(Subscribe to our Today's Cache newsletter for a quick snapshot of top 5 tech stories. Click here to subscribe for free.)
Even after a year into the pandemic, one of the challenges doctors continue to face is how to predict a COVID-19 patient’s condition - - will it improve or worsen in a few days? Answer to that critical question will help hospitals better manage resources.
Facebook, in collaboration with NYU Langone Health’s Predictive Analysis Unit and Department of Radiology, has developed three machine-learning (ML) models to help doctors predict a patient’s condition. These models can help hospitals optimise their resources.
The first model can predict a patient’s deterioration based on a single X-ray. The second one can recognise a patient’s condition with a sequence of X-rays. This model can predict up to four days in advance whether a patient may need more intensive care. And the last ML model can predict how much supplemental oxygen a patient might need based on a single X-ray study, Facebook said in a statement.
Also Read | British hospitals use blockchain to track COVID-19 vaccines
These predictions could help doctors avoid sending at-risk patients home too soon, thereby mitigating complications.
Previous approaches to solving the resource use problem have relied on supervised training, where the programme is specifically told what to look for in the form of input data and expected output. But labeling data is time-consuming and limiting.
Also Read | Coronavirus | Facebook bans false claims about COVID-19 vaccines
Facebook used large amounts of non-COVID chest X-ray data to train a neural network to extract information. And the company said it has open-sourced the models.
“We have been able to show that with the use of this AI algorithm, serial chest radiographs can predict the need for escalation of care in patients with COVID-19,” says William Moore, MD, a Professor of Radiology at NYU Langone Health.
