How machine learning is powering the cognitive search for COVID-19 answers
The pandemic has threatened the health of every person on earth. The internet has responded to the crisis with information that answers the multitude of questions people have. But it isn't always clear where to find trustworthy information.
Along with the rest of us, medical researchers and professionals working to treat those affected by COVID-19 struggle to sift the facts out of the mountains of information on the internet - right or wrong, current or outdated.
Microsoft cognitive search specialist Jennifer Marsman wanted to fact-check a claim that giving ibuprofen to children could make COVID-19 symptoms worse. She found a lot of conflicting information that wasn't useful.
"I just wanted to find a source of information I could trust," she says.
That problem sparked an idea. What if she could use her skills to help medical professionals locate accurate and timely information?
She explains, "As a software developer, I'm not going to be making any medical innovations, but how can I help scientists and healthcare workers make their innovations faster?
Research on the pandemic is pouring in from research facilities, labs, and universities all over the world - the kind of research that provides new ways to treat COVID-19.
Medical professionals are also putting in countless hours of time and effort to help those in need, and having access to the most relevant and up-to-date data on specific cases is of huge importance to improving patients' treatment (and survival).
Marsman and her colleague Liam Cavanagh, alongside the Microsoft Cognitive Search team, developed a new online search function that leverages machine learning. The free online tool is called COVID-19 Search, and it is designed to make it easier to find timely, relevant information about COVID-19.
People can search for COVID-19's impact on the body, which medications are effective, the role genes play, and articles by specific contributors.
Says Marsman, "The World Health Organisation warned against giving children ibuprofen, then recanted that advice. Using the Search, people can narrow the parameters so they're only getting the most up to date information, which helps remove confusion and saves them time. And, we hope, that helps save lives.
Built on Microsoft's Azure cloud platform, the search uses machine learning to scan research and medical journals in the COVID-19 Open Research Dataset (CORD) database. It then provides content summaries of the most important information, cross-references search terms, and even generates graphs to show links between information. Marsman hopes to add more functionality over time.
"Microsoft very much recognises that we have the tools, intelligence, and the resources to make a difference. When we get the chance, why wouldn't we?
Marsman and the Microsoft Cognitive Search team are far from alone in their mission to improve healthcare through technology.
Microsoft New Zealand partner director Matt Bostwick says that local partners and customers are using technologies like augmented reality and artificial intelligence to create amazing solutions for healthcare.
For example, New Zealand firm CloseAssociate developed a COVID-19 modelling tool, and Southern Institute of Technology nursing students are using Microsoft HoloLens to learn how to treat virtual COVID-19 patients. Aceso and Whanau Tahi are two other Microsoft partners helping join up health services across New Zealand at the highest levels down to the grassroots.
"Even though we've largely been isolated from the worst of the pandemic, we're innovating along with the best," concludes Bostwick.
