This image provided by the Korea Institute of Oriental Medicine shows its newly-developed method that extracts color and shape variables for each region of patient’s face.
DAEJEON, April 15 (Korea Bizwire) — The Korea Institute of Oriental Medicine said Wednesday that it can detect high blood pressure patients simply by the color and shape of an individual’s face in a photo.
On the basis of about 1,000 photos of high blood pressure patients and normal persons, the research team quantitatively extracted color and shape variables for each region of their faces using ‘mang-jin,’ the method of examining patients by observing symptoms on their face and skin employed primarily by doctors in Oriental medicine.
After setting a reference point in the face area, information like length and angle was extracted for shape variables. Color variables were extracted on the basis of the brightness and color by each face area.
By applying a statistical machine learning technique that combines artificial intelligence (AI) machine learning with statistics, the research team compared and analyzed the facial characteristic variables that are different between high blood pressure patients and normal persons.
With this method, the research team deduced the facial characteristic variables that enable prediction for high blood pressure.
The analysis showed that the variables with the biggest facial difference between high blood pressure patients and normal persons were the angle and width of nose (shape variables) and the color of forehead and cheek (color variables).
By gender, the biggest facial difference between high blood pressure patients and normal persons were the shape of eye and the angle of nose for females and the width of nose and the color of cheek for males.
Kevin Lee (kevinlee@koreabizwire.com)
