Microsoft’s Cloud and AI Services Tapped in Coronavirus Fight

24Mar - by aiuniverse - 0 - In Microsoft Azure Machine Learning

Source: rcpmag.com

Microsoft recently described the ways in which its artificial intelligence (AI) and machine learning offerings, as well as its Azure cloud, are being used by researchers and other public health groups responding to the ongoing coronavirus (COVID-19) pandemic.

In one collaboration announced Friday, Adaptive Biotechnologies Corp. and Microsoft are using Azure to map the immune system’s response to threats, including COVID-19. Adaptive is using Azure’s machine learning capabilities to sift through data on the body’s T-cell receptor sequences, which get generated in response to antigens in the blood. The machine learning process is used to refine a “map” of those sequences, which is done by “matching trillions of T cells to the diseases they recognize,” per Adaptive’s fact sheet (PDF download).

Leveraging Microsoft’s hyperscale machine learning capabilities and the Azure cloud platform, the accuracy of the immune response signature will be continuously improved and updated online in real time as more trial samples are sequenced from the study,” the announcement stated.

The aim of the effort is to more accurately detect diseases via blood tests. The partnership between Microsoft and Adaptive on the T-cell antigen map isn’t new, but dates back to 2018. It’s already led to a proof-of-concept for identifying Lyme disease. Adaptive plans to apply for clinical trials with the U.S. Food and Drug Administration sometime this year.

Adaptive is promising to share T-cell response signatures from COVID-19 with other researchers via a soon-to-come “open data access portal.”

“These data will be made freely available to any researcher, public health official or organization around the world via an open data access portal,” the announcement stated regarding the COVID-19 data.

Adaptive plans to use de-identified blood samples from individuals diagnosed with COVID-19 to bolster its data samples, and is seeking additional contributors. It’s working with LabCorp on the blood collection effort. Adaptive also is currently collaborating with the Providence health group, whose hospital in Seattle “treated the first U.S. COVID-19 patient.”

COVID-19 Assessment Bot
In addition to these efforts, Microsoft announced on Friday that it is supporting the U.S. Centers for Disease Control and Prevention’s (CDC’s) newly released COVID-19 assessment bot, which is “powered by Microsoft Azure.” The bot, which asks users health questions to assess possible COVID-19 infections, is based on Microsoft’s Healthcare Bot service and uses artificial intelligence in the screening process.

“Microsoft’s Healthcare Bot service is one solution that uses artificial intelligence (AI) to help the CDC and other frontline organizations respond to these inquiries, freeing up doctors, nurses, administrators and other healthcare professionals to provide critical care to those who need it,” the announcement explained.

Hospital systems are already using Microsoft’s bot to screen COVID-19. They include Seattle-based Providence, Virginia Mason Health System in the Pacific Northwest and Novant Health in the Southeast. Customized versions of the bot are currently handling “more than 1 million messages per day.”

Microsoft is helping other organizations build their own health bots by providing four “COVID-29 response templates.” Two of the templates are based on the CDC’s protocols. Organizations can modify the templates, if wanted.

COVID-19 Portal
The CDC reports U.S. cases of COVID-19 at this page, which includes a U.S. “heat map” illustration of the disease’s distribution. Cases have been reported in all 50 states.

Another site showing COVID-19 distribution around the world is nCoV2019.live. The site was put together by Avi Schiffmann, a high school junior from Mercer Island near Seattle, according to a report by the Democracy Now! news program. Schiffmann used Web scraping to pull the data together from various government sites.

Facebook Comments