A new type of AI health assistant can accurately identify health conditions just by looking at the tongue. Medical researchers have designed the assistant to look for specific colors and conditions on the tongue. The algorithm behind the AI has proven so effective that it can accurately identify the patient’s conditions in 98 percent of cases.
The new AI is part of research described in an article published in the journal TechnologiesThe researchers used 60 photos from teaching hospitals in the Middle East and fed them into the algorithm. The photos were all taken by different patients, all suffering from different illnesses such as COVID-19 and even gastrointestinal problems.
In these tests, the AI healthcare assistant was able to accurately identify what disease the patient was suffering from based on the color of the tongue alone. However, developing this system was no easy task. The researchers first took 5,200 photos of the patients’ tongues and then added the relevant information to their diagnoses using six different ML algorithms.
This allowed them to accurately predict tongue color under all lighting conditions, the paper says. The researchers say that tongue color, thickness and shape can reveal a “litany of health conditions,” and that developing an AI that can understand this will help doctors detect and diagnose conditions much more reliably.
While the process relies on a healthcare AI assistant to do the heavy lifting, it actually builds on medical practices that date back centuries, the researchers note in their paper. The ability to understand the condition of the tongue can help us understand what’s going on inside a patient’s body without having to invade their privacy more than necessary, and this type of medical AI will only make that easier.
