Google Develops AI Tool to Assess Body Fat from Photos

Researchers at Google have introduced a new technology that enables body composition assessment using standard smartphone photos. The system, called PhotoScan, leverages artificial intelligence to analyze a series of 2D images and determines not only overall fat levels but also their distribution throughout the body.

To train the model, Google utilized data from X-ray scans (DXA method) and MRI studies. This allowed the system to learn how to identify key metrics, such as the ratio of fat in the abdominal area to the legs (A/G) and the ratio of visceral to subcutaneous fat (V/S). Wearable devices using bioelectrical impedance cannot provide such detailed data.

In addition to body composition assessment, PhotoScan can predict insulin resistance—a condition often linked to type 2 diabetes. This research direction is particularly relevant for Google, which is already integrating similar features into its devices, such as the Pixel Watch smartwatch.

Despite the technology’s potential, several challenges must be addressed before its widespread adoption. Users may be reluctant to share photos of their bodies, even if the system guarantees privacy. Additionally, it remains unclear how to effectively motivate people to change their lifestyles based on the data provided.

Google has not yet announced plans for the commercialization of PhotoScan, but the technology could potentially become part of Google Health services in the future.