Sign in to view
member only content

Spotted: In ancient Chinese medicine, the condition of the tongue was used as an important diagnostic tool, indicating the potential presence of disease. Now, building on this idea, an algorithm has achieved a 98 per cent accuracy in predicting multiple diseases by analysing the colour of the human tongue.

The system is based on the fact that the colour and condition of the tongue can actually be used to diagnose a number of health conditions. For example, people with diabetes often have a yellow tongue; a white tongue can indicate anaemia; and an indigo- or violet-coloured tongue might indicate gastrointestinal issues or asthma.

The imaging system has been developed by engineering researchers from Iraq’s Middle Technical University and the University of South Australia, who used 5,260 images to train machine learning algorithms to detect tongue colour.

Images of tongues from people with various health conditions were supplied by two teaching hospitals in the Middle East. After training, the AI was able to match the tongue colour with 98 per cent accuracy. Although the algorithm wouldn’t replace a real doctor, it could help speed up diagnosis, saving money as well as time. The research team has recently published a paper in Technologies, outlining how the proposed system could provide on-the-spot diagnosis.

Written By: Lisa Magloff