Artificial intelligence researchers at Stevens Institute of Technology have created a system that detects when someone is eating and calculates their carbohydrate intake.
Stevens' Samantha Kleinberg said, "This brings us a step closer to the holy grail — an 'artificial pancreas' that can quickly detect glucose changes, and correct them with an insulin pump, without the user having to do anything."
The algorithm compares changes in glucose in the body to an archive of curves representing glucose changes resulting from eating many different foods, and identifying the curve resembling fluctuations and activity levels read over a certain period.
The algorithm refers to the dietary information for what it concludes the person just consumed, calculates the carbohydrate number, and dispatches the appropriate dose of insulin.
The algorithm can model carbohydrate intake to within 1.2 grams and identify meals within an average of 25 minutes.
From Stevens Institute of Technology
View Full Article
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA