Facebook wants to help doctors fight COVID-19 with AI and X-rays

Facebook has spent the better part of this month addressing criticism around the world for amendments made to its privacy policy in connection with how it deals with the data of users interacting with businesses. The company has now stated that it wants to help doctors working on the frontline with its in house AI technology.

The novel coronavirus has been absolutely devastating on most countries healthcare systems and economies, with many healthcare workers fatigued or pushed to exhaustion. Facebook’s AI division has been conducting research into leveraging AI to help make more informed decisions on treating patients during the pandemic, the company said in a blog post.

Read more: Researchers introduce first artificial intelligence tool to detect COVID-19 probability

The researchers at Facebook worked with NYI Langone Health's Predictive Analytics Unit and Department of Radiology while creating three machine learning models for healthcare workers that can help with projections on a patient’s condition, for better or worse. One model predicts deterioration based on an X-ray, the second can predict deterioration on the basis of multiple X-rays and the last model can help doctors decide how much oxygen support a patient may eventually require.

According to the company, the second machine learning (ML) model, which uses sequential chest X-rays, is effective to the extent that it can predict whether a patient needs more intensive care solutions up to four days in advance. That 96-hour window might actually prove critical for doctors looking to treat patients with comorbidities or monitor at-risk patients for a little longer before they’re sent home.

Facebook’s team also stated that they were open sourcing their ML models and publishing the research “so the broader community can benefit from and build on what we’ve done.” Considering the company’s claims that the AI can ‘outperform predictions by human experts’, it seems like this could be a good addition to the tools used by frontline workers while fighting the pandemic.