For 2,000 years, traditional Chinese medicine has been diagnosing diseases by examining a person's tongue coating. Now, artificial intelligence and machine learning technologies are enhancing this approach. A collaborative study by Iraqi and Australian researchers shows that a computer-assisted tongue diagnosis system can accurately identify conditions such as diabetes and kidney failure in 94% of cases. This diagnostic method, which often uses simple devices such as smartphones, offers a promising and cost-effective alternative to remote health monitoring, especially as medical services are limited by the COVID-19 pandemic.
Traditional Chinese medicine has been examining the tongue to spot signs of disease for 2,000 years, and now computer scientists are using machine learning and artificial intelligence to do just that.
A study by Iraqi and Australian researchers provides more evidence that the technology is increasingly accurate at detecting disease.
Engineers at Baghdad's Medium Technical University (MTU) and the University of South Australia (UniSA) used USB webcams and computers to capture images of the tongues of 50 patients with diabetes, kidney failure and anemia and compared their colors with a database of 9,000 tongue images.
Using image processing technology, they correctly diagnosed 94% of cases compared to lab results and were also able to send a voicemail via text message to the patient or designated healthcare provider explaining the tongue color and disease.
In a new paper published in the AIP conference proceedings, Ali Al-Naji, adjunct associate professor at MTU and UniSA, and colleagues review global progress in computer-aided disease diagnosis based on tongue color.
He said: "Thousands of years ago, Chinese medicine pioneered the practice of examining the tongue to detect disease. Traditional medicine has long recognized this method, proving that the color, shape and thickness of the tongue can reveal signs of diabetes, liver problems, circulatory and digestive system problems, and blood and heart diseases. Taking this a step further, artificial intelligence and cameras can now be used to even is a new way to diagnose diseases remotely from the appearance of the tongue using a smartphone. Computerized tongue analysis is very accurate and can help diagnose diseases remotely in a safe, effective, easy, painless and cost-effective way. This is especially important in the wake of a global pandemic like COVID, where accessibility to medical centers may be affected."
Diabetic patients often have yellow tongues, cancer patients have purple tongues with a thick layer of grease, and acute stroke patients have red tongues that are often curved.
A 2022 study in Ukraine analyzed tongue images of 135 COVID patients via smartphones and showed that 64% of patients with mild infection had light pink tongues, 62% of patients with moderate infection had red tongues, and 99% of patients with severe COVID-19 infection had dark red tongues.
Previous studies using the tongue diagnostic system have also accurately diagnosed appendicitis, diabetes and thyroid disease.
"It is possible to diagnose more than 10 diseases that cause noticeable changes in tongue color with an accuracy of 80%. In our study, we achieved a diagnostic accuracy of 94% for three diseases, so we have the potential to further improve this research," Associate Professor Al-Naji said.