Fb, NYU Langone Well being Use AI to Predict Covid-19 Affected person Well being


Facebook's use of machine learning to predict the types of content users are most interested in in their news feeds has been recorded many times over the years. However, the social network is also working on potential ways to predict treatment and resource needs for Covid-19 patients as part of its ongoing collaboration with the Predictive Analytics Unit and the Radiology Department of NYU Langone Health.

The two parties developed three machine learning models that physicians can potentially use to predict how their patients' condition will develop and ensure they have adequate resources to care for those patients:

  • A model for predicting patient deterioration from a single x-ray.
  • A model for predicting patient deterioration based on a series of X-rays.
  • A model for predicting how much additional oxygen (if any) a patient might need based on a single x-ray.

Research engineer Matthew Muckley, research assistant Koustuv Sinha, research engineer Anuroop Sriram, and program manager Nafissa Yakubova wrote in a blog post: “Our sequential chest x-ray model can predict up to four days (96 hours) in advance if a patient can need more intensive care solutions that generally exceed the predictions of human experts. These predictions could help doctors avoid sending high-risk patients home too early and help hospitals better predict needs for supplemental oxygen and other limited resources. "

William Moore, Professor of Radiology at NYU Langone Health added, “We have shown that this artificial intelligence algorithm can use serial chest x-rays to predict the need for escalation in care for patients with Covid-19. With Covid-19 remaining a major public health concern, the ability to predict a patient's need for increased care – such as ICU admissions – is vital for hospitals. "

Facebook and NYU Langone Health are making their pre-built models available as open-sourced products and releasing their research so that the wider community can benefit and build on them.


Jeffrey Rabinowitz