, experts claim that Apple Watch & Cardiogram’s DeepWatch neural network together have the ability to detect the diabetes and inform patients to follow up with physicians in case of any issues. Some of the key officials of Cardiogram have also stated that the study has depicted how a heart rate sensor used in the Apple Watch can help in detecting diabetic symptoms in an individual.

Post Cardiogram’s announcement of the Apple Watch being able to detect diabetic symptoms, it has been speculated that this new study can prove to be highly beneficial for individuals detected with this disease, helping them conduct timely treatment.