Dr. Dan Ponticiello, 43, and Dr. Gabriel Gomez, 40, intubate a patient with coronavirus disease (COVID-19) in the COVID-19 ICU at Providence Mission Hospital in Mission Viejo, California, on January 8, 2021.
Lucy Nicholson | Reuters
Facebook’s artificial intelligence researchers say they have developed software that can predict the likelihood that a Covid patient will deteriorate or need oxygen based on chest x-rays.
Facebook, which worked with academics from the predictive analytics unit and NYU Langone Health’s radiology department in the research, says the software could help doctors prevent at-risk patients from being sent home too soon. at the same time it would help hospitals plan oxygen demand.
The ten researchers involved in the study, five from Facebook AI Research and five from the NYU School of Medicine, said they have developed three “models” of machine learning in total, which are slightly different.
One tries to predict patient deterioration based on a single chest x-ray, another does the same with an X-ray sequence, and a third uses a single x-ray to predict the amount of supplemental oxygen (if any). that a patient may need. .
“Our model using sequential chest X-rays can predict up to four days (96 hours) in advance if a patient may need intensive care solutions, usually exceeding the predictions of human experts,” the authors said in a published publication on the blog on Friday.
William Moore, a professor of radiology at NYU Langone Health, said in a statement: “We have been able to show that with the use of this AI algorithm, serial chest x-rays can predict the need for an escalation. of care in patients with Covid-19. “
He added: “As Covid-19 remains a major public health problem, the ability to predict a patient’s need to elevate care (e.g., ICU admission) will be essential for hospitals. “.
In order to learn how to make predictions, the AI system received two sets of chest x-ray data from non-covid patients and a data set of 26,838 chest X-rays from 4,914 covid patients.
The researchers said they used an artificial intelligence technique called “impulse contrast” to form a neural network to extract information from chest X-ray images. A neural network is a computer system loosely inspired by the human brain that can detect patterns and recognize relationships between large amounts of data.
The research was published by Facebook this week, but experts have already questioned the effectiveness of artificial intelligence software.
“From a machine learning perspective, we should study the extent to which this translates into new and invisible data from different hospitals and patient populations,” said Ben Glocker, who researches machine learning for imaging in the world. ‘Imperial College London, by email. “From my described reading, it appears that all the data (training and testing) comes from the same hospital.”
Researchers from Facebook and NYU said: “These models are not products, but research solutions, designed to help hospitals in the coming days and months with resource planning. Although hospitals have their own sets of data, often lack the computational power needed to form deep learning models from scratch. “
“We are opening up our pre-treated models (and publishing our results) so that hospitals with limited computing resources can adjust the models using their own data,” they added.