Google Research uses smartphone photos to estimate body composition and predict insulin resistance

soumitrashukla9 · x · 2026-08-18

Google Research published a blog post introducing PhotoScan, a deep learning approach that estimates body composition from smartphone photos. In a clinical research setting, the method predicts insulin resistance with accuracy comparable to DXA scans. The study aims to identify metabolic risks earlier, such as insulin resistance (HOMA-IR score > 2.9), by combining wearable sensor data with body composition assessment before blood sugar levels rise.

Original post →

More from Apps

Apps channel →