Google’s camera tool estimates body fat more accurately than wearables
▼ Summary
– Google Research developed PhotoScan, an AI system that estimates body composition from 2D smartphone photos, trained on DXA X-ray and MRI data.
– PhotoScan estimates body fat more accurately than wearable bioelectrical impedance analysis (BIA) sensors, and can also measure A/G and V/S fat ratios that BIA cannot.
– The system shows potential for predicting insulin resistance, aligning with Google’s broader health initiatives like the Pixel Watch’s insulin resistance trends feature.
– A key challenge is user reluctance to share unflattering body photos, despite privacy safeguards, which could hinder adoption.
– No product plans are announced yet, but the technology may appear as a future Google Health offering; the article notes that motivating lifestyle changes remains a separate, difficult issue.
The numbers are stark. According to the NIH, roughly 73.1% of American adults are currently classified as overweight or obese. But for most people, understanding their true health profile goes far beyond a simple number on a bathroom scale. The most precise methods for measuring where we store fat and how it’s distributed typically require sophisticated X-ray technology. Researchers at Google believe they have found a far more accessible alternative, one that relies on the camera already sitting in your pocket.
The system, known as PhotoScan, uses artificial intelligence to estimate body composition from a series of standard 2D images. To build its foundation, Google fed the AI with data derived from dual-energy X-ray absorptiometry (DXA) scans. These scans provided baseline metrics on total body fat, along with critical ratios such as the distribution between core and limb fat (A/G ratio) and the balance between visceral and subcutaneous fat (V/S ratio).
The team then layered in MRI data and refined the model further using actual photos snapped on smartphone cameras. The final result was a model that could visually estimate body fat with greater precision than the bioelectrical impedance analysis (BIA) sensors found in most modern wearables. Perhaps more impressively, PhotoScan demonstrated strong accuracy in predicting the A/G and V/S ratios, measurements that BIA technology is simply not equipped to calculate.
The potential applications extend beyond just tracking fat. Google Research also explored whether PhotoScan could serve as a predictor for conditions like insulin resistance. This fits into a broader pattern of the company’s recent focus on metabolic health, a priority also visible in the upcoming Insulin Resistance Trends feature planned for the Pixel Watch’s Health Guardian suite.
The science behind the project appears robust, and the appeal of a non-invasive, low-cost method for obtaining meaningful health data is obvious. However, the path to real-world adoption is fraught with hurdles. Convincing users to upload a series of unflattering photos to a digital tool requires a significant leap of faith, even with a bulletproof privacy policy in place. And beyond the initial data collection lies the most stubborn challenge of all: translating that information into lasting dietary and lifestyle changes. Showing someone their visceral fat ratio is one thing; motivating them to alter their daily habits is an entirely different, far more complex battle.
Google has not announced any concrete plans to launch PhotoScan as a consumer product. Still, given the trajectory of its health-focused initiatives, it would not be surprising to see this technology resurface in a future Google Health offering, potentially giving us all a clearer look at what is happening beneath the surface.
(Source: Android Authority)
