AI identifies heart disease in trial using face and tongue study
An artificial intelligence system trained on photographs of the human face and tongue correctly identified every patient diagnosed with coronary heart disease in a small trial conducted in China, according to a study published in Scientific Reports.
-
Key Clinical Takeaways:
- The AI system achieved up to 98% accuracy in a test of 104 people by analyzing separate photographs of the face and tongue.
- Researchers recruited 518 participants for the study, split between 200 diagnosed coronary heart disease patients and 318 healthy volunteers.
- David C. Gaze, a senior lecturer in chemical pathology at the University of Westminster, notes that while the results are promising, the study does not establish causation between visible facial features and blocked coronary arteries.
How did the artificial intelligence system test for coronary heart disease?
Researchers recruited 518 individuals for the study, which included 200 people with a confirmed diagnosis of coronary heart disease and 318 healthy volunteers. Each participant had their face and tongue photographed using a purpose-built scanning system under controlled lighting. A deep learning AI analyzed these images by examining the face and tongue photographs separately before combining the data to predict who had the condition. The best-performing version of the model demonstrated 98% accuracy in a secondary test of 104 people, correctly identifying all 40 patients with coronary heart disease, though it misclassified two healthy participants as positive.
What specific facial and tongue features did the algorithm target?
The deep learning model concentrated its analysis on the center of the tongue, the surface coating of the tongue, the skin surrounding the mouth, and the cheeks. Patients with coronary heart disease frequently displayed darker tongues, thicker surface coatings, or noticeable changes in facial pigmentation. The research team noted that these targeted anatomical zones align closely with traditional Chinese medicine teachings regarding the observed physiological connections between the heart, face, and tongue.
Why do medical experts urge caution despite the high accuracy rates?
Writing in an analysis republished from The Conversation, David C. Gaze emphasized that reading disease from physical appearance is not unique to traditional medicine, pointing out that blue lips signal low oxygen levels and yellowish eyelid deposits called xanthelasma indicate very high cholesterol. However, Gaze cautioned that skin tone, ambient lighting, age, medications, and smoking habits all alter facial and tongue appearances. The study design compared already-diagnosed patients against healthy volunteers rather than screening a general population, meaning the model has not yet been validated across diverse ethnic groups, external hospitals, or varying healthcare settings.
Disclaimer: The information provided in this article is for educational and scientific communication purposes only and does not constitute medical advice. Always consult with a qualified healthcare provider regarding any medical condition, diagnosis, or treatment plan.