AI to see a doctor: an Indian study judges colds by analyzing sounds, with an accuracy of 70%

On the morning of April 10, Beijing time, it was reported that researchers at the National Institute of Technology in Sadar Walrab, India, analyzed the voices of 630 Germans, of whom 111 had colds. The researchers want to determine who is healthy and who has a cold by analyzing the sound.

▲ image source piqsels researchers mainly analyze the rhythm of human speech, which is equivalent to harmonics, and the amplitude decreases as the frequency increases. The researchers believe that a cold will break the conventional sound pattern. The researchers then input the information into the machine learning algorithm to distinguish the difference between the normal sound and the patient's voice.

The subjects counted from 0 to 40, then told the researchers what they had done over the weekend, and then recited Aesop's fable, the North Wind and the Sun. The study found that AI is 70% accurate in judging whether a person has a cold.

As for the value of the technology, the researchers explained that if AI can judge whether a human has a cold, there is no need to waste money or spend time seeing a doctor. Of course, it will be much more difficult for employees to pretend to be ill and ask for leave in the future.

Last year, a team of researchers tried to diagnose Parkinson's disease by a patient's breathing pattern, and using AI to analyze his voice could even determine whether the patient had depression or cancer.