Google Research has described a method for estimating cardiometabolic risk from ordinary smartphone photographs. The approach uses computer vision to analyze images of the body, potentially capturing signals that go beyond what body mass index (BMI) provides. BMI is a crude metric that does not distinguish muscle from fat or reflect fat distribution, which is closely tied to metabolic health.
The researchers trained models on a dataset of images and corresponding clinical measurements, allowing the system to predict risk-related indicators from photos alone. If validated further, the technique could make preliminary risk screening more accessible, especially in settings where clinical exams or lab tests are not readily available.
Because the work is presented in a research blog, it should be read as a progress report rather than a proven clinical tool. The post does not claim that smartphone imagery can replace medical diagnosis, but it suggests the approach could complement existing screening methods.