Google's AI camera tool estimates body fat with near-DXA accuracy, research shows
Google's investigational PhotoScan AI estimates body composition from smartphone photos with near-DXA accuracy, according to new research.
The feature builds on Google's new Insulin Resistance Trends health feature, which was introduced alongside the Pixel Watch 5 last week. Insulin Resistance Trends provides month-by-month estimates of how the body responds to blood sugar spikes.
Body composition is a key marker of insulin resistance and metabolic health. Medical-grade DXA scans use X-rays but lack scalability. Wearable devices often use bioelectrical impedance analysis, or BIA, which sends a low-level electric current through the body to estimate fat, muscle, and bone percentages. Examples include smart scales and smartwatches such as the Samsung Galaxy Watch Ultra 2.
According to Google's research, PhotoScan was developed by matching smartphone photos with medical-grade DXA scans and other subject information, then running the data through an AI model. The model achieved strong DXA agreement when estimating body fat percentage and other metrics, leading to near-DXA accuracy for predicting insulin resistance.
The researchers wrote that clinical DXA imaging is accurate but lacks scalability, while wearable BIA sensors are convenient but limited to basic body fat percentage. They described PhotoScan as offering a promising middle ground, estimating granular body composition from standard smartphone imagery with near-DXA accuracy. The technology is not yet available to the public, but the research demonstrates the feasibility of smartphone-based body composition estimation.