Can AI detect certain diseases by looking at images of faces ?

Current AI systems can extract suggestive signals from facial photographs—changes in texture, asymmetry, pigmentation, and subtle swelling—that correlate with certain metabolic, cardiac, and endocrine disorders, but these cues are not disease-specific and often overlap with normal variation or other conditions. Research groups have reported modest accuracies (often 60–80 % AUC) for detecting diseases such as diabetes, chronic kidney disease, or coronary artery disease, relying on large datasets and deep-learning models trained on tens of thousands of labeled images. Because facial biomarkers are indirect and influenced by age, sex, lighting, and ethnicity, the technology remains investigational and is not approved for clinical diagnosis. It is currently used mostly in research settings and as an adjunctive screening tool rather than a diagnostic standard.

— Enriched May 13, 2026 · Source: Nature Medicine

Researchers have been exploring the use of artificial intelligence to detect certain diseases by analyzing images of faces, a field known as facial phenotyping. This approach is based on the idea that certain diseases can cause subtle changes in facial features, which can be detected using computer vision algorithms. For example, some studies have shown that AI can be used to detect genetic disorders such as Down syndrome and DiGeorge syndrome by analyzing facial images. Other diseases, such as Parkinson's disease and Alzheimer's disease, have also been the focus of facial phenotyping research. The use of deep learning techniques, such as convolutional neural networks, has improved the accuracy of facial phenotyping systems. However, the development of these systems is still in its early stages, and more research is needed to fully realize their potential. Facial phenotyping has the potential to provide a non-invasive and low-cost method for disease detection, which could be particularly useful in resource-poor settings. The technique is not yet widely used in clinical practice, but it has shown promising results in research studies.

— Enriched May 13, 2026 · Source: National Institutes of Health