AI Can Make Sense of Covid-19 Data and Speed Vaccine
Technology firms and medical researchers are working with artificial intelligence to beat the coronavirus pandemic.
They’re already using AI to analyze big data and track the biological development of the Covid-19 disease and the coronavirus that causes it. Researchers hope that AI — which means teaching machines to think more like humans — will eventually help answer questions about how the virus spreads.
It could ultimately help identify treatments and a vaccine.
AI forms an essential bridge between raw data and meaning. Published research on coronaviruses over recent years including Covid-19 have created a substantial body of intelligence. But the sheer scale of that data makes it hard to extract information that would lead to medical breakthroughs.
Ultimately AI could help develop drugs to treat and vaccinate against Covid-19 by speeding up the time to research medication from four years to less than one.
Special White House request
A consortium of technology and academic bodies on March 16 launched the Covid-19 Open Research Dataset, or CORD-19, at a request from the White House. This database contains 24,000 research articles on the pandemic. The articles have been converted into machine-readable files so engineers can create machine learning algorithms to make sense of the data. Machine learning is a form of AI that identifies patterns in large data sets. The technology can also help scour through data on existing drugs to identify the ones likely to be useful.
Researchers will apply another form of AI, natural language processing, to aid in understanding the research and identify any links between disparate sets of data.
The open research project encourages researchers to develop machine learning tools that scan data for patterns. The articles are hosted on the Semantic Scholar academic search engine operated by the Allen Institute for AI.
Researchers can share any data mining tools for CORD-19 file analysis on the website of Google-run data science community Kaggle. Kaggle’s 4.3 million participating data scientists could lead the way in building machine learning tools to extract useful insight from the files. They will seek to answer questions raised by the World Health Organization about the transmission, progress and cure for the disease.
In a similar vein, a team at Boston Children’s Hospital is using machine learning to scour social media posts, news stories and health data to track progress of the virus.
Shot in AI’s arm
If AI can fight the coronavirus, the results could help combat longstanding negative perceptions of the technology. AI has been heavily criticized for promoting “algorithmic bias” – embedding social prejudices into everyday life – and for the spread of facial recognition technology associated with authoritarian governments bent on surveillance.
A win in medicine has been expected all along, too. The coronavirus behind Covid-19 was spotted by an AI-powered early warning system before health authorities knew about it. Credit for that finding goes to Toronto-based start-up BlueDot, which identifies infectious diseases by analyzing news reports in multiple languages, government health data, international travel and information about human and animal movement.
The system identified the cluster of unusual pneumonia cases near a market in Wuhan, China, in December 2019 — nine days before the World Health Organization announced its emergence.
Successful introduction of a new technology usually depends on finding a “killer application,” a transformational use that makes it popular and indispensable. This time, could it kill a virus and help rescue the world?
