Much work has been done on telemedicine technology during the Corona epidemic. Now the University of Washington has developed a video technology that can accurately detect the breathing and pulse rate of a person in a video in real time. It will be demonstrated at the April 8 ACM Conference on Health, Interface and Acquisition.
This technique notes both important medical symptoms by looking at the face. It is now expected that this will boost telemedicine. Machine learning classifies images, said Zhen Lai, a doctoral student at the University of Washington. Now, thanks to this, physical and physical conditions can be revealed by looking at the face. But everyone is different in their nature. Therefore, the whole system must be rapidly changing so that it can monitor the physical changes of every person, regardless of their environment and background.
The video takes care of personal privacy as its data is limited to the phone instead of the cloud. By changing the light and shadow on the face, the machine learning process keeps track of blood flow and respiratory fluctuations. He has been trained on the original video and pictures of the people. Breaths and pulses were also recorded by instruments and combined with video.
After a few days, the software enabled itself to record videos of people walking and even their breathing and pulse rate. But it performed better on some datasets while it did not perform better on other datasets and this process is called ‘overfting’ in machine learning.
The machine was then trained as an individual with good results. However, for black people, it needs more training. However, smartphone video can also be used to detect breathing and pulse rate.
