Google AI Can Detect Body Fat More Accurately Than a Smartwatch
Technology-Portfolio.Net - Knowing body composition usually requires using X-ray analysis. However, a number of Google researchers are developing an easier way to provide similar results with a phone camera.
Google Research's system, PhotoScan, can estimate body composition using AI to analyze 2D body images.
The PhotoScan method initially trained the AI system using data from DXA and MRI scans, and was then refined using photos taken with a phone camera.
As a result, PhotoScan has proven capable of accurately estimating body fat compared with using sensor-based tools such as bioelectrical impedance analysis (BIA) on wearable devices.
Advantages of PhotoScan
In detecting body fat, the tool can measure the waist or leg circumference fat ratio (A/G ratio) as well as the ratio of visceral fat to subcutaneous fat (V/S ratio).
In addition, Google Research also wants to see how PhotoScan results can predict conditions such as insulin resistance.
This development effort can be seen in the Insulin Resistance Trends feature that will be available in Health Guardian on the Pixel Watch.
Although it appears promising, there is a user trust issue when it comes to sharing a number of photos that may look less than ideal through the PhotoScan tool.
Even with the highest level of privacy and efforts to encourage users to change their eating habits and lifestyle, this will be the biggest issue to be faced.
Nevertheless, it is hoped that this tool can make things easier for users and eventually become one of Google Health's services in the future.