Researchers at the University of California San Francisco have conducted a study demonstrating the feasibility of DeepHeart, a deep neural network programmed to detect diabetes.
DeepHeart can accurately predict cardiovascular risk while requiring less labeled data than conventional deep-learning techniques.
The researchers validated the accuracy of DeepHeart in distinguishing between people with and without diabetes, realizing 85% accuracy on a large dataset.
"This is the first large-scale study that shows a regular heart rate sensor can be used in conjunction with an [artificial intelligence]-based algorithm to identify early signs of diabetes," says Hot Hardware journalist Paul Lilly.
The algorithm is designed to measure shifts in the pattern of heart rate variability, which manifest as people develop early stages of diabetes.
In addition, DeepHeart aims "to help screen for diabetes in people who otherwise had no idea they were at risk for it."
From Tech Xplore
View Full Article
Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA
No entries found