Malnutrition is a typical downside at long-term care properties, as many residents merely do not eat all of the meals which is served to them. An experimental new AI-based system is designed to assist, by analyzing pictures of their plates.
The know-how is presently being developed through a partnership between Canada's College of Waterloo, Schlegel-UW Analysis Institute for Growing old, and College Well being Community.
Placing it merely, the system compares images of platefuls of meals. One photograph is taken earlier than the meal is served, whereas the opposite is taken as soon as the resident has completed consuming – ideally, these pictures could possibly be obtained by workers through a pill.
As a result of the software program utilized by the system is linked to the care dwelling's recipes and meal plans, it already is aware of the kind and the dietary worth of the meals included in every explicit meal. By analyzing elements equivalent to the colour and depth of the meals that is left on the plate, it's subsequently in a position to decide how a lot of every meals kind was consumed, and thus how a lot diet the resident acquired.
If it is discovered that the resident has an inadequate or nutritionally unbalanced food regimen, caregivers will probably be notified.
And whereas workers at care properties do already estimate meals consumption by manually checking how a lot meals is left on plates, the scientists state that such methodology has been proven to have an error charge of no less than 50 p.c. Against this, the AI-based system is claimed to be correct to inside 5 p.c.
A paper on the research, which is being led by Waterloo researchers Kaylen Pfisterer and Robert Amelard, was just lately printed within the journal Scientific Reviews.
Supply: College of Waterloo
Post a Comment