Facebook has created a system for predicting the condition of a patient with coronavirus – UNIAN

The system is designed to help hospitals provide sufficient resources for patient care.

Facebook has created a system that can predict the condition of a patient with coronavirus photo REUTERS

Facebook has developed a system that can help doctors determine how a person with COVID-19 can develop.

The work was carried out jointly with the Department of Forecasting Analytics and the Department of Radiology of NYU Langone.

There are three options:

  • a model for predicting the deterioration of the patient’s condition based on a single X-ray,
  • a model for predicting the deterioration of the patient’s condition based on a sequence of X-rays,
  • a model for predicting the need for additional oxygen based on a single X-ray.

A model that uses sequential chest x-rays can predict up to four days that a patient may need additional intensive care solutions. These predictions can help physicians avoid sending patients at risk too quickly and help hospitals better predict the need for additional oxygen and other limited resources.

Experts use two publicly available sets of chest x-rays, using a self-study technique called Momentum Contrast. “This allowed us to use large volumes of non-COVID chest X-rays to train a neural network that could extract information from chest X-rays,” Facebook said in a statement. kommersant.

News on the topic

Read the latest news of Ukraine and the world on the UNIAN channel Telegram

If you find an error, select it with the mouse and press Ctrl + Enter


Source: Інформаційне Агентство УНІАН by covid.unian.ua.

*The article has been translated based on the content of Інформаційне Агентство УНІАН by covid.unian.ua. If there is any problem regarding the content, copyright, please leave a report below the article. We will try to process as quickly as possible to protect the rights of the author. Thank you very much!

*We just want readers to access information more quickly and easily with other multilingual content, instead of information only available in a certain language.

*We always respect the copyright of the content of the author and always include the original link of the source article.If the author disagrees, just leave the report below the article, the article will be edited or deleted at the request of the author. Thanks very much! Best regards!