Skip to content 
An AI system trained on photos of the face and tongue correctly identified every patient with coronary heart disease (CHD) in a small test, according to a new study from China.
The researchers recruited 518 people: 200 with diagnosed CHD and 318 healthy volunteers. Each had their face and tongue photographed using a purpose-built scanning system. The images were analyzed using deep learning, a type of AI that learns to see patterns in images.
The system examined the tongue and face photos separately, then combined what each captured to predict who had CHD. Using both together performed better than either alone. The best-performing version was 98% accurate in a test of 104 people, correctly identifying all 40 patients with CHD, although two healthy participants were incorrectly classified as having the condition – so-called “false positives”.
The AI focuses on the center of the tongue, its surface coating, the skin around the mouth and the cheeks. Some patients with CHD had darker tongues, thicker coatings, or changes in facial pigmentation. The researchers say that the areas highlighted by the AI are consistent with traditional Chinese medicine teachings about links between the tongue, face and heart.
Reading illness from appearance is not unique to traditional Chinese medicine. Doctors have long recognized that some visible features point to underlying medical problems. Blue lips can signal low oxygen levels in the blood, and greasy yellow deposits around the eyelids, known as xanthelasma, can indicate very high cholesterol, often caused by an inherited condition.
Whether a tongue or a cheek can detect blocked arteries is another matter. Skin color, exposure, age, medications and smoking all change how a face and tongue look, and the study does not find that these features are caused by blocked coronary arteries. AI can pick up associations that humans struggle to recognize, but that doesn’t explain what causes them.
CHD develops when the arteries that supply the heart muscle are narrowed due to fatty deposits called plaques, a process known as atherosclerosis. This restricts blood flow and can eventually cause a heart attack. For many people, the first sign of a problem is chest pain, a heart attack or another life-threatening event, so getting the disease early is an important target of preventive medicine.
Doctors currently estimate risk with factors including age, smoking, blood pressure, cholesterol and diabetes. If CHD is suspected, tests may include an electrocardiogram (ECG), blood tests, or specialized scans. A quick, painless, low-cost photographic test could help decide who needs these investigations, especially in communities where access to specialist health care is limited.
A promising result, with questions remaining
The study, published in Scientific Reports, compared people who have already been diagnosed with CHD against healthy volunteers, which is very different from general screening, where people have a mix of other diseases and levels of cardiovascular risk. The photos were also taken under controlled lighting with specialized equipment.
The model has not been tested on people from other hospitals or settings. AI systems sometimes perform very well on the data used to build them, but less well in different hospitals, ethnic groups or healthcare facilities.
Not all of the healthy volunteers were evaluated with coronary imaging, so undetected disease cannot be excluded. The AI may also have recorded differences related to age, other diseases or other characteristics of the participants rather than heart disease.
Screening adds another problem. Even a very accurate test can produce a large number of false-positive diagnoses when used on a population in which the disease is relatively uncommon.
Larger studies following people over time would need to show whether this approach can pick up a disease that goes unnoticed, distinguish it from other conditions and lead to better health. Researchers would also have to directly compare it to existing ways of assessing risk. If it proves useful, it could eventually be used alongside standard heart disease risk tests, cholesterol tests and blood pressure measurements.
