DigiNews

Tech Watch by Johan Denoyer

← Back to articles

Seeing beyond BMI: Estimating cardiometabolic risk with smartphone imagery

Quality: 8/10 Relevance: 9/10

Summary

Google Research outlines PhotoScan, a deep learning framework that estimates body composition from smartphone images to predict insulin resistance with accuracy near DXA. The approach uses pre-training on UK Biobank data, fine-tuning on the PhotoBIA cohort, and independent validation to demonstrate clinically meaningful performance, suggesting smartphone imagery can augment traditional adiposity metrics like BMI for cardiometabolic risk assessment.

🚀 Service construit par Johan Denoyer