AI system identifies coronary heart disease via facial and tongue photos

A new study from China has demonstrated that an AI system trained on photographs of the face and tongue can identify patients with coronary heart disease (CHD). In a study involving 518 participants—including 200 with diagnosed CHD and 318 healthy volunteers—researchers used deep learning to analyze images captured by a purpose-built scanning system.

The AI examined tongue and facial photographs both separately and in combination, finding that using both methods together yielded better predictive results. In a test of 104 people, the most effective version of the system achieved 98% accuracy, correctly identifying all 40 patients with CHD, while recording only two false positives.

The AI identified patterns in the center of the tongue, surface coatings, and skin around the mouth and cheeks. Researchers noted that these findings align with traditional Chinese medicine teachings regarding the links between the tongue, face, and heart. However, the study noted that factors such as skin tone, lighting, age, medication, and smoking can influence these visual features.